If you’re responsible for growth in any business today, the system you’re operating is becoming increasingly difficult to manage.

Acquisition costs have increased by more than 60% over the past five years, according to ProfitWell, but the signals that once made that spend intelligible are fading. Privacy changes and platform fragmentation have materially reduced attribution fidelity, as reported by AppsFlyer and Meta, making it harder to understand what is actually driving performance.

At the same time, customer expectations have shifted. People no longer experience brands through a single channel. They expect interactions to feel continuous, relevant, and immediate, regardless of where they happen.

Most retail organizations are still operating on a model that was built for a different environment. It is a model designed around discrete transactions, mediated by platforms that once offered both reach and clarity.

That model is no longer holding.

What is striking is not just the pressure itself, but how consistently companies respond to it. Rather than rethinking the system, they look for ways to extend it. And increasingly, that extension takes the form of a mobile app.

The Misunderstood Role of the App

The decision to build an app is rarely irrational. It is usually framed as a way to regain control. An app promises stronger loyalty, more repeat purchase, and better attribution. It appears to offer a direct relationship with the customer in a landscape where intermediaries are becoming more powerful.

But the promise of the app is often misunderstood.

An app is not, in itself, a solution to any of these problems. It is simply another surface through which a system operates. If the underlying system does not change, the outcomes will not change either.

This is why so many retail apps fail to deliver on their promise. Data from Adjust shows that more than 90% of users abandon an app within the first 30 days, which means the problem is not distribution but sustained relevance. They are built as extensions of the website, or as containers for existing programs, rather than as systems designed to influence behavior over time.

An app only becomes valuable when it becomes part of a customer’s routine. Without that, it remains inert. It may be downloaded, occasionally opened, and largely forgotten, in a landscape where attention is already concentrated in a small number of dominant apps, as consistently shown by data.ai.

The difference between those two outcomes is not the quality of the interface. It is the design of the system behind it.

Subscribe now

Loyalty as an Outcome of the System

This becomes most visible in how retailers approach loyalty.

The instinct is to begin by embedding the loyalty program within the app. On the surface, this makes sense. If loyalty is the goal, then the program that represents loyalty should be central to the experience.

But this approach confuses representation with causation.

Loyalty programs do not create behavior. They record it. They provide a way to track and reward actions that have already occurred, but they do not, on their own, generate those actions. Research from Bond Brand Loyalty shows that more than half of loyalty program memberships are inactive, reinforcing how rarely these systems translate into ongoing engagement.

Customers return to products that consistently deliver value in moments that matter to them. That value has to feel timely, relevant, and easy to access. It has to align with the customer’s context, rather than forcing the customer to adapt to the product.

When that kind of value is present, behavior follows. When it is absent, no amount of points or rewards will compensate for it.

In that sense, loyalty is not something you build into the product. It is something that emerges from how the product behaves.

The System Behind Repeat Purchase

The same dynamic applies to repeat purchase.

Retail has historically been structured around acquisition. The system is designed to bring customers in, convert a percentage of them, and then begin the process again. Each visit is treated as an isolated event, and the relationship between brand and customer is largely reset after each transaction.

This model is effective, but it is also inherently fragile. It depends on continuous spend, and it does not accumulate knowledge or context about the customer in a way that meaningfully shapes future interactions. The economic implications are significant. According to Bain & Company, increasing retention by just five percent can increase profits by as much as ninety-five percent, while Invesp estimates that existing customers are up to five times more likely to convert than new ones.

An app introduces a different kind of system.

Instead of starting from zero with each interaction, the relationship persists. The customer is no longer anonymous. Their behavior unfolds over time, and the brand has the ability to respond to that behavior in a way that feels continuous rather than episodic.

At that point, the nature of the problem changes. Growth is no longer about capturing intent when it appears. It becomes about creating the conditions under which intent re-emerges.

Repeat purchase is not simply a reflection of customer preference. It is a function of how well the system is able to re-engage the customer at the right moment, with the right context, in a way that feels relevant rather than intrusive.

When that system is well designed, repeat behavior becomes more predictable. In practice, many ecommerce businesses already see a disproportionate share of revenue come from returning customers, often approaching forty percent according to RJMetrics. When it is not, it remains dependent on external triggers and continued spend.

Attribution and the Question of Control

What often appears, on the surface, to be an attribution problem is in fact a consequence of this deeper structural issue.

Retail marketers are operating inside environments where they do not control the underlying system. Data is fragmented across platforms, and visibility into behavior is partial at best. Surveys from Gartner show declining confidence in attribution models, while research from McKinsey & Company highlights how incomplete and fragmented customer journey data has become.

In that context, attribution becomes an attempt to reconstruct a system that is no longer fully observable.

An app changes that dynamic, but not because it solves attribution directly. It changes who owns the system in which behavior occurs.

Within an app, identity persists across interactions. Behavior can be observed as it unfolds. Engagement is not inferred from external signals but measured directly within the system itself. The loop between action and outcome becomes visible in a way that is increasingly difficult to achieve elsewhere.

The result is not just better measurement. It is a different way of making decisions. When behavior can be observed clearly, the need to rely on assumptions diminishes. The system begins to reveal what is actually working.

When this system is designed well, the performance difference becomes clear. Mobile apps consistently outperform mobile web experiences, with data from Criteo showing conversion rates that are often two to three times higher. Engagement follows a similar pattern, with Airship reporting significantly stronger retention among app users.

Over time, that clarity compounds into advantage.

What Building Outlook Mobile Actually Revealed

This dynamic is not unique to retail.

While building Outlook Mobile at Microsoft, we faced a different set of surface-level problems, but the underlying system was the same. The challenge was not simply to build a product that people could use, but to create a system that people would return to.

Growth did not come from adding features in isolation. It came from understanding how users moved through the product, identifying where they disengaged, and redesigning those moments to reduce friction and increase relevance.

Every return was the result of value being delivered in a way that aligned with the user’s context. Over time, those returns accumulated into habit.

What became clear through that process was that growth is not a function of what the product contains. It is a function of how the system behaves.

The Missing Discipline

This is the part that most retail strategies underestimate.

The difficulty is not in building an app. The difficulty is in designing and operating the system that gives the app meaning.

Creating a mobile experience that people return to requires a different kind of organizational capability. It requires instrumentation that captures behavior from the beginning. It requires continuous experimentation, not as a separate initiative, but as a core part of how the product evolves. It requires aligning teams around outcomes that are expressed in behavior, rather than outputs that are expressed in features.

Most retail organizations are not structured in this way. Their strengths lie in brand, merchandising, and supply chain. Those capabilities remain essential, but they do not, on their own, produce the kind of system required to shape customer behavior over time.

The result is a gap between intention and execution. The desire to create loyalty and repeat purchase is present, but the system required to produce those outcomes is not.

Subscribe now

From Channels to Systems

What is happening here is part of a broader shift.

For a long time, growth was treated as a function of channel performance. The focus was on acquiring more traffic, improving conversion rates, and increasing efficiency within the funnel. The underlying assumption was that if the inputs could be optimized, the outputs would follow.

That assumption is becoming less reliable.

The companies that are succeeding in this environment are not simply optimizing channels. They are designing systems that shape behavior over time. They are creating environments in which customers return, engage, and deepen their relationship with the product without requiring constant re-acquisition.

This is a different way of thinking about growth. It is less about driving isolated events and more about creating continuous interaction.

In that model, the app is not the product. It is the interface through which the system operates.

The Strategic Shift

The distinction between web and mobile becomes clearer when viewed through this lens.

The web is well suited to capturing intent at the moment it arises. It performs effectively as a transactional surface, converting demand that already exists.

Mobile, when designed correctly, functions differently. It creates the conditions for that demand to reappear. It enables a relationship that persists between transactions, rather than resetting after each one.

As acquisition becomes more expensive and less predictable, this distinction becomes more important. More than seventy percent of consumers now expect personalized experiences, according to Salesforce, and companies that deliver on that expectation generate materially higher growth, as shown in research by McKinsey & Company. Brands that rely entirely on external platforms for discovery are increasingly dependent on systems they do not control.

The companies that win will be those that build systems of their own. Systems that allow them to maintain a direct, persistent relationship with their customers, and to shape behavior over time rather than simply reacting to it.

This is not a shift in channel strategy.

It is a shift in how the business operates.

At its core, growth in retail is no longer driven primarily by traffic. It is driven by behavior. And behavior, in turn, is a function of the system you design.

The brands that recognize this, and build accordingly, will not just perform better within the existing model.

They will define what the next model becomes.

The cost of producing software has collapsed so quickly that most founders haven’t updated their mental model of what building a product now means. Describe a mobile app in plain English. Watch a credible prototype take shape in hours. Connect the APIs. Wire up authentication. The result is real software. It runs. It looks convincing.

AI makes apps easy to prototype. It does not make them work in production. And the gap between those two things is where most mobile efforts die.

The prototype illusion

A prototype gives the comforting illusion of progress because it demonstrates possibility. It proves the interface can exist, the APIs can connect, and the core journeys can be rendered on a device. That is genuinely useful. But it also creates a specific kind of danger for founders: the demo looks like a product when it is actually just a surface.

Production is an entirely different state. Production means the app has entered an ecosystem where reliability matters, compliance matters, performance matters, and every weakness is exposed not in a demo environment but in the hands of real customers. The question is no longer whether the app can be built. The question becomes whether you have built the systems required to operate it.

That is where the hard part begins.

Subscribe now

Where the System Breaks

The first thing founders discover is that the platform is not neutral. Apple and Google are not passive distribution channels. They are active gatekeepers with evolving rules around privacy, permissions, payments, and data use. A prototype that works perfectly in a test environment can still fail app review, and even after approval, the ground continues to shift. Policies change, operating systems update, and what passed yesterday may need to be rebuilt tomorrow. Shipping is not a milestone. It is the beginning of an ongoing negotiation with an ecosystem you do not control.

Even if you pass that threshold, a different problem emerges. Once the app is live, it has to become legible to you. Installs begin to accumulate, usage begins to form, and very quickly you realize you cannot answer the most basic questions about what is actually happening. Which users are real, where they drop off, what brings them back, and whether the last release improved anything at all. Without instrumentation designed into the system from the start, the app becomes opaque. You are no longer building based on reality, but interpreting fragments of it.

Then comes retention, which is where most systems fail. An app only becomes valuable if it creates a reason to return, and return is not driven by intention but by structure. It depends on whether the system can recognize behavior, respond to it in context, and do so with enough precision that the experience feels relevant rather than generic. Without that, engagement degrades quickly. Messages are sent, but they are poorly timed. Personalization exists in theory, but not in practice. The app may still be installed, but it is no longer part of the user’s daily routine.

As usage grows, quality becomes unavoidable. The app that worked in controlled conditions is now exposed to the full variability of real environments. Different devices, different operating systems, unstable networks, edge cases that were never anticipated. Failures begin to surface in ways that are difficult to reproduce but impossible to ignore. Each release introduces risk, and without a disciplined approach to testing, monitoring, and rollback, that risk accumulates. What initially felt like speed starts to reveal itself as fragility.

And beneath all of it sits security, which becomes more consequential as the system succeeds. The app is no longer just an interface. It is a surface connected to identity, payments, behavioral data, and a growing network of dependencies. Weaknesses that were invisible in a prototype become meaningful in production. Trust, once broken, is difficult to recover. Security is not a layer you add once the product works. It is part of what allows it to work in the first place.

None of these issues exist in isolation. They are not separate problems to be solved independently. They are expressions of the same underlying reality: the app is not the system. It is the entry point into one.

Why point solutions fail

Once founders recognize these realities, the natural instinct is to assemble solutions from tools. One for analytics, another for messaging, another for attribution. It feels efficient to construct a stack where each component is best in class.

But a mobile app does not behave like a collection of independent parts. It behaves like a tightly coupled system. The way behavior is captured determines what can be seen. What can be seen determines what can be acted upon. What can be acted upon determines whether engagement is effective. When these layers are disconnected, the system fragments. Data becomes inconsistent. Signals lose fidelity. Teams operate on different versions of reality.

The creation of the app, the measurement of behavior, the orchestration of engagement, and the economics of acquisition are not separate problems. They are a single loop. When that loop is intact, the system improves itself. When it is broken, no amount of additional tooling compensates.

Before the System Exists

Most founders only realize the gap between prototype and product after they have already crossed it.

At first, the signals look encouraging. Installs begin to accumulate. Usage appears to grow. But very quickly, basic questions become difficult to answer. It is not clear which users are real, where they drop off, or what brings them back. The system produces activity, but not understanding. Without instrumentation designed in from the start, the product becomes something you can observe, but not explain.

At the same time, re-engagement is often mistaken for messaging. Notifications are sent, campaigns are scheduled, and activity is generated, but the system lacks the structure to respond to behavior in context. What looks like a retention strategy is often just a calendar. It works until it doesn’t, and when it breaks, it does so quickly.

Underlying both of these is a more fundamental issue. No one actually owns the system that sits beneath the app. Compliance, release discipline, QA, and store operations are treated as secondary concerns, distributed across the team rather than held as a primary responsibility. The product exists, but the system that allows it to operate reliably does not.

By the time these gaps become visible, the prototype has already become a product in name, but not in reality.

The actual hard part

Building Outlook Mobile taught me that the visible software is only one layer. Underneath it sits a set of technical, operational, and behavioral systems that determine whether an app becomes an enduring channel or a forgettable artifact. The hard part was never getting the app to exist. The hard part was turning it into something dependable enough, measurable enough, and useful enough that people returned to it until it became part of their daily lives.

Vibe coding can help you build the surface quickly. That matters. But the surface is not the moat. The system beneath it is.

The question is no longer whether you can build an app. It is whether you can operate the system that gives that app meaning.

If SaaS were just interfaces, it would already be over.

The assumption behind the SaaSpocalypse narrative is that software is just a collection of UIs. And if Artificial Intelligence can generate software on demand, companies will no longer buy it.

Across the technology sector, investors increasingly believe that enterprises will soon start vibe-coding their own CRM, ERP, HRIS, and other internal systems. If AI can generate software on demand, the thinking goes, why pay expensive SaaS subscriptions at all?

That belief has already erased hundreds of billions of dollars in market value from cloud software companies. Across the sector, valuations for many SaaS leaders have fallen sharply from their peaks as investors price in a future where artificial intelligence eliminates traditional applications.

The reasoning seems straightforward.

For two decades the SaaS model appeared simple: build a database with a graphical interface, charge a subscription for every employee who logs in, and rely on switching costs to keep customers from leaving. Software could be written once and sold many times. Every new hire created another seat.

Artificial intelligence, the story goes, has now broken that model.

Software is becoming trivial to build. Automation will reduce the number of humans inside organizations, which means fewer seats to charge for. And the dashboards that defined the SaaS era may disappear entirely as AI agents interact directly with APIs.

From this perspective, the future belongs not to SaaS companies but to the infrastructure platforms that enable autonomous agents.

It is a clean story.

It is also wrong.

Because SaaS companies don’t really sell software.

They sell systems that run businesses.

To understand why the current narrative around AI is wrong, it helps to look at how enterprise software systems are actually structured.

Artificial intelligence does not eliminate the need for those systems. If anything, it increases their importance.

The Myth of the Dashboard

The extinction narrative assumes SaaS companies primarily sell software artifacts: code, screens, and features.

If that were true, and software were simply capabilities wrapped in a user interface, recreating them with AI would be trivial.

But enterprise software inside real organizations does not behave like interchangeable features.

Over time it becomes something far more consequential. It becomes a system through which work is organized, decisions are made, and accountability is enforced.

Companies do not buy enterprise software because they enjoy dashboards. They buy it because businesses need shared systems of record that multiple teams can trust. They need processes that ensure work happens, approvals are tracked, and responsibility is clear. They need permissioning systems, audit trails, operational reliability, and compliance safeguards.

Artificial intelligence can accelerate these systems.

It does not remove the need for them.

Subscribe now

The Salesforce Paradox

Salesforce succeeded not because of its interface, but because it became the system of record for revenue.

Pipelines live inside it. Forecasts come from it. Marketing attribution depends on it. Compensation plans reference it. Leadership meetings revolve around the reports it produces.

Once a system sits at the center of revenue operations, it stops behaving like a tool and starts behaving like infrastructure.

Replacing it is no longer about rebuilding screens. It means reconstructing the network of workflows, integrations, data dependencies, and operational habits that formed around it.

The interface was never the moat.

The system was.

Control Points and the Microsoft Environment

Two other examples reveal the deeper structure of enterprise software power.

My experience working on Outlook Mobile inside the Microsoft ecosystem showed how powerful infrastructure can be. Even when the tools themselves are far from perfect.

Few companies have deeper enterprise infrastructure than Microsoft. Identity runs through Active Directory. Documents live in SharePoint and OneDrive. Email flows through Exchange Online. Teams coordinates communication.

But the experience of using them is messy.

File sharing can be confusing. Collaborative editing often breaks. The distinction between OneDrive and SharePoint is unclear even to experienced users. The interplay between Teams and Outlook frequently requires manual work to bridge communication and coordination.

And yet Microsoft still sits at the center of enterprise workflows.

The reason is that Microsoft controls several critical control points inside the enterprise.

Identity is a control point. Data is another. Communication is a third.

When a company controls those control points (identity, documents, communication) it becomes extremely difficult for workflows to exist outside its environment.

Work flows through the system whether the experience is elegant or not.

Microsoft owns the workflows of the enterprise in spite of the tools, not because of them.

The Slack Lesson

The opposite lesson comes from Slack.

Slack delivered one of the most elegant enterprise user experiences ever created. It replaced chaotic email threads with fast, conversational communication. Teams organized themselves around channels instead of inboxes. Conversations became searchable, persistent, and collaborative.

For many people, Slack felt like the first piece of enterprise software designed for humans rather than IT departments. Teams loved it and adoption spread inside organizations with remarkable speed. But Slack never became the operating system of the enterprise.

The reason is structural. Slack sits at the communication layer around work, not at the system where the work itself is defined and executed.

Inside Slack, teams discuss deals. But the deal still lives in Salesforce. They coordinate around documents. But the document still lives in Microsoft 365. They talk about incidents, product launches, and customer problems. But the systems that actually track those activities live elsewhere.

Slack became the place where work is discussed, but it never became the place where work is actually done. And that distinction explains why even a product with extraordinary design and massive adoption could not displace the deeper systems of the enterprise.

Slack proved something important about software. A beautiful interface can change how people communicate. But power belongs to the systems that define how work actually happens.

Seen together, these examples reveal a deeper structure inside enterprise software.

Systems that control the interface shape how people interact with work. Systems that control the data become the record of the business. Systems that control identity and governance determine where workflows can exist at all.

Power in enterprise software does not come from the surface of the product.

It comes from the layers beneath it.

The Operational Software System

A growing belief in the technology industry is that AI will allow companies to build their own software instead of buying it. In this view, enterprises will simply describe what they need, AI will generate the code, and entire categories of SaaS products (CRM, ERP, HRIS, and ticketing systems) will disappear.

The argument sounds plausible because it focuses on the most visible part of software: the interface. If AI can generate dashboards, forms, or workflows on demand, why license an expensive SaaS platform at all?

The problem with this reasoning is that software systems are not defined by their interfaces. They are defined by the operational layers that sit beneath them.

What appears to users as a single application is actually a stack of systems that coordinate how the business runs. At the top sits the interface layer, where humans interact with dashboards, copilots, or AI agents. Beneath that sits the workflow layer, where business processes execute approvals, automations, campaigns, and operational tasks.

Supporting those workflows is the control layer, which governs identity, permissions, and authority across the organization. Below that sits the data layer, the system of record that stores the company’s operational history.

Finally, at the foundation lies the trust layer, which provides security, compliance, reliability, and support. Without this layer, no organization would entrust its operations to software in the first place.

Seen through this lens, the idea that AI will allow companies to replace SaaS by generating their own systems becomes much less obvious. But building a durable operational system requires far more than code generation. It requires identity frameworks, governance models, data architectures, reliability guarantees, and the institutional trust that enterprises rely on to run their businesses.

AI may dramatically change how software is built and interacted with. But the deeper operational system that makes software reliable, governable, and trustworthy does not disappear.

Subscribe now

Prototypes Are Easy. Trust Is Hard

Yes, you can replace a vertical SaaS UI in days with vibe coding.

Give a model the schema, sketch the workflows, and ask for a UI. Within hours you can have something that looks like a CRM, a support system, or a marketing tool. Dashboards appear. Forms work. Records save. The prototype feels surprisingly real.

This is the moment that fuels the narrative that SaaS is finished. But building a convincing interface is not the hard part of enterprise software.

The hard part begins the moment the system becomes responsible for real data and real operations. Production systems must preserve data integrity over years of use, across thousands of edge cases and evolving schemas. They must enforce complex permission models, maintain audit trails, withstand security threats, and comply with regulatory frameworks that change constantly. They must stay available when infrastructure fails and recover gracefully when something breaks.

And something always breaks.

When it does, someone has to fix it. Someone has to understand the architecture, the edge cases, the historical decisions embedded in the system, and the expectations of the thousands of businesses relying on it. That knowledge lives not just in the code, but in the organization that operates the system.

Artificial intelligence dramatically reduces the time required to build software. But building software is not the same thing as building systems businesses trust.

AI collapses time-to-prototype. It does not collapse time-to-trust.

Businesses run on trust, not demos.

Fewer Humans Doesn’t Mean Less Software

The seat-tax era is ending.

Many SaaS companies historically priced their products per seat, which tied software revenue directly to the number of employees in a company. If automation reduces headcount, then the revenue base appears to shrink with it.

But shrinking teams makes software more valuable, not less.

When organizations operate with fewer people, each individual becomes responsible for a broader set of outcomes. Systems must automate more work, integrate more information, and support more decisions.

What changes is the pricing model, not the need for the system.

Seat licenses may give way to usage-based pricing, outcome-based pricing, or value metrics tied directly to business results.

But the underlying infrastructure remains essential.

The Interface May Change. The System Remains.

A related argument suggests that the graphical user interface itself may disappear. Instead of humans clicking through dashboards, AI agents will interact directly with APIs.

There is truth in this observation. Human interaction with software will likely change dramatically in the coming decade. But removing the dashboard does not remove the system behind it.

APIs still require business logic that determines what actions are allowed. They require state management that tracks what has happened and what must happen next. They require governance structures that define who has authority over which actions.

The deeper infrastructure remains.

Artificial intelligence may also make it easier to write code. But it does not make it easier to earn trust from buyers, pass security reviews, or navigate procurement processes inside large organizations.

It does not automatically replicate the partner ecosystems, training programs, and customer success teams that help organizations adopt and operate complex systems.

And it does not solve the problem of accountability.

When a business-critical workflow fails, someone must diagnose the issue, correct it, and ensure that it does not happen again. Companies therefore continue to look for vendors who will stand behind the systems they sell. They expect service-level agreements, escalation paths, and support teams that understand the context in which the software is used.

Over time these relationships embed the software inside the operating environment of the organization itself. Workflows evolve around the system. Data structures expand. Integrations form with dozens of other tools.

Eventually the software stops behaving like a tool people choose to open.

It becomes part of how the business runs.

Fit Matters More Than Flexibility

One of the promises of AI-generated software is infinite flexibility. If applications can be assembled dynamically, perhaps rigid systems become obsolete.

But businesses do not usually want infinite flexibility.

They operate under constraints imposed by regulations, industry standards, internal policies, and the practical realities of coordinating large groups of people. Systems that reflect these constraints are often more valuable than tools that can be endlessly reconfigured.

Organizations want software that fits the structure of their work. They want predictable behavior, clear boundaries, and systems that evolve in ways they can anticipate.

The more a system reflects the operational logic of an industry, the harder it becomes to replace.

The Real Correction

Artificial intelligence will reshape the software industry.

That part is obvious.

What is less obvious, at least judging from the market reaction, is what exactly is being reshaped.

The dominant story today is that SaaS itself is in danger. If AI can generate software, then enterprises will simply build their own systems. CRM, ERP, HR, support, entire categories of software will be recreated internally instead of purchased from vendors.

But this story assumes that SaaS companies were primarily selling software.

They weren’t.

The most successful companies of the cloud era were building the operational infrastructure through which modern businesses run. Their systems became embedded in workflows, integrated across organizations, and trusted to manage the records and processes that companies depend on every day.

This is what I saw as we were building Microsoft 365 into companies all over the world.

That kind of infrastructure does not disappear simply because software becomes easier to generate.

The core problem enterprise software solves has not gone away. If anything, it has become more important.

The real correction underway in the market is not an extinction event. It is a credibility test. Artificial intelligence is forcing a distinction that the SaaS boom often obscured.

Some companies built dashboards.

Others built systems that businesses actually run on.

Artificial intelligence will make that difference impossible to ignore.

The New Retail Power Structure

Amazon recently began rolling out a new capability called Shop Direct. At first glance, it appears to be a simple extension of Amazon’s marketplace. In reality, it signals a tectonic shift in the direction of modern commerce.

Shop Direct allows Amazon to surface products from merchants’ own websites inside Amazon search results. A customer can discover a product on Amazon and then complete the purchase on the brand’s website rather than through Amazon’s marketplace.

On the surface, this sounds like a win for retailers. In reality, it has massive strategic implications. Amazon no longer needs to own the transaction in order to control the retail ecosystem.

Amazon no longer needs to own the transaction to control the retail ecosystem. What matters is owning discovery. If consumers begin their buying journeys on Amazon, the company gains enormous influence over what brands get seen, what products get considered, and ultimately which companies grow. The transaction itself becomes almost irrelevant.

The power lies in controlling the starting point.

Discovery Is Being Centralized Again

Over the past decade, the internet has gradually reorganized itself around a small number of discovery platforms. Consumers increasingly begin their buying journeys inside a handful of environments where products are surfaced, ranked, and recommended.

Amazon is one of those environments. So are TikTok, Google, and Meta.

These platforms do not merely drive traffic. They shape the economic structure of entire industries. When discovery becomes centralized, brands are forced to compete within the rules of the platform that controls it.

We have already seen how this dynamic plays out inside Amazon’s marketplace. Brands that once built direct relationships with customers now compete inside a search engine where rankings, pricing pressure, and customer acquisition costs are largely determined by Amazon itself.

Shop Direct extends that dynamic beyond the marketplace.

Amazon is no longer limiting its discovery engine to products sold on Amazon. It is expanding that discovery layer to products sold anywhere on the internet.

In effect, Amazon is attempting to become the search engine for retail.

Subscribe now

The Real Strategic Shift

Most retailers will interpret this as a marketing problem. They will ask how to optimize their Amazon presence, how to improve advertising performance, or how to compete more effectively inside Amazon search.

But the real shift is structural. Discovery and customer relationships are beginning to separate.

Large platforms are increasingly becoming the places where customers first encounter products. They control the ranking, the visibility, and the moment of discovery. But discovery is only the beginning of the relationship.

The brands that succeed in this environment will not be the ones that control traffic. They will be the ones that control what happens after discovery.

The future of retail will not be determined by who owns the search result. It will be determined by who owns the customer relationship.

The Lesson From Outlook Mobile

This dynamic is not unique to retail. I first encountered it while working on Outlook Mobile at Microsoft.

When we launched Outlook Mobile, we quickly realized that a great product alone would not determine success. The real question was whether we could design a system that kept users engaged after installing the app and guided them through moments of value that naturally expanded into the broader Microsoft 365 suite.

We focused on building continuous engagement through notifications, integrations, and workflows that made Outlook part of a user’s daily routine. The goal was not simply to acquire users. It was to create a product system that turned acquisition into long-term retention.

That same principle now applies to retail.

Discovery can happen anywhere. But the companies that win are the ones that turn that discovery into an ongoing relationship.

Why Mobile Apps Are Becoming the Economic Engine of Retail

This is where mobile apps are becoming the most important strategic asset a retail brand can build.

A website is fundamentally a transaction surface. Customers arrive, browse products, and complete purchases. A mobile app creates something much more powerful. It creates a persistent connection between the brand and the customer.

In modern retail, discovery increasingly happens on platforms, but the relationship lives inside the brand’s app.

Through personalized experiences, loyalty programs, and behavioral insights, the app becomes the place where the relationship actually lives.

In a world where discovery increasingly happens on external platforms, that persistent connection becomes the engine of customer lifetime value.

Product-Led Economics in Retail

At Bryj, we describe this shift through the lens of Product-Led Economics.

Great products do not win markets by themselves. The companies that win build systems that turn product value into sustained economic advantage.

In retail, the mobile app is becoming the core of that system.

It enables brands to continuously engage customers, personalize experiences, and drive repeat behavior in ways that traditional digital channels cannot match. Over time, this engagement compounds into something much more powerful than a single transaction. It becomes a relationship that drives loyalty, frequency, and lifetime value.

In other words, the mobile app is not just another channel. It is the infrastructure that converts discovery into durable economics.

The next era of retail will be defined by three layers: platforms that control discovery, brands that own the relationship, and the economics that compound from that relationship over time.

The Future of Retail

Amazon’s Shop Direct initiative reveals something deeper about where commerce is heading.

For most of its history, Amazon’s power came from controlling both discovery and transaction inside its marketplace. If a brand wanted visibility, it had to sell on Amazon.

Shop Direct changes that equation.

Amazon is no longer limiting discovery to products sold on its platform. It is extending its discovery engine to products sold anywhere on the internet. A customer can now discover a product on Amazon and complete the purchase somewhere else entirely.

That is a much bigger strategic move.

Amazon is positioning itself as the starting point for product discovery across the entire retail ecosystem. If consumers begin their buying journeys there, Amazon gains influence over what gets seen, what gets considered, and ultimately what gets bought — even when the purchase itself happens somewhere else.

But that also clarifies the strategic divide that will shape the next era of retail.

Platforms will compete to control discovery.

Brands will compete to control the relationship.

The companies that win will not necessarily be the ones that generate the most traffic. They will be the ones that build the systems that turn discovery into loyalty, engagement into habit, and purchases into long-term customer relationships.

As discovery becomes more centralized, ownership becomes more valuable.

In the next era of commerce, discovery may happen anywhere.

The the brands that define the market will be the ones that own what happens next.

Product Is Not the Company

Most founders are taught to believe that great products build great companies. It is one of the most persistent ideas in technology, and one of the most misleading. Great products matter enormously. They can create delight, trust, habit, even love. They can solve real problems with elegance and precision. They can generate intense user pull and create extraordinary outcomes. But on their own, they do not build great companies. In many cases, the opposite is true. Great products destroy their companies all the time.

Ask Evernote, Vine, or Pebble. Great product. Real love. Weak company. Or at least, weaker than the product deserved.

That is the trap. In technology, we romanticize the product. We tell ourselves that if users truly love what we have built, the rest will follow. It often does not. A loved product can still sit inside a weak business model. It can still depend on the wrong buyer, the wrong channel, the wrong pricing architecture, or the wrong expansion path. It can generate usage without creating leverage. It can become essential to users without becoming durable for the company behind it.

A great product proves that a company has discovered how to create value. It does not prove that the company knows how to turn that value into an enduring business. The history of technology is full of products users loved and companies that never became what those products seemed to promise. In each case, the product was not the problem. The problem was that the value it created was never translated into a system that could produce lasting economic power.

What Product-Led Growth Did and Did Not Solve

That distinction has become easier to miss over the last decade because so much of the conversation has been organized around product-led growth. Product-led growth was an important correction. It forced companies to take product experience seriously as a driver of adoption, activation, and expansion. It shifted attention back toward the user and away from the old fantasy that distribution could indefinitely compensate for weak software. It was, in many ways, exactly the right response to a world of bloated enterprise products and top-down selling.

But product-led growth was never the whole story. It described how a product could become a more effective mechanism for acquiring and activating users. It did not fully answer the harder question of what happens next. It did not tell us why some products become enduring companies while others remain isolated successes. It did not explain why some products produce leverage for the company behind them while others produce affection without durable power.

PLG solved for adoption. It did not solve for durability.

That is where the conversation needs to go next. The real question is not whether a product can attract users. It is whether the company has designed the economic system around the product well enough to convert product value into durable growth, retention, expansion, and strategic power.

Subscribe now

What I Saw at Outlook Mobile

I came to this view through operating experience rather than theory. At Outlook Mobile, I saw what happens when a great product becomes the entry point into a much larger system. Outlook Mobile worked because it solved a real problem elegantly. It was fast, clean, useful, and respectful of the user in a category that had long tolerated compromise. People adopted it because it was simply better. Product quality created trust, momentum, and a wedge.

But the more important lesson was not that a great mobile app can win users. It was what happened because it won them. Outlook Mobile did not remain an isolated product success. It became a product-led entry point into broader Microsoft 365 adoption. The product created pull at the edge of the system, and the company was built to convert that pull into a much larger commercial outcome. What began as product-led growth helped enable product-led sales. The product generated value, but the surrounding architecture turned that value into leverage.

That, to me, is the difference between a great product and a great company. The product creates the opening. The company determines whether that opening becomes a system.

What I Saw at Slack

I saw the same pattern, though from a different angle, at Slack. Slack was one of the great products of its era. It changed how teams communicate. It spread quickly because it solved a real problem in a way that felt obvious only after it existed. It had clarity, craft, and a depth of product instinct that very few companies ever achieve. It also created an exceptional company outcome.

But a great acquisition is not the same thing as an industry-shaping platform. A product can be extraordinary. A company can be enormously successful. And still, the broader lesson remains the same: a great product does not automatically create the kind of economic system that reorganizes a market around itself for decades. That takes something more deliberate than adoption alone.

Slack became incredibly important software. But importance at the level of user experience is not the same thing as inevitability at the level of industry structure. Product excellence and company design are not the same discipline.

The Missing Discipline

Most company-building frameworks assume that once value creation is established, the rest of the system will emerge naturally. It often does not. In practice, company building happens in two phases.

In the first phase, the company discovers how to create value. It identifies a real need, builds a product that meets it, and proves that customers care. This is the phase most startups are built to navigate. The central questions are familiar: does the product solve a meaningful problem, do users adopt it, do they return, and can the company acquire them efficiently enough to keep going?

The second phase begins once the company has proven it can create value. This is where the company has to decide whether that value will remain a product experience or become the basis of an enduring business. The questions change. What behavior does the product produce in customers over time? Which of those behaviors matter economically? How does usage translate into retention, expansion, pricing power, or strategic dependence? How does the business become stronger as the customer becomes more successful?

This is the phase in which companies separate. Some continue to create real value but fail to capture enough of it. Some capture value briefly, but in brittle ways. A much smaller number design systems in which product value, customer behavior, and business economics reinforce each other. Those are the companies that move beyond having a good product. They become structurally stronger as they scale because the product is not operating alone. It is embedded in an economic design.

The Framework: Product-Led Economics

This is the lens I have increasingly come to use to understand the difference. I think of it as Product-Led Economics.

Product-Led Economics is the discipline of designing the system through which product value becomes customer behavior, customer behavior becomes economic outcome, and those outcomes align the company around what matters most.

Products do not create businesses directly. They create value. The question is whether that value produces the right customer behaviors, whether those behaviors produce the economics the business needs, and whether Product, Engineering, Design, GTM, and Finance are aligned around reinforcing that system.

Most companies start with the product and hope the economics follow. Product-Led Economics starts from the opposite direction. It begins with the economics the business needs to produce, works backward to the customer behaviors required to create them, and then asks what kind of value design inside the product is most likely to drive those behaviors.

When those layers reinforce each other, product value begins to compound into what Product-Led Economics calls an economic engine. A system that continuously converts customer value into strong unit economics and, over time, durable market power.

Value Is Not Enough

Once you start looking for this pattern, you see it everywhere. Apple did not become Apple because it made beautiful devices. Plenty of companies have made beautiful devices. Apple built an ecosystem in which hardware, software, identity, services, payments, and developer participation reinforce each other. The product was the point of entry, but the company was built in the relationships between the products.

Costco did not become Costco because it sold low-priced goods in warehouses. It built a membership model that changed the economic logic of the relationship and aligned customer behavior with the company’s advantage. In each case, the product mattered. In neither case was the product alone the explanation.

That is why so many conversations about strategy stay too close to the surface. They focus on what the product does, but not on what the product causes. They focus on value creation, but not on the chain of behaviors through which value becomes an economic outcome. And they focus on the experience of the user without asking what that experience makes possible for the business over time.

A product can create enormous value and still remain economically weak if that value does not produce the right downstream effects. Users may engage without converting. They may adopt without expanding. They may love the experience without deepening their dependence. They may return without ever becoming more valuable customers.

The real question is not simply whether a product creates value. It is whether the product is designed so that the value it creates produces the customer behaviors and organizational alignment required for a durable business. Product-Led Economics begins there.

Subscribe now

Why This Matters More in the Age of AI

This is a systems problem, not a product problem. It sits at the intersection of product, growth, sales, pricing, packaging, success, and finance, which is precisely why so many companies miss it. Each function optimizes its own slice of the business, but too few leadership teams step back and ask whether the whole system is designed to turn product value into compounding advantage.

When that coherence is missing, the company leaks strength in ways that are hard to diagnose if you are looking only at the product. It may have strong usage but weak monetization because the path from utility to willingness to pay is poorly designed. It may have a loved product but no real expansion path because the user, buyer, and budget owner sit in different places. It may acquire customers efficiently but fail to retain them because the value it delivers is episodic rather than cumulative. In each case, the product may be doing exactly what it was designed to do. The problem is that the surrounding system was not.

That matters far more in the age of AI. As software becomes easier to build, product quality alone becomes a weaker source of advantage. Functional products will be cheaper to create. Polished experiences will appear faster. Features will be copied more quickly. More companies will be able to build something good enough to win initial attention.

In that world, product quality becomes the price of admission. System design becomes the source of advantage.

The winners will not simply be the companies with the best product. They will be the companies with the best-designed system around the product. They will know how to turn value into retention, adoption into expansion, and product strength into economic power. When more companies can build useful software, the advantage shifts to those that can design what that software does next.

From Product Quality to Economic Design

That is the discipline that sits beyond the usual conversation about product-market fit or product-led growth. Product-market fit tells you that you have found a real problem and built something people want. Product-led growth tells you that the product can help drive adoption. Both matter. Neither is enough. The real question is whether the company has designed the system around the product well enough to turn value into something durable.

That is what I mean by Product-Led Economics. Not a theory of product quality, but a discipline for turning product value into durable economic power.

Great products do not build great companies. They create the possibility of one. What determines the outcome is everything that follows: the behaviors the product creates, the economics those behaviors support, and the alignment inside the company required to reinforce them.

That is the shift. Not from product to growth, but from product quality to economic design. Great products may win users. Only well-designed systems win industries.

If you only watch revenue, you are always late.

That’s true in startups. It’s true in scale-ups. It’s true in public companies. And it’s true in retail right now.

When cycles turn, revenue doesn’t collapse first. Behavior does. The problem is that most leaders are trained to watch the wrong signals. Revenue. Growth rate. ARR. GMV. Topline momentum. Those numbers are clean, reportable, board-friendly. They are also lagging indicators.

By the time revenue rolls over, the structural weakness has already been building for months.

Every downturn I’ve lived through follows roughly the same sequence.

First, behavior shifts. Then unit economics soften. Revenue holds for a while. And then, eventually, revenue breaks.

Most teams notice it at the last stage. The disciplined ones see it at the first.

What Deteriorates First

It’s not revenue that deteriorates first. It’s the economics inside each customer relationship.

Repeat frequency softens. Conversion tightens slightly. Promotional reliance increases. Acquisition costs creep upward. Margin per unit compresses.

Nothing collapses. Everything strains.

The business still “looks” healthy. The dashboard still trends up and to the right. But the engine underneath is running hotter than it used to.

Retail is a useful live case study right now.

On paper, the macro environment looks stable enough. Nominal retail sales are still up year-over-year. Unemployment remains low. Markets are strong. If you scan headlines, there’s no obvious crisis.

And yet, if you talk to operators, the tone is different.

Repeat purchase isn’t as predictable. Customers are taking longer to convert. Promotions are doing more of the work. Paid acquisition feels heavier than it did twelve months ago. Margins don’t stretch the way they used to.

Revenue hasn’t fallen. But density has.

That’s how downturns start. Not with dramatic headlines, but with compounding micro-frictions.

Subscribe now

Consumers don’t suddenly stop spending. They hesitate. They comparison shop. They stretch purchase cycles. They become more selective. That selectivity shows up in cohorts long before it shows up in earnings.

If your 60- or 90-day repeat intervals are drifting outward, that isn’t churn. It’s caution. And caution compounds.

To defend volume, businesses lean a little harder on promotions. Discounts get slightly deeper. A larger share of revenue becomes incentive-driven. Pricing power weakens at the margin.

At the same time, acquisition costs rise just enough to be uncomfortable. Conversion rates soften a few points. It takes more effort to generate the same order.

Individually, none of these signals trigger panic.

Collectively, they compress contribution margin.

And contribution margin is where recessions begin inside a business.

Scale vs. Resilience

Expansion cycles reward scale. Traffic, reach, paid efficiency, rapid acquisition. Those levers look brilliant when demand is abundant and conversion is forgiving.

Compression cycles reward resilience.

Resilience looks less glamorous. It shows up in repeat behavior. In margin discipline. In cohort visibility. In the ability to influence customer behavior without paying to reacquire the same person over and over again.

Most companies optimize for scale when conditions are favorable. Few build resilience while revenue is still growing.

That’s the mistake.

Because resilience is slow to build and fast to need.

When demand is expanding, you can rent growth. You can buy traffic. You can tolerate inefficiency. You can subsidize acquisition and hide it inside topline expansion.

When elasticity rises (when customers become more selective and price-sensitive) rented growth becomes fragile. You pay more for customers who convert less reliably and return less frequently.

That’s when infrastructure matters.

Infrastructure isn’t a feature. It’s not a campaign. It’s not a growth hack.

It’s the set of systems that give you direct visibility into behavior, control over engagement, and leverage over repeat. It’s the ability to influence the customer relationship without resetting it every time a session ends.

In retail, that might mean mobile as an owned channel. In SaaS, it might mean product-led retention and expansion loops. In marketplaces, it might mean depth of supply and embedded workflows.

The form changes. The principle doesn’t.

Owned leverage beats rented growth in selective cycles.

A Simple Diagnostic

If you want to know whether your business is entering a margin recession, don’t start with revenue.

Start with repeat interval trends. Look at contribution margin per cohort. Watch CAC quarter over quarter. Measure the percentage of revenue that requires an incentive.

If those are softening, revenue will eventually follow.

It always does.

The most dangerous cycles are the quiet ones. The ones where revenue still looks fine and the board feels calm. The team keeps pushing growth because nothing appears broken.

Until margin gives out.

The companies that outperform aren’t the ones that react fastest to revenue decline. They’re the ones that respond earliest to behavioral signals. They invest in infrastructure before it becomes urgent.

That’s what “beyond the build” actually means.

It’s not shipping faster. It’s not adding features. It’s not chasing traffic.

It’s strengthening the underlying economics of the business while the surface still looks healthy.

Revenue flatters you.

Density protects you.

If you’re only watching the first, you’ll miss the turn.

If you’re building the second, you’ll survive it.

Subscribe now

Building Outlook Mobile at Microsoft nearly failed. More than once. In hindsight, it looks inevitable - Microsoft, email, hundreds of millions of users. It seems obvious, but it wasn’t. It was one of the hardest product challenges I’ve ever worked on, and it completely reshaped how I think about mobile, engagement, and growth.

It also explains why most consumer brands are underestimating what it really takes to build a successful app. And why they shouldn’t try to do it alone.

The Myth: “We Just Need an App”

When marketers talk about building an app, the conversation usually starts with surface-level questions: What features should it have? How long will it take? How much will it cost?

Those aren’t the hard questions. The hard questions are the ones you don’t even know to ask until you’re deep into it. The hard questions are operational. Who owns the funnel? How will we instrument behavior? What actually drives habit? How fast can we iterate without breaking the experience? What actually drives revenue?

At Outlook Mobile, we had to solve four complex problems, simultaneously. All while integrating into the Microsoft ecosystem and culture and building to become “Enterprise Grade”.

1. Alignment Is the First Gate to Velocity

The first challenge wasn’t code. It was people and their alignment. We were lucky in that the core team of engineers was already present - world class engineers who went on to build incredible products at companies like Instacart.

But we had to move beyond engineers. We had to assemble a cross-functional team that could move fast while fixing things.

We needed instrumentation, analytics, and a framework to understand them. We needed growth leadership, both for demand generation and in-app engagement. And we needed clear shared metrics to ensure decisions were made quickly.

Product velocity is dependent upon decision velocity.

Without shared metrics of success, opinion creeps in. Politics follows opinion. And before you know it, meetings run into hours and velocity dies. Your product dies. Your business dies. You’re done.

We set three audacious north star metrics:

1. 100 million MAU

2. Five star app rating

3. Shipping to both app stores every week

And one grounding principle. To build an app “Loved by users. Trusted by IT.”

That clarity and precision gained immediate alignment and mattered more to the product’s success than any other single feature we shipped.

Build Your Own Mobile App

2. Assumptions are Expensive

We shipped features we thought were valuable. But users didn’t behave the way we expected, and features we thought were critical were ignored.

For example, a sophisticated Files experience. We assumed mobile users struggled with document workflows. And so we had a Files tab at the bottom of the screen alongside “Mail” and “Calendar”.

No one used it.

We joked that it was the place code went to die.

We also thought writing emails was important. A reasonable assumption to make, given the application and what it was for, right? Wrong.

When we instrumented the app, we saw the average session length of a user was 24 seconds. That single metric changed everything.

Outlook Mobile wasn’t a composition engine. It was a consumption engine. Users weren’t writing emails. They were triaging it.

Without instrumentation, we would have been optimizing the wrong thing for years. That’s the cost of building blind.

3. Engagement is Designed, Not Discovered

The (relatively) easy part about building an app is getting someone to download it. The hard part is getting the user to build a habit around it.

To get to our first 10 million MAU, we obsessed over onboarding. Making it as quick and easy as possible. We looked at all of the key workflows and removed friction relentlessly. We experimented with push notifications, ensuring we were only sending ones that added value, not noise. And we shipped constantly without destabilizing the product.

And we did that while competing against Apple and Google, whose apps are pre-installed on the device. So we had to build a habit around Outlook Mobile, while breaking a habit built around a product and ecosystem they’d been using since they bought the phone.

Engagement doesn’t happen by accident. It’s engineered through relentless attention to detail.

4. The App Store is Not Distribution

Even inside of Microsoft, we had almost no central marketing support. We didn’t even have a clear funnel from Microsoft.com to the app.

And so, like everyone else, growth had to come from building the best app in the category. This earned us press attention that gave us credibility. We got the app rating above 4 stars and that earned us features on the app stores.

All of this created curiosity. Curiosity we had to convert into loyalty through instrumentation, interviews, and rapid iteration.

The app store isn’t a distribution channel. It’s a conversion funnel you have to drive. And it’s ruthless.

Now Imagine Doing That as a Retail Brand

If this was hard inside Microsoft, with engineering depth, infrastructure, and brand reach, imagine doing it as a $150M apparel brand. Retailers don’t wake up wanting to become mobile software companies.

Top of mind for them are core business metrics. Higher repeat purchases, stronger loyalty engagement, lower blended CAC, clearer channel attribution, and higher LTV.

But those outcomes require a strong brand identity, purchase funnel instrumentation and ownership, continuous experimentation and iteration. And most of all, operational alignment across all of these motions.

That is a mobile operating model. If this nearly broke inside Microsoft, what do you think it does to a retail org built around merchandising and paid acquisition?

This is Why Bryj Exists

If building Outlook Mobile required alignment, instrumentation, iteration, and relentless focus on habit, retail brands shouldn’t have to recreate that machine from scratch.

Bryj exists to give you the operating model without forcing you to become a software company.

We do it in two ways.

Bryj Beam: Start Fast. Own the Funnel.

Bryj Beam gets you live fast. But more importantly, it gets you instrumented from day one, so you don’t launch blind. You know who installed but never purchased; who purchased once but never came back; where users drop off in the funnel; and which campaigns actually drive LTV.

Identity is built in. Event tracking is wired, and engagement tools are native.

Your marketing team can test, learn, and iterate weekly without waiting on engineering cycles. You can trigger the right message to the right segment at the right time. You can see what worked, and double down.

Beam turns your app into a measurable growth channel. Not a brand accessory. Not a loyalty wallet.

A revenue engine.

Bryj Managed: Sophistication at Scale

Some brands need more than speed. They need deep CRM and POS integrations; custom UI that reflects the brand precisely, complex backend logic, global scale, and enterprise-grade security

Bryj Managed delivers that level of sophistication, without losing the growth engine underneath.

You get a custom-built experience where it matters, and infrastructure that scales. You get ongoing optimization, not a one-time build. And critically, you still get the instrumentation, funnel visibility, and engagement infrastructure that drives repeat purchase and LTV.

Managed builds the foundation.

Beam drives the compounding growth.

That’s the difference.

You don’t have to become a mobile software company.

You just have to decide whether you want to own the relationship, and operate it deliberately.

Subscribe now

What This Means for Marketing Leaders

If you’re a CMO, Head of Growth, or Digital leader at a consumer brand, here’s the uncomfortable truth. You are already being measured on metrics that require a mobile operating model.

But most brands are trying to improve those metrics with rented channels. Paid social. Search. Marketplaces. Email blasts and LinkedIn fighting for inbox survival.

You don’t control those channels. You don’t own identity inside them. And you don’t see the full picture within those channels. They’re a black box.

When we instrumented Outlook Mobile properly, everything changed. Decisions got faster. Assumptions disappeared. Growth became deliberate instead of hopeful.

That’s the real shift.

It’s not “do we need an app?” It’s:

Do we want to own our customer relationship, or rent it?

AI Accelerates the Mobile Urgency

And here’s where it gets more urgent. AI is increasing abstraction.

Consumers will increasingly ask assistants what to buy. They will let algorithms choose options. Product will be discovered through intermediaries, and your website will be skipped completely.

AI agents, marketplaces, aggregators - they all sit between you and your customer.

Every layer of abstraction reduces your visibility. Every intermediary weakens your relationship. Every algorithm owns more of the decision. And this is not slowing down, it’s accelerating.

Essentially, websites become endpoints and aggregators become gatekeepers. CAC gets out of your control and becomes more volatile. Attribution becomes more opaque unless you own a direct channel.

Inside an app, you own identity. With identity, you own behavior. Once you track behavior, you can gain insight. And with insight, you gain leverage.

That leverage compounds. Again, this isn’t about “having an app.”

It’s about defending and strengthening your relationship with your customer in a world where AI is making that relationship easier to lose.

The question isn’t whether mobile matters. It’s whether you want to rent your customer relationship, or own it, as AI tightens the funnel.

Because the brands that own identity and engagement will outlearn, out-iterate, and outlast the ones that don’t.

That’s what building Outlook Mobile taught me.

And it’s the decision every consumer brand needs to make. Sooner than they think. Because the next five years won’t reward brands with the best ads. They’ll reward brands with the strongest direct relationships.

Microsoft is trying to monetize AI through engineering brute force instead of user experience. And that gap between capability and usability is where value goes to die.

The headline that everyone sees is that there’s growing skepticism about whether enterprise AI can actually deliver the returns implied by today’s valuations. On the other side, there’s the sheer scale of capital being committed to data centers before the revenue model is truly proven.

But Microsoft’s problem isn’t an AI problem. It’s a user-experience problem. It’s a product culture problem.

The value of AI is real. We’ve already seen it at the leading edge. Early adopters who are genuinely changing how they work every day. I’m one of them. I use AI constantly. It works.

The real question is how that value gets distributed to the early majority, and how Microsoft captures it as durable revenue.

Microsoft’s Strengths

Historically, Microsoft’s strength has come from two places. First, engineering. The platforms behind Microsoft 365 are extraordinary. The backend systems, scale, reliability, and security are insane.

Second, the field organization. Sales and customer support at Microsoft is a machine. Having seen it firsthand, I can say it’s one of the most metrics-driven, strategically disciplined GTM engines in enterprise software. From deal close to activation to usage to expansion, it’s executed with near-perfect rigor.

But neither of those strengths guarantees success in the AI era. Because AI doesn’t create value in isolation. It creates value inside the product experience. And that’s where Microsoft is struggling.

Microsoft is Still Shipping its Org Chart

Once you move past early adopters, tolerance for friction collapses. Complexity, unreliability, and “weirdness” are no longer acceptable. The early majority doesn’t want to learn AI. They want it to quietly make their work easier.

Yet when you look at Outlook, Word, PowerPoint, Excel, it’s hard to argue that the experience has meaningfully evolved in years. More worrying than the stagnation of individual apps is the lack of progress in how they work together.

The seams are still painfully visible. You can see organizational boundaries at every click.

File-centric workflows. Awkward transitions between apps. Collaboration that feels bolted on instead of native. The vision around “flow” is directionally right, but flow isn’t branding. It’s lived experience.

Layer AI on top of that, and the problem compounds.

Copilot as a Passive Observer

Copilot is present. It’s visible. But in practice, it rarely delivers value in a way that feels natural or indispensable. Worse, it’s teaching users to ignore it. And once AI becomes background noise, monetization becomes nearly impossible.

That’s the concern of investors. That’s the fundamental challenge Microsoft faces right now: how to make AI feel inevitable, and not optional.

Solving that requires design leadership. End-to-end ownership of the experience across products. Clear decision-making authority rooted in human workflows, not system architecture.

From the outside, it still looks like engineering has the steering wheel. And while engineers are exceptional at building platforms, they rarely optimize for user experience. The result is software that is powerful, but not delightful. AI that is capable, but not habit-forming.

Product Culture is the Constraint

The good news? If anyone can change that culture, it’s Satya Nadella.

I watched him do it before, and it was incredible to see. He took Microsoft from defensive, inward-looking, and platform-protective to cloud-first, open, and developer-friendly. That transformation reshaped the company and its valuation.

This moment calls for another one.

Because in the AI era, engineering excellence is table stakes.

Design-led experiences are the differentiator.

Subscribe now

Hi! I’ve been quieter here than usual.

Over the past couple of months, I stepped away from writing to focus on building Bryj. I joined for one reason: to help democratize the building of mobile apps.

I learned the hard way how difficult and expensive mobile can be. While leading product for Outlook Mobile at Microsoft, I saw firsthand what it took to build, scale, and maintain a world-class app. The teams, the cost, the complexity. Most brands don’t have the technical experience, and could never afford it. For a long time, that was fine, because a website was enough.

That’s no longer true.

AI has changed how consumers discover, decide, and buy. The website is no longer the center of the relationship. And that makes a mobile app no longer a “nice to have,” but an existential requirement for brands that want to survive the next decade.

This piece explains why, and why now every brand needs a mobile app, not just the few who can afford to build it the hard way.

Subscribe now

How AI Changed Everything for Marketers

Since the web started to gain traction for retailers 30 years ago, the website has been the center of gravity for marketing.

It’s where brands told their story. Where campaigns landed. Where traffic converted. Where revenue was generated, and where attribution lived.

Marketing strategy assumed one core truth: If you could get someone to your website, you controlled the relationship.

Since ChatGPT burst onto the scene in 2023, that assumption is no longer true. Not because the website disappeared, but because AI stepped in between consumers and the web.

In the “old days”, marketers worked hard on Search Engine Optimization and paid search advertising to ensure the gatekeeper of the web, Google, would put their website at the top of the first page of consumer search results. The “blue link” on Google was the top of the funnel. Once marketers got that secured, they could work their campaigns to move prospects through their funnel.

Today, consumers aren’t going to Google as much. According to a Datos/SparkToro report, Google searches are down 20% The phenomena is worldwide, but best exemplified by the impact on Hubspot, who saw a 70-80% drop in organic traffic between 2024 and 2025. The drop in organic traffic was noted by Hubspot’s CEO Yamini Rangan, who stated on LinkedIn, “If HubSpot, with one of the best SEO teams in the world, can experience this, none of us are safe”.

There are two ways marketers can respond to this phenomena, because it’s only going to get more pronounced.

1. You can invest in GEO - Generative Engine Optimization

2. You can invest in a mobile app

Everyone is scrambling to implement GEO, but it’s a moving target. If you think optimizing for Googles PageRank algorithm was hard, optimizing for discovery by an LLM whose creators don’t even know how it works is going to take a lot of investment, iteration, and learning. Read, time and money.

A mobile app is how brands reclaim the relationship, owning identity, engagement, and loyalty in a world where AI mediates everything upstream. It also enables brands to meet consumers where they are - on their phones!

AI changed who owns the customer moment

How did we get here? And how did we get here so quickly? Traditionally, marketing worked like this:

1. Capture attention (ads, SEO, social)

2. Drive traffic to your site

3. Educate, persuade, convert

4. Retarget and repeat

Your website was the hub, and every dollar you spent was designed to push people toward it. AI collapses this entire system.

Today, consumers increasingly ask AI what to buy, which brand is best, to compare options, and to ultimately ask AI to take action → buy! Consumers don’t browse, don’t research, don’t click ten tabs. Consumers ask, and the AI decides. And when AI resolves the decision, your website never gets a chance to do its job.

AI disintermediates brands from consumers

As AI now sits between your brand and your customer, it is a filter you have no control over. AI is not just providing your customers with a link to your products. It is providing your customer with a narrative, a perspective, a feeling even.

AI has converted the pass-through sign post that was a Google search result and turned every interaction into a conversation about your brand. A conversation you are not in!

When consumers stop browsing and start asking, traffic thins, brand recall weakens, attribution disappears, and differentiation shifts upstream into the AI.

The website doesn’t vanish, but it becomes supporting infrastructure, not the center of the relationship.

So the real question becomes unavoidable.

If the website is no longer where the customer relationship lives… where does it live?

Learn From Microsoft’s Mistake

For many marketers, a “mobile first” experience misses the point. They’ve been trained by platforms like SquareSpace and Substack that it’s just a responsive design of your website and a better checkout UX to fit on a smaller screen.

I learned this lesson first hand. It’s the same mistake Microsoft made when they developed their first Outlook Mobile application. They thought they could just shrink down the functionality of Outlook desktop into a smaller interface and you’re done. Wrong. That app flopped.

After the acquisition of Acompli, we scrapped the old app and rebranded Acompli as Outlook mobile. Within a year, we went from zero to 30M monthly active users.

At the core of this decision is a grounding principle. Mobile isn’t an afterthought. It’s the center of your digital go-to-market and monetization strategy.

What Mobile Gets You

A mobile app gives you a known customer, persistent identity, first-party data, and direct reach and connection. You’re in your customers pocket able to reach out, connect, and build a relationship on your customers terms and in a way that serves your brand. Through push notifications, personalization, saved preferences, and habitual engagement.

These aren’t features. They are relationship control.

In a world where AI mediates discovery, the only way to protect brand value is to own the relationship after the decision is made.

That ownership lives on mobile.

AI becomes the acquisition layer - Mobile becomes the brand layer

All of this requires a change in mindset. Increasingly, AI will influence what consumers consider, recommend where they buy, provide a shortcut comparison, and automate decisions.

You cannot outbid an AI. You cannot out-optimize a prompt. You cannot funnel-hack a conversation you don’t participate in.

But you can own the experience that follows. With your mobile app.

Your mobile app will anchor your brand identity. It will create loyalty through interactions and the building trust. Your app will help create engagement and purchasing habits. All of which will compound as lifetime value - for you and your customer.

All of this goes to say that the future marketing stack looks like this:

• AI → discovery and recommendation

• Mobile → relationship, loyalty, and repeat value

• Web → supporting surface, not the core

Web-first marketing optimizes for clicks. Mobile-first marketing optimizes for trust

This distinction between clicks and trust is critical. Web-first marketing is an exercise in statistics and the power of large numbers. How do we get more traffic? How do we improve conversion? How do we lower CAC?

A mobile-first marketing person has to be more user-centric. Building the intangibles that will convert into revenues over time. Why should a customer stay connected to us? What ongoing value do we provide? How do we earn a permanent place in their behavior?

What this means for marketing leaders

This shift forces a rethink of what marketing success actually means. You have to go beyond sessions, page views, CTR, conversion rate, etc.

You need to think in terms of repeat engagement, owned audience, first-party signals, lifetime value. Monthly active usage and engagement within your mobile app funnels.

The old metrics don’t go away. You need to build a model above that and start thinking in the way mobile app developers have for a long time.

The brands that win will own the relationship layer

AI won’t replace brands. But it will commoditize the ones that don’t own their customer connection. The brands that will survive and thrive reduce dependence on rented attention (ads and search sponsorship), invest in direct, owned experiences, and treat mobile as a strategic asset, not a technical project.

In the AI era, this means that websites will inform users and AI LLMs. AI will decide, but it will be a mobile experience that will connect your customers to you, your product, and your brand.

That’s the new reality.

AI didn’t kill the website - It ended its reign

Your website still matters. But it no longer defines your brand. AI now controls discovery. Mobile must control the relationship.

If your marketing strategy is still web-first, you’re optimizing for a world that no longer exists. And your customers have already moved on.

Get Your Mobile App

Every growing product company eventually hits the same wall. Customers, prospects, and sales reps generate an endless stream of feature requests. Engineering capacity feels stretched. Marketing wants fresh stories to tell. Leadership is pushing for revenue growth, shorter sales cycles, and clear differentiation to drive up average contract value or sales prices.

The result, too often, is a roadmap that feels more like a reaction log than a strategy. The result is product teams are driven to implement features from the the customer or account executive who pushes most persistently (Squeaky Wheel); the person in the room with the biggest paycheck (HiPPO); or the last customer someone spoke to (recency bias).

If this is you, you’re in a feature factory. It is not sustainable, and the opportunity costs are enormous.

A roadmap should be a strategic asset. It should capture where the company is focusing, what customer problems are worth solving, and how product investment will drive business outcomes. The challenge is to take the long flat list of features the company could build and turn it into a thought-through roadmap of what they should build to meet the company’s goals.

Using examples from my experience at Slack and elsewhere, we’ll walk through a detailed process to turn chaos into clarity. A long list of seemingly disconnected features into a coherent roadmap.

Here are the steps I’ll detail. (🔒 for our paying community subscribers.)

1. Market segmentation: deciding who you’re building for and get alignment across the product and go-to-market organizations

2. 🔒 Product Domain Mapping: how to get out of the weeds of features and focus on what matters

3. 🔒 Jobs-to-be-Done: how to become customer-centric and business value-oriented

4. 🔒 Kano value stack classification: how to make tradeoffs explicit and intentional

5. 🔒 Goals, OKRs, and KPIs: how to align customer needs and business goals

6. 🔒 Bring it all together: decide how to allocate investments across the value stack

7. 🔒 Build the roadmap: where prioritization becomes real by balancing engineering capacity

By the end, you’ll see how to move from a messy list of asks to a structured system for prioritization that is repeatable, transparent, and strategic.

Why Every Product Company Needs a Prioritization Framework

Ideas are cheap and easy to come by. The difference between company success and failure is the speed of execution. And the greatest drag on execution is the speed and clarity of decision-making.

Without a clear decision-making framework, roadmaps become dominated by short-term pressure. This may feel like a successful approach as code is shipped and some sales are made. But it’s not sustainable as over time growth slows and frustration builds across the organization:

• Sales feels blocked in deals.

• Customer success struggles to manage churn.

• Engineering suffers from whiplash as priorities shift.

• Leadership sees a limited connection between roadmap items and financial outcomes.

In short, you’re living inside a Feature Factory at 4% annual revenue growth. And from the investors’ perspective, you’re building a lifestyle business and not a “rocket ship” that will deliver outsized returns on their investment.

A prioritization framework doesn’t eliminate tradeoffs, it makes them transparent. Instead of fighting over features, the team debates segment focus, value creation, and alignment with company goals. More than anything, it’s these debates and the decisions made where the value lives. These tough conversations are what build a healthier organization delivering better outcomes.

So, how do we do this? Read on!

Step 1: Define Market Segments

The first step is to get clarity about who you are serving now. Different customer segments have different requirements, costs to serve, and revenue potential.

For example:

• An enterprise company with complex infrastructure may care deeply about integrations, compliance, and security.

• A mid-market business using modern SaaS stacks may prioritize speed of setup and in-product dashboards.

In the early days of a company, when founder-led sales dominate and you’re in search of Product-Market Fit (PMF), you’ll take any customer meeting, close any deal, and build any feature just to show traction and validate hypotheses.

That’s great, to a point. Once you’ve gathered all of this data, you need to take a step back and ask yourself, “Who do we serve now?”. Even if you’re product is horizontal, you need focus to direct your finite resources. Product teams need to focus on a coherent set of capabilities to build, and your go-to-market (GTM) team needs to focus on a coherent set of prospects to increase sales predictability and velocity.

Treating every prospect as “the customer” does not help prioritization and focus.

How to define segments:

• Vertical: industry, e.g., consumer packaged goods (CPG) , fintech, healthcare

• Opportunity sizing: Total Addressable Market, number of prospects, growth profile

• Size: SMB, mid-market, enterprise

• Tech stack: Salesforce vs HubSpot, Slack vs Teams

• Geography: US, EU, global

• Regulatory profile: strict vs light

• Buyer and user mix: marketers, IT, agencies, operations

Anchor each segment with real customer examples. Instead of abstract labels like “tech-forward enterprise,” use “Nike”. It makes planning and conversations more tangible. These are customers you’ve engaged with and understand their needs. Customer names connect the product and GTM teams and facilitate alignment.

The output of this exercise is a one-slide grid showing the different market segments and the key attributes of each. These can be stack-ranked by factoring in opportunity sizing, current product fit, and current GTM fit. Then choose one market segment to focus on for the next twelve months.

Step 2: Map Product Domains

All across the spectrum of company size, I have seen the same problem persist. At the early stage startups and scale-ups like Spendesk, mid-sized late-stage private companies like Envoy, growth rockets like Slack, and the global titan Microsoft. Energy-sapping debates pitting feature against feature with no context other than what each feature does for the immediate customer need.

If you’re building a product, and not just a feature, your users will traverse workflows through your product and across it. These User Journeys are a guide to break down a monolithic experience into different product domains. Each domain addresses a specific customer pain or need and represents a collection of features that, when used in concert, solve that pain, meet that need, and deliver value to the customer.

Using Slack as an example, domains could look like this:

• Onboarding: how to quickly get users into the product

• File Management: how users can easily see, manage, and share files within Slack

• Calls: how users can easily communicate over voice and video within Slack

• Time Management: how users can easily manage their time in Slack

• Personalization: how a user can connect, configure, and otherwise customize Slack

Align on the language for what the product domain actually does. You now have a shared product map and vocabulary. As a first approximation, color each domain with a red, yellow, and green code as an indicator of how confident you are in that domain. This could be in terms of functional completeness, quality, or performance. You’ll refine what this means relative to the goal, but for now it will give folks a common sense of how the product is doing from an engineering perspective.

Crucially, instead of debating narrow features, the team can now discuss whether to strengthen “Time Management” or “File Management” domain and gain alignment across product, marketing, and sales about where you should invest and when without getting bogged down in the specifics of individual feartures.

Step 3: Capture Requirements as Jobs-to-Be-Done

Customers don’t buy features, they hire products to do jobs in the real world. They don’t want to buy a drink, they want to not feel thirsty. That’s why Jobs-to-Be-Done (JTBD) is a powerful framing to push your conversations and decision making away from your product and features and to the user and their needs. This is how you truly actualize customer centricity. Through deep empathy and understanding of the user’s context.

Within each defined domain, and for the example customer in your target market segment (Nike, in our example), capture the top two or three JTBDs. Larger companies like Slack and Microsoft have user research teams that can perform this exercise exhaustively (and we did). The reality is, for an earlier stage company with finite time and resources, you should be able to capture a tight list of JTBD within a focused brainstorming session involving folks who are close to the customer. Product Manager, Customer Success, Product Marketing, etc.

There are three components to a JTBD. The situation, the motivation, and the desired outcome. For example, “after running, I want a drink, so I no longer feel thirsty”.

In our Slack example, it looked something like this.

To rank and select your primary JTBD, consider the customer pain/need and the frequency they feel it. The higher the pain, the more valuable the solution. The more frequent the pain is felt, the higher the utility of the solution. A high-pain frequently felt is a valuable JTBD and somewhere resources should be focused to deliver a high-value solution that will increase your ACV, for example.

Once you’ve selected your JTBD, take a pass over the list and mark each with a simple red, yellow, green traffic light to indicate how well the product meets that need today. That gap analysis gives a clear and visual way to reveal your product strengths and weaknesses relative to your customer’s needs.

But don’t leap to any conclusions yet. We still have some important steps to go through. We need to understand and balance the value of the solution to the customer and the value of the solution to you - the company and business!

Step 4: Apply the Kano Model

Not all product gaps or JTBD have equal impact. On the customer or on the business. We need a methodology that enables us to map each opportunity relative to the potential return so we can rank and then plan what to build.

For this exercise, we use the Kano model. (I’m consistently shocked by how useful this model is for decision making and how little it is known and applied. Why is that?)

In summary, the model maps capabilities (JTBD) against two axes. The x-axis represents the degree to which the capability is built or delivered. On the far left, the feature is absent or poorly implemented (“missing”). Moving rightward, the feature is present, basic, functional, and eventually best-in-class or highly polished. It’s essentially the supply side: how much capability the product provides in the feature.

The y-axis represents the emotional response or satisfaction the customer feels from the presence of the capability. At the bottom, the feature causes frustration or dissatisfaction if missing or poorly done. In the middle, customers are neutral, they don’t notice or care much about the feature. At the top, the feature creates delight and excitement, often going beyond expectations. It’s essentially the demand side: how customers perceive the value of what’s provided.

Within this framework of satisfaction vs functionality, there are three types of capabilities or features. Table stakes, Satisfiers, and Delighters.

Some features are mandatory, others create linear value, and a few create delight. The Kano model is the tool for sorting them.

• Table stakes: If missing, the product feels broken. Compliance, exports, and reliability often fall here. Once covered, more polish adds little incremental value.

• Satisfiers: Direct levers of value. Investment translates proportionally into customer satisfaction and willingness to pay. Think dashboards, speed, or automation.

• Delighters: Unexpected, memorable features. They create pull and pricing power. For example, an AI “coach” suggesting next best actions.

By classifying requirements this way, teams can avoid overinvesting in Table Stakes or chasing endless “nice to haves.”

Here’s how we thought about “calls” in Slack and how it could be mapped into the three Kano categories with distinct use cases.

Table Stakes

Use Case: Basic One-to-One Calling

• Situation: A user needs to escalate a quick conversation from text to voice.

• Expectation: They can click a button in Slack and instantly start a basic voice or video call with another user.

• Why it’s Table Stakes: If Slack didn’t have this, the product would feel broken compared to alternatives like Teams or Zoom. But once the feature exists and works reliably, adding more polish doesn’t meaningfully increase perceived value. ie. It doesn’t close more deals or increase price.

Satisfier

Use Case: Reliable, High-Quality Group Meetings

• Situation: A team of 6-10 people schedules a weekly check-in and decides to run it entirely within Slack.

• Expectation: The call has reliable connectivity, screen sharing, recording, and integration with Slack channels so discussions stay linked to ongoing work.

• Why it’s a Satisfier: The better the quality, stability, and feature richness, the more valuable the experience. And the more likely customers will adopt Slack as their primary communication hub. Performance scales linearly. Every improvement drives more satisfaction and willingness to use Slack over Zoom or Meet. More deals are closed and higher prices can be charged once the perceived customer satisfaction measure has cleared zero.

Delighter

Use Case: AI-Powered Meeting Summaries in Context

• Situation: A manager joins a Slack call but can’t stay for the full meeting.

• Expectation: After the call, Slack automatically generates a summary, action items, and links them back to relevant channels and files.

• Why it’s a Delighter: This goes beyond what most users (used to) expect from a call tool. It feels like magic. Saving time, improving productivity, and increasing Slack’s stickiness. Over time, of course, competitors may catch up, and this feature could migrate into the “satisfier” category. Which is the case with all of the AI notetakers now on the market. And why these companies are in danger of becoming “features” rather than sustainable products and companies.

Note that within each of these use-cases there are dozens of individual features. Hundreds of user stories and backlog tickets. It’s important to keep at the high-level and out of the weeds of individual features.

Step 5: Balance with Goals, OKRs, and KPIs

Now we have product capabilities mapped against the value customers would perceive the solution to have. Table Stakes remove blockers to close deals. Satisfiers deliver incremental value that help us penetrate more deeply into customers. And Delighters differentiate us from competitors and enable us to position and price as a premium provider. But we can’t do it all.

With competing priorities and resources, we need goals to break the deadlock. An OKR, Object and Key Result, is a measurable goal. And the company needs them to inform decisions and bring clarity to “the why” we’re building something for the business now.

A company of any size should have no more than three OKRs. When I joined Slack, we had eight and it resulted in poor prioritization and a lack of focus in our efforts. When everything is a priority, nothing is. The following year we had three and life was much easier.

It is tempting to have revenue as an OKR. Unfortunately, for a growth startup in particular, that’s kind of like the business equivalent of making “lines of code” a metric of success for engineering. Neither of these metrics help make decisions and both of which can result in poor behaviors.

Decide on OKRs that are going to encourage desired and consistent behaviors in and across teams and ease decision making when conflicts or disagreements arise. For example, increase the number of new customer logos by 20%. Increase our Monthly Active Users (MAU) within target accounts by 25%. Increase our Average Contract Value from $25,000 to $50,000 by the end of the year.

Notice, all things being equal, accomplishing each of these goals will increase the revenue of the company. But in each case we have goals that are actionable and can be accomplished in different ways.

Side note: That last piece is very important. For a company or a team to be able to reach a metric (KR) in different ways is the definition of “Agile”. Keep the target constant and give the empowered team autonomy to adapt and make efficient progress quickly.

Step 6: Bring it All Together

From the top, we now have a defined target market segment with measurable characteristics exemplified by our canonical customer, Nike.

We have a product marketing architecture (”markitecture”) showing domains that collect a coherent set of capabilities or features to help us reason and prioritize.

We have the high-priority Jobs-to-be-Done for each domain so we’re no longer thinking about features, but are now customer-centric in our conversations and our thinking.

We have each JTBD mapped against perceived customer value and the level of completeness in delivering on those customer expectations.

We now have a small, measurable, and well understood set of company goals the whole organization is aligned on.

As a rule of thumb, most SaaS products teams should have a balanced portfolio per planning cycle. 2-3 Table Stakes (close gaps, unblock sales); 3-5 Satisfiers (drive measurable revenue impact); 0-1 Delighters (innovation bets with asymmetric upside).

Table Stakes

You should only commit to as many as necessary to unblock adoption or remove critical objections. These don’t create delight or pricing power, but missing them means the customer will not buy. For most teams, this means a handful at a time (2-3 per quarter). enough to close competitive gaps or enable sales, but not so many that you sink time polishing things customers won’t pay more for.

Satisfiers

This is where you should spend the majority of your investments. These features directly influence ARR, ASP, and NRR (Net Retention Rate) as every incremental improvement increases value. Typically, you’d want 3-5 major performance bets per quarter, often aligned with your key KPIs (Key Performance Indicators (e.g., faster onboarding, deeper integrations, higher reliability). These should map directly to revenue efficiency metrics (ARR/FTE, win rate, retention).

Delighters

Commit to very few at once. These investments are expensive, risky, and often multi-quarter bets. A realistic ratio is 1 big delighter per cycle (or every 2-3 quarters for early-stage teams). Think of these as innovation bets that can justify price uplifts, differentiate in crowded markets, or generate PR buzz. The goal isn’t volume, but impact: one breakthrough delighter can be more valuable than ten incremental improvements.

Step 7: Build the Roadmap

With segments, domains, jobs, Kano categories, and company goals in hand, it is time to translate strategy into execution. This is where prioritization becomes real, because the roadmap is constrained not by ideas, but by engineering capacity.

Allocate Engineering Resources Intentionally

Engineering effort is finite. Each quarter or cycle, teams should allocate resources across three categories:

• Table Stakes: Allocate just enough to unblock adoption or remove critical objections. These do not justify ongoing heavy investment, but they are necessary to close competitive gaps or unlock deals.

• Satisfiers: The majority of engineering capacity belongs here. These are the features that directly drive revenue and usage at lower risk.

• Delighters: Reserve a smaller portion for one big bet. This allocation may span multiple quarters and requires leadership conviction.

As a heuristic, many high-performing teams target something like: 20% Table Stakes, 60% Satisfiers, 20% Delighters. Ratios can shift based on stage of company, competitive pressure, or major customer commitments.

Manage Trade-Offs and Resource Contention

The hard part in this whole process is not identifying opportunities, but making tradeoffs explicit. Some examples:

• Closing deals vs scaling architecture: Do you fix the missing compliance feature to win an enterprise deal, or invest in workflow scalability that serves all future customers?

• Speed vs depth: Do you push out a lightweight version of analytics for mid-market accounts, or double down on advanced reporting for enterprises?

• Revenue this quarter vs strategic differentiation: Do you deliver the one feature blocking several deals, or continue investing in a multi-quarter delighter that could shift your category positioning?

By framing these conflicts in terms of segments, goals, and Kano classification, teams can debate openly and align on tradeoffs without devolving into politics and feature squabbles.

Account for Dependencies Between Domains

Roadmaps are not just lists of features; they are systems of interdependent capabilities. Dependencies between domains and the underlying architecture need to be acknowledged early:

• Foundational layers: Improvements to API, permissions, or data model may be required before higher-level features can be delivered.

• Cross-domain orchestration: Adding workflows often depends on two or more domains being sufficiently robust.

• Analytics and activation: Advanced reporting or AI inference requires reliable data capture, as well as scalable infrastructure for storage and querying.

• Delighters on shaky ground: A flashy AI assistant or advanced automation won’t create lasting value if core reliability or integration gaps undermine trust.

Mapping these dependencies visually helps you plan sequencing. A high-scoring feature may not be deliverable until its architectural prerequisites are in place.

Deliverables of Step 7

At the end of this step, the team should have:

• A quarterly roadmap with P0/P1/P2 items, mapped to OKRs and Kano categories.

• Capacity allocation plan across Table Stakes, Satisfiers, and Delighters.

• Tradeoff log capturing the biggest conflicts discussed and the rationale for the decision.

• Dependency map identifying technical prerequisites and sequencing needs.

This is the moment where strategy turns into execution. The roadmap should not just reflect what the company hopes to build, but what it has consciously chosen not to build, at least not yet. That discipline is what keeps product organizations focused and credible.

The biggest risk for any product team is not choosing the wrong feature, but choosing too many without a framework. Prioritization is not about saying yes to everything, it is about saying yes to the right things with conviction and not yet to the rest with confidence.

The process outlined here gives you the tools to do just that: define your segments, capture jobs-to-be-done, map domains, apply Kano, balance with goals, and build a roadmap that reflects both customer needs, business strategy, and engineering capacity.

The best way to start is simple: gather your product, sales, and engineering leaders for a two-hour workshop this week. Pick one segment, build the first problem × domain matrix, and classify the top three jobs with Kano. Once you take that first step, you’ll have a shared language and a foundation for every roadmap discussion that follows.

Don’t let the perfect get in the way of the good.

Start now!

Product leaders often focus inward on vision, roadmaps, delivery velocity, or customer delight. But the CPO role does not exist in a vacuum. The real measure of product strategy is how well it aligns with the company’s growth engine and capital efficiency. Two people in the executive suite hold the keys to that alignment: the Chief Financial Officer (CFO) and the Chief Revenue Officer (CRO).

These roles are often cast as guardians of constraints. The CFO is seen as the one policing spend. The CRO is seen as the one pushing for short-term revenue. In reality, they are the CPO’s most important allies in building sustainable growth. Each brings a sharp perspective on value creation.

The CFO is focused on capital efficiency, cash flow, margins, and investor confidence. Their language is unit economics and valuation multiples. They want to know: Will this product strategy create more value for every dollar invested?

The CRO is focused on revenue velocity, win rates, deal size, and expansion. Their language is sales funnel metrics and customer outcomes. They want to know: Will this product strategy help my team win more often, faster, and at higher contract values?

If you can show how your product roadmap improves CAC payback, gross margin, win rates, and ACV, your CFO and CRO will not just support your strategy. They will champion it. That is when product stops being “the feature factory” and becomes the growth engine.

Collaborating with the CFO: Speaking the Language of Capital Efficiency

Recent ProductLed benchmark data shows that only 96% of SaaS leaders think PLG is relevant to their business. Yet 58% say their product is PLG ready. This mismatch reveals an enormous opportunity. Most companies want to be product-led, but their products are not designed for it.

For a CPO, this is not just a product readiness issue. It is a capital efficiency issue. And efficiency is the CFO’s love language. Every improvement in capital efficiency strengthens the business and lifts valuation multiples. Investors reward efficient growth, and CFOs know this better than anyone.

So when you sit down with your CFO, you are not just pitching features. You are co-architecting the company’s economic engine. These conversations should cover three areas: where you will invest and why, the expected ROI at a company level, and how your roadmap affects cost structure and pricing power.

The first area is where you will invest and why. A product roadmap is not just a list of features. It is an allocation of scarce company resources. Every investment you propose, whether in infrastructure, onboarding automation, or a new market, represents capital that could be deployed elsewhere. The CFO will want to understand the rationale for these bets. Why this market? Why this capability? Why now? As CPO, you need to show not only the customer problem being solved but also the expected business impact.

The second area is ROI at a company level. CFOs are not satisfied with feature-by-feature justification. They want to know how your strategy translates into revenue growth, margin expansion, or improved cash flow. That means connecting the dots: how a new onboarding flow shortens CAC payback (Customer Acquisition Costs), how a new premium tier raises ASP (Average Selling Price), or how infrastructure investments reduce churn. By grounding product choices in financial outcomes, you give the CFO a story they can share with the board and investors.

Subscribe now

The third area is how your roadmap affects cost structure and pricing power. Every major product decision ripples through the financial model. Automation reduces reliance on costly human support and improves gross margin. Enterprise features open opportunities to raise prices or close larger deals. A new vertical may increase sales costs or require more services, which pressures margins. CFOs think in terms of these trade-offs, and they expect product leaders to do the same.

When you address investments, ROI, and cost structure in language the CFO understands, you transform product planning from a technical roadmap into a financial strategy. That alignment gives your CFO confidence to advocate for your priorities with the board and investors.

CAC Payback: The CFO’s North Star

Among all the metrics CFOs track, CAC Payback is one of the most important. It shows how long it takes for new revenue to cover the cost of acquiring a customer. Shorter payback means more efficient growth.

CAC itself is straightforward. Add up sales and marketing spend (ads, events, and headcount) then divide by the number of new customers. Payback adds time: how many months of gross-margin revenue does it take to earn that money back? If CAC is $1,000 and monthly gross-margin revenue is $50, payback is 20 months. Best-in-class SaaS startups target between 13 to 15 months. According to ICONIQ’s State of Software 2025 report, startups with ARR (Annual Recurring Revenue) of less than $50M should be aiming for ~13 months; and late-stage private companies with ARR greater than $200M are pushing around 15 months. It’s a tight range across the entire spectrum of private companies.

Why does this matter? Investors use CAC Payback as a proxy for growth health. Companies that turn marketing dollars into revenue quickly are seen as more resilient, scalable, and fundable.

For the CPO, there are two main levers.

The first is lowering CAC. Self-serve funnels let prospects experience the product without needing sales involvement. Product-led onboarding removes friction and guides users to their “aha moment” quickly (Time to Value). Viral loops turn existing users into an acquisition engine. Each of these scales with software instead of headcount. This is why PLG matters to the CFO: it allows the company to acquire more users at much lower marginal cost.

The second lever is increasing gross margin. Differentiated features justify higher Average Selling Prices (ASP) and make the product harder to replace. Strong design reduces reliance on costly professional services or custom integrations, which erode margins. Every time you reduce the need for human labor, more revenue drops to gross profit. From the CFO’s perspective, higher margins mean faster CAC payback, more investment capacity, and a stronger story for investors.

Share

Explaining PLG to the CFO

To many CFOs, especially those from enterprise backgrounds, PLG can look like an indulgence. Free tiers, automations, and experiments may feel like diversions from core features. But the most expensive part of CAC is people. Without PLG, sales costs scale linearly with revenue. With PLG, software automates parts of the funnel and changes that equation.

Show the CFO how PLG reduces CAC Payback. With no PLG, median payback is around 20 months. With partial PLG, it drops to 12 months. With core PLG, it shrinks to eight months. This is not just a GTM experiment. It is capital efficiency at work, and that is music to the CFO’s ears.

Gross Margin and Product Value

The other lever in CAC Payback is gross margin. The higher the margin, the faster the payback. And gross margin is a product story. Stronger value increases willingness to pay, raises ASP, and helps sales close bigger deals at healthier margins.

Your CFO will want to know which roadmap bets will increase ASP, how features support premium tiers, and what investment is required to deliver them. Frame your roadmap in those terms, and you provide the CFO with a narrative they can take into boardrooms and investor updates.

This is where the story gets practical. Up to now we’ve explored why CFOs care about efficiency and how product can move metrics like CAC Payback and gross margin. But the real unlock comes when you bring the CRO into the equation.

In the next section, I break down:

• How to build a true partnership with your CRO that makes product a revenue multiplier

• The two metrics CROs live and die by (and how product leaders can directly move them)

• A practical playbook you can use to align CFO, CRO, and CPO strategies into one growth story

That full playbook is available for paying subscribers.

Collaborating with the CRO: Turning Product Into a Sales Multiplier

While the CFO is focused on efficiency, the CRO is focused on revenue velocity. And while PLG may reduce CAC, most SaaS companies at scale need to blend it with Product-Led Sales (PLS). This is especially true when moving into enterprise.

Enterprise sales cycles are long, involve many stakeholders, and require heavy customer success engagement. In these contexts, the product itself becomes a sales weapon. That is why collaboration with the CRO is essential.

At Microsoft, I helped design and implement the company’s first PLS strategy. Outlook Mobile was the tip of the spear to pull through Microsoft 365 upgrades. All driven by an account targeting platform that aggregated data across the app, the Intune SDK, Exchange Online, the Intune backend, and the field’s internally developed CRM.

The lesson was clear: when sales teams use product signals such as usage data, activation milestones, and expansion triggers, they sell more effectively. Deals close faster, win rates rise, and expansion revenue compounds. We went from a standing start to 85% enterprise penetration in two years.

ProductLed’s survey of 600 SaaS companies shows that sales drives 25% of free-to-paid conversions, ahead of product at 18%. In complex products or immature onboarding journeys, sales, customer support, and customer success are indispensable. But sales without product partnership is inefficient. The CRO needs product to deliver signals, proof points, and value props.

Two metrics define the CRO’s world. The first is opportunity-to-won rate. If it is above 20% and stable, the value proposition is strong and objections are manageable. If it is below 20% and falling, the value proposition is weakening, often due to competitive pressure. Marketing may boost pipeline temporarily, but if product truth does not match the message, churn rises. Product must address the recurring gaps that hold win rates down.

The second is Average Contract Value (ACV). If ACV is steady or rising, the product is delivering strong, recognized value. Rising ACV signals that sales can close bigger deals at larger customers. If ACV falls, either customer needs have shifted or competition is winning the value argument. Here product must expand the value footprint with new modules, premium features, or platform extensions.

The CRO does not just want more leads. They want deals that close faster, win more often, and land at higher values. The first way product helps is by sharing usage data. Sales thrives when it knows which accounts are primed to buy. Usage milestones and adoption patterns are the signals. Surfacing these in real time allows sales to focus where it matters most.

The second way is by shaping the roadmap around recurring objections. Sales hears daily why deals stall or go to competitors. Product leaders who use these patterns as input can unblock sales velocity. This does not mean building every feature request, but it does mean addressing the objections that hold the most deals back.

The third way is through packaging and tiering. The way a product is bundled and priced has as much impact on ACV as the features themselves. By aligning tiers with customer segments, the CPO creates natural upgrade paths. Entry-level packages expand reach, while premium tiers unlock larger budgets. When done well, packaging turns product innovation into revenue leverage.

When usage insights, roadmap alignment, and thoughtful packaging come together, the result compounds. Product becomes a multiplier on the CRO’s success. And when the CRO wins, the whole company gains momentum.

Bringing It Together: The CPO as Economic Integrator

The CPO’s job is not just building great products. It is connecting product bets to the metrics CFOs and CROs care about most. When your roadmap reduces CAC Payback and raises win rates and ACV, you create an integrated economic story. That story earns board confidence, investor enthusiasm, and organizational alignment.

Too many product leaders see finance and sales as adversaries. The reality is the opposite. They are leverage points. They give your product strategy credibility in the market and in the boardroom.

The most effective CPOs are translators. They turn product roadmaps into capital efficiency for the CFO and revenue acceleration for the CRO. That is how you build not just great products, but great companies.

The CPO’s Playbook for CFO and CRO Alignment

To make this concrete, here is a simple playbook for engaging your CFO and CRO. Use it as a translation guide between product strategy and economic impact.

With the CFO: Speak the Language of Efficiency

• Frame investments as capital allocation decisions. Show why this market, capability, or infrastructure deserves capital over other options.

• Tie features to financial outcomes. Connect roadmap items to CAC Payback, gross margin, cash flow, or valuation multiples.

• Show ROI at the company level. Highlight how each dollar invested multiplies into revenue and margin improvements.

• Be transparent about cost structure. Demonstrate that you understand how product choices ripple into margins, pricing power, or service costs.

• Position PLG as efficiency. Explain how self-serve funnels, onboarding automation, and viral loops bend CAC, not just how they grow users.

With the CRO: Position Product as a Revenue Multiplier

• Provide actionable usage data. Equip sales with signals that identify accounts ready to buy or expand.

• Address recurring objections. Use sales feedback to prioritize roadmap items that unlock stalled deals and raise win rates.

• Collaborate on packaging. Design tiers and bundles that expand ACV and create natural upsells.

• Build features that enable larger deals. Focus on integrations, compliance, or enterprise controls that open bigger markets.

• Measure impact together. Track win rates and ACV alongside product metrics to show how roadmap bets drive revenue performance.

When you apply this playbook, you shift the perception of product from a cost center to a growth engine. You are not just shipping features. You are shaping the financial destiny of the company.

Enterprises are in the middle of a gold rush. AI pilots are being launched in every function: customer support, HR, finance, product, sales. Board decks are littered with “AI strategy” slides.

And yet, here’s the uncomfortable truth (according to MIT): by the end of 2025, 95% of these AI experiments will be terminated.

Now, the statistic may not be accurate and open to debate. However, for such an outsized metric, we can see that there is a problem with AI.

This isn’t a model problem. It isn’t about weak algorithms or underpowered GPUs. It’s about how enterprises are approaching AI adoption. And why they’re failing to translate pilots into production.

I’ve spent my career scaling products at Microsoft, Slack, and scaleups like Spendesk. All of which had AI capabilities. I’ve seen firsthand what it takes to make software stick inside organizations. With AI, the challenge isn’t building a clever demo or checking off features, like “AI”. It’s integrating intelligence into the DNA of how companies operate.

So why are most enterprise AI initiatives headed for the graveyard? And what will the few that succeed do differently?

Subscribe now

The Learning Gap

Enterprises treat AI like a static piece of software, not a dynamic system. They roll out generic models like ChatGPT, or Anthropic, expecting them to transform workflows overnight. But without contextual grounding, feedback loops, or governance, these systems remain surface-level toys.

The pilot phase often shows some promise (a chatbot answers FAQs, a summarizer reduces email load) but the gains evaporate when the model can’t adapt to the enterprise’s real workflows. AI isn’t plug-and-play. It has to learn. And most organizations don’t yet know how to teach.

ROI Shortfall

CIOs are discovering that AI experiments rarely deliver measurable ROI. Not because the models are bad, but because they’re deployed as add-ons rather than embedded workflows. A chatbot layered onto an old system doesn’t magically improve customer satisfaction. A text summarizer on top of email doesn’t change productivity at scale.

These projects rack up cost, infrastructure overhead, and compliance headaches, but fail to deliver measurable business value. When budgets tighten, pilots without ROI are the first to die. And the macros are incredibly uncertain right now.

Data and Infrastructure Problems

AI thrives on clean, connected data. Enterprises are notorious for the opposite: siloed systems, inconsistent metadata, poor labeling, sensitive data tied up in compliance knots.

When data is fragmented, AI experiments produce inconsistent results. Accuracy falters. Trust erodes. Executives lose patience. Without AI-ready data infrastructure, even the best models can’t succeed.

Want to discuss this or get advice on how to build successful products? Schedule a call.

Schedule a Call

Workflow Misalignment

Here’s the biggest mistake I see: AI projects are launched as side experiments instead of being woven into core processes. An HR chatbot doesn’t matter if employees still file requests in a legacy portal. A finance summarizer doesn’t move the needle if approvals still happen in spreadsheets.

For AI to succeed, it has to be embedded into the flow of work. That requires cross-functional alignment between IT, operations, and business units. Most enterprises don’t have that muscle yet.

Culture and Governance

Technology adoption isn’t just technical. It’s cultural. Enterprises often force AI initiatives top-down, telling employees to “use this tool” without buy-in, training, or clear guidelines. The result? Resistance, fear, and (ironically) shadow AI adoption as workers find their own solutions.

At the same time, lack of governance raises real risks: hallucinated outputs, compliance violations, or security leaks. Without clear guardrails, many AI pilots get shut down before they create real harm.

Culture eats strategy. Governance protects trust. Without both, AI doesn’t stick.

Cost Overruns and Security

Finally, there’s the brutal math. Running AI at scale isn’t cheap. Infrastructure costs balloon. Security reviews slow everything down. Integration with legacy systems requires consultants and contractors.

The result? AI projects look promising in pilot, but the economics collapse when leaders try to scale them across thousands of users. That’s why so many AI initiatives quietly vanish. Not because the idea was bad, but because the cost and risk outweighed the benefit.

What the Successful 5% Will Do Differently

There’s good news. Some companies will succeed, and they’ll define what AI-native enterprises look like. They’ll start small, but integrate deeply. They’ll invest in data infrastructure. They’ll build feedback loops so the AI learns continuously. They’ll prioritize governance and trust. And they’ll measure the right metrics: not “chatbot sessions,” but time-to-value, intent accuracy, and trust compounding.

These companies won’t treat AI as a feature. They’ll treat it as the environment. And that’s the difference between a terminated pilot and a transformative shift.

Enterprise history is full of tech pilots that fizzled: early mobile apps, VR headsets, blockchain proofs of concept. AI risks becoming the next casualty, unless we learn from those failures.

The winners of this era will be the ones who understand that AI isn’t about experiments. It’s about redesigning workflows, data infrastructure, and culture for a new environment.

95% will fail. But the 5% who succeed won’t just adopt AI. They’ll become AI-native. They’ll become AI-native. Embedded into the DNA of the company.

👉 Curious how the AI-native playbook is reshaping SaaS and growth? I unpack it here.

The Old Playbook That Built SaaS

When I think about my career building products at Microsoft, Slack, and Spendesk, etc. I realize how much of the products’ success was tied to the Product-Led Growth (PLG) playbook of the 2010s and early 2020s.

At Microsoft, we scaled Outlook mobile from 1M to 60M MAU through relentless friction reduction, usage loops, and enterprise readiness. At Slack, my teams pushed beyond messaging into platform growth, architecting integrations that made the product indispensable, and leveraging email to drive the funnel. We built the fastest-growing PLG machine in the fastest-growing PLG company.

These were category-defining wins, and they were made possible by a PLG system built on assumptions that users:

• Were willing to invest time learning your product.

• Wanted to explore value progressively.

• Would tolerate onboarding journeys to get going quicker.

• Measured you against competitors in the same category.

Those assumptions no longer hold.

AI has changed the game. Today’s users don’t want progressive disclosure. They don’t want to learn your UI. They don’t want onboarding tours (they never really did!). They don’t even want to sign up if another tool (or ChatGPT itself) can solve the problem in seconds.

The PLG era we built SaaS in is over. But PLG isn’t dead. It’s being rewritten.

In this piece, I’ll outline four shifts every founder and product leader must grapple with. And in doing so, I’ll connect the old playbook, the one that worked for me and countless other leaders, with the new realities of the AI-native era.

Subscribe now

Time-to-Value Expectations Have Collapsed

In the old world of PLG, the magic came from progressive disclosure. You revealed value step by step, pulling users deeper into your product over days. A free trial, a guided tour, maybe some nudges and lifecycle emails. These worked because users were willing to spend the time. It was what users were used to and it was better than what had come before. Products like Photoshop or Microsoft Word - powerful tools, but with every feature imaginable vomited onto the screen for the user to pick through.

At Outlook mobile we leaned into the concept of progressive disclosure. Once we realized the mobile experience was about consumption and not creation, we found ways to remove features from the top level of the information architecture. This was a fight, because customers and folks in Redmond expected Outlook mobile to look and behave like Outlook on Windows. This was why Microsoft had lost in the past. And going against this with Outlook mobile was a key reason why we won. Against Gmail and Apple mail no less.

But that was 2016. Today, it would be too slow. More importantly, it was a fixed approach. We were making decisions that would improve the experience and drive engagement and growth. But it wasn’t adaptive.

There hasn’t been a massive amount of research in this domain, but we’re already seeing user expectations moving in this direction. And as with many growth philosophies and approaches, it’s coming from the domain of eCommerce and sales. According to McKinsey, 71% of customers expect companies to understand their needs and personalize their experience. According to Comviva, customer acquisition and retention costs can be reduced by 28% as a result of personalization.

Why does this expectation translate? Because AI has trained them. When ChatGPT, Perplexity, or Claude can generate an answer in three seconds tailored to your query, every second of lag in a SaaS product feels like friction. Expectations have been reset, and the bar is high.

In the AI era, time-to-value isn’t a funnel metric. It’s a second-by-second expectation.

What This Means for PLG Today

Freemium alone is no longer enough. Users won’t wait days to find value. The “aha” moment must happen in seconds, not sessions. Immediately upon sign-up, the user expects to be receiving value. The expectation is that within those few seconds the user can accomplish what they had in mind the moment when they clicked your sign-up button.

AI-powered personalization is mandatory. The single prompt input box of your favorite GenAI tool is just the starting point. While there is customization in the text response, there is no personalization in the visual experience. Just look at how ChatGPTs UI quickly becomes a disaster once you have more than five projects running in parallel. The point is, your product must infer the context of the user and then adapt to it. What is their role? What is their real-world workflow? What is their goal and objective? Your product must do that before the user clicks anything. And then adapt to meet those inferred expectations.

Metrics shift. There needs to be a change in what we measure, how we measure it, and what does success look like. Back at Slack in 2021, success was defined by the number of messages sent within five days by how many people within the channel - days. Now we need to measure the accomplishment of perceived value in how many seconds. Instead of “days to first value,” we measure “seconds to perceived value.”

The companies that win won’t just remove friction. They’ll erase it.

The Learning Curve is Dead

Classic PLG assumed users were willing to invest time learning your product. Tooltips, help docs, and sandbox accounts were all part of the journey. The early masters of this was Evernote. Remember them? Once upon a time, they were the productivity innovation darlings of Silicon Valley. They developed a novel approach to note taking based upon a deep understanding of how the user thinks about taking notes and how they actually use them after the fact. This novelty introduced complexity to the user and therefore friction to the creation of habits which impacted retention and growth.

Rather than hide all of the complexity (as Notion originally did), they provided in-app tool tips and educational moments delivered just at the right time. It was a blessing and a curse. Everyone fell back on this affordance to build engagement and drive retention at the expense of thoughtful design. But it worked.

At Microsoft, we obsessively reduced friction in Outlook mobile. We simplified setup, built intuitive gestures, and optimized the core mail–calendar integration. That focus was key to differentiating and scaling from 1M to 60M MAU.

We always had “educationals” on our backlog, but we kept on deprioritizing it. Emphasizing the “less is more” philosophy. A good example was the removal of the files tab. The outrage! At Slack we took a hybrid approach as we were better resourced both in design and engineering. Prompts were introduced, and you see them today. Nudges to move people along. We were also constantly paring back on design. Keeping the visual noise to an absolute minimum, following what I call the “successive revelation” approach to the user experience.

But here’s the truth: even the level of friction we considered a triumph then would kill adoption today. AI has flipped the relationship. Products can no longer expect users to learn them. Products must learn the user.

The numbers back this up. According to MarketsandMarkets, the conversational AI market is projected to grow at a 19.6% CAGR, from $17.05B in 2025 to $49.8B in 2031. Why? Because users don’t want menus, inputs, or training. They want to describe what they need and have the system respond.

But this only tells part of the story. It’s directional, but shouldn’t be taken literally. In that timeframe, software will move from reactive to responsive to proactive. Even now, ChatGPT and Claude are learning what you want in reaction to what you input. You’re not clicking a single mouse button on a menu, you’re tapping on hundreds of keys of the keyboard. You’re telling the AI what you want and slowly the system starts to learn who you are and how you like to be communicated with. But it’s still reactive. It’s not responsive to your needs. And it certainly isn’t proactive.

From Interfaces to Personalized Interactions

In the past, and with many legacy products used every day all over the world generating billions of dollars in revenue, the message has been “Here’s our interface. Invest time and you’ll unlock value. Let me teach you how”.

Now, we have the rush to conversational AI. Just this week, the Wall Street Journal stated that “nearly every software-as-a-service company is now selling agent tools to their customers”. But these are bolt-on solutions created after the fact. They are reactive to the user’s needs, as they are expressed explicitly. But as the MIT finding recently cited by Fortune that 95% of enterprise AI solutions will fail by the end of the year indicates, it is because of the continued learning gap. Tools are too hard to learn for individuals and organizations. Crucially, generic tools like ChatGPT “stall in the enterprise because they don’t learn from or adapt to workflows”. Essentially, the model is “Tell me what you want, and I’ll give it to you instantly. But you have to tell me. And all you have is this chatbox.” And that’s not going to cut it. The learning curve isn’t just steep, it’s unacceptable.

Going forward, we need to move to a model of “I know you….here, let me help.” We need to combine with design-led principles that have served us well in the past the power of inference to proactively adapt the user experience to enable each user accomplish their goal in a personalized way.

For PLG leaders, this means adoption loops look entirely different:

• Interfaces shrink. Prompt fields and conversational UIs replace complex menus.

• Education disappears. The best products reveal functionality through user intent, not tours. And each user gets a unique experience personalized to their needs.

• Retention shifts. Stickiness isn’t about mastering features, it’s about trusting the product to deliver without instruction. And each user journey can be different.

The products that scale in the AI-native world will be the ones that act like partners, not platforms. This is a tall order and will require deep exploration and experimentation. The first iteration of which we’re seeing with the advent of Agents. But we have a long way to go.

Onboarding as Discovery vs. Education

Onboarding used to be a core PLG lever. It was where you taught users how to get started, introduced features, and nudged them toward activation.

At Slack, we invested in explaining integrations, showing teams how to create them in channels and use email to bring colleagues into the channel. That worked and Slack grew explosively, and in many enterprise accounts, we out-competed Microsoft Teams.

But in 2025, those same tours would be attention killers.

At Outlook mobile, we designed beautiful guided tours. We soon ripped them out as they were actually driving down activation. It worked and we started to see the climb up towards our first 60M MAU milestone. We know they don’t work.

Why? Because onboarding as education assumes patience. And patience is gone. Userpilot still advocates in-app educationals, but warns that they need to be broken down to bite-sized chunks. But that’s missing the point and not looking forward to what users really expect.

AI Flips the Script - Product discovers the user’s intent

We need to harness the power of AI to deliver the experience and value AI has taught users to expect. We need to catch up with our users. This means that we need focus more than ever on the user’s intent. And use the product and onboarding moment to discover it.

This is the single biggest shift for PLG leaders. Onboarding is no longer about showing what your product can do. It’s about figuring out what the user wants to do, instantly.

This shift is made possible by the range of signals every product can now observe. Explicit inputs like search queries or button clicks, implicit behaviors like navigation paths or dwell time, and contextual information like user role, account stage, or connected data all become raw material. By combining these signals with lightweight rules, semantic matching, and AI models, products can predict what a user wants with increasing accuracy—whether it’s creating a card, importing a dataset, or setting up a recurring workflow.

The interaction model also changes. Instead of long-form tours or 20-minute training videos, intent discovery should feel conversational and lightweight: a one-line prompt, a smart default, or a pre-filled option that can be corrected with a single click. When confidence is high, the product can even “ghost” an outcome. Like pre-composing a chart or issuing a test card, and simply ask the user to confirm. Trust is built not through lengthy education, but through fast, accurate suggestions that prove the system understands you.

Done well, this turns onboarding from education into discovery. The product doesn’t ask users to learn; it shows it has already learned about them. Time-to-value shrinks to seconds, the learning curve disappears, and adoption loops accelerate because users feel like the system is anticipating needs rather than demanding attention. That’s the AI-native standard, and the new measure of whether a PLG motion is competitive.

Onboarding is no longer about teaching users what your product can do. It’s about proving your product already understands what they want to do.

Competition is Invisible and Instant

In the old PLG playbook, your competition was another SaaS tool and the battle lines were clear. When we were fighting to unseat Apple Mail, the customer pulled out a spreadsheet of all the features we “have to build”. And then at Slack in head-to-head duels at enterprise customers against Microsoft Teams, Microsoft used to love turning the conversation to a simple spreadsheet of check-boxes. Showing all of the features they’d copied (but poorly implemented).

As I used to say at Outlook mobile and at Slack, “be competitor aware, not competitor obsessed”. And particularly in the case of Teams, understand the whole picture, not just what’s in front of the user. (But that’s another article.)

If you’re worried about feature gaps now, you’re in even worse shape in a world where AI-native apps don’t care about feature matrices. Today, your biggest competitor is ChatGPT, Claude, or Perplexity solving the user’s problem in 30 seconds. Without signup, without a credit card, without onboarding.

Google is the first obvious victim. As Google execs show concern that users don’t want to click, read, click, click, click to find an answer. They just want the answer. But these general purpose AIs are also coming for you. The flexibility and power hidden behind the prompt box is real.

This is the existential threat for PLG leaders. The user’s question has shifted:

• Old: “Which SaaS product should I use to solve this problem?”

• New: “Why can’t I just describe what I want and have it work?”

Implications for PLG

Differentiation moves up the stack. Competing on features has always been a race to to the bottom. AI makes this instant. You must compete on data, workflow integration, and outcomes. Focus on the user intent, the outcome, and the value. And deliver it beautifully.

Moats look different. It’s no longer features vs. competitor X. It’s whether your product is the fastest, most trustworthy orchestrator of intent into outcome.

Distribution must change. SEO, ads, and freemium are still useful, but the real battle is being the invisible layer inside workflows (APIs, agents, embedded AI copilots).

The competitive battlefield is no longer a Gartner quadrant. It’s the user’s brain in the split second they decide whether to open your app, or just ask ChatGPT.

Conclusion: PLG is Dead. Long Live PLG.

The old PLG playbook isn’t irrelevant, it built SaaS as we know it. I’ve lived its power firsthand. At Microsoft, reducing friction was the lever that scaled Outlook mobile adoption against entrenched giants like Google and Apple. At Slack, carefully crafted integrations and thoughtful onboarding prompts unlocked massive enterprise adoption. Across both Microsoft and Slack, freemium economics delivered efficiency in growth that would have been impossible through sales-led motions alone. And in both cases, the combination of go-to-market strategy and deep product integration created durable category leadership.

But what worked then won’t work now.

In the AI-native world, the rules of growth have fundamentally shifted. Time-to-value must be instant, not measured in days or sessions. Learning curves must be inverted, with products adapting to the user instead of the other way around. Onboarding must become discovery, revealing intent rather than teaching features. And most importantly, competition is invisible and instant. The threat is no longer the SaaS rival across the street, but an AI tool that can solve the same problem in seconds without sign-up or onboarding.

This is both terrifying and exciting. For founders, product leaders, and investors, it means rethinking every assumption we’ve held about PLG.

That’s exactly what I’ll be sharing at the Product-Led Summit in San Francisco on September 16th, where I’ll present “PLG is Dead. Long Live PLG: How AI is Rewriting the Rules of Product-Led Growth.”

If you’re attending, I’d love to continue this conversation in person. And if you’re reading this later, know this: the companies that thrive in the next decade won’t just use AI. They’ll rebuild PLG itself around it.

In the AI era, PLG doesn’t guide users to value. Products must deliver it before they ask.

💡 Enjoyed this essay?

• 👉 Subscribe free to get new posts in your inbox each week.

• 🔒 Paid subscribers unlock full archive access, including past essays, deep-dive playbooks, and private frameworks I use with SaaS founders and investors.

  Subscribe now

Read more

If you’re thinking about raising money, or eyeing an exit, the latest PitchBook Enterprise SaaS report provides key insights.

The SaaS landscape has changed, and the rules of the game aren’t going back to 2021. Valuations are lower. Growth expectations are tighter. Profitability isn’t just a nice-to-have; it’s the price of admission. PitchBook’s Q2 2025 report shows the median public SaaS multiple hovering around 3.8x revenue, less than half of where it was just two years ago.

But that doesn’t mean opportunity is gone. Quite the opposite. The report highlights a breakout quarter for tech IPOs and a renewed focus on operational efficiency. For late-stage founders, this is the blueprint for building a company investors will fight to back. For earlier-stage CEOs, it’s a reminder that every funding round now comes with sharper questions around unit economics, runway, and path to profitability.

So, what does this all mean for you as a founder? Here’s a breakdown of the key PitchBook insights and what they signal for private markets. And the steps you should be taking now to set your company up for success.

Key Takeaways from the Report

1. Valuation Trends

   • Median EV/TTM revenue multiples for public enterprise SaaS companies settled at 3.8x in Q2 2025, up slightly from 3.7x in Q1, but still far below 2022 levels (8.2x).

   • Forward revenue multiples remain weak, reflecting investor caution and broader macroeconomic uncertainty.

2. Revenue & Growth Outlook

   • Revenue growth is decelerating to high single-digit to low double-digit rates in 2025, compared to historical growth of 15–30%.

   • The HR/workforce management and marketing/customer engagement sectors will see the sharpest slowdowns, while analytics & BI shows modest growth resilience.

3. Profitability Emphasis

   • Gross margins are expected to rise to 76% in 2025, with the strongest improvement in ERP and CRM segments.

   • EBITDA margins increased to 21% in 2024, up from 17% in 2023, and are forecast to strengthen further in 2025 as companies prioritize operating efficiency.

4. IPO Market Recovery

   • Q2 marked a resurgence season for tech IPOs, with listings from CoreWeave, MNTN, eToro, Circle, and Chime. Pending IPOs (e.g., Klarna, StubHub, Cerebras Systems) suggest investor interest is returning, though selectively.

• Winners and Losers in Valuations

  • Block (+196%), Twilio (+75%), and Domo (+65%) saw major EV/TTM revenue multiple gains.

  • Sprout Social (-58%), Braze (-50%), and MongoDB (-42%) were among the biggest decliners.

• Sector Divergence

  • Marketing/advertising SaaS will see margin pressure (-1% gross margin in 2025), while ERP, CRM, and analytics segments are projected to grow margins.

Implications for Startups

• Lower Valuation Benchmarks

  Public SaaS multiples anchor private valuations. With median multiples under 4x, late-stage startups will face lower funding rounds or structured terms (e.g., down rounds, higher liquidation preferences).

• Shift Toward Profitability & Efficiency

  Investors now prioritize gross margin strength, operating leverage, and positive EBITDA over aggressive top-line growth. Startups will need to demonstrate a clear path to profitability at scale.

• IPO Market Reopening but Selective

  The IPO recovery suggests public exits are possible but only for companies with compelling growth and profitability narratives. Early-stage founders should consider how their metrics and unit economics align with public comparables.

• Segment-Specific Headwinds

  Startups in HR, workforce management, and martech may face slower revenue multiples and funding appetite due to segment deceleration. Analytics and ERP-focused startups may benefit from relative resilience.

Founders & CEOs Action Plan

We are in a new and established SaaS market reality: valuations are lower, growth is slower, and profitability is the new currency. (Of course, there are exceptions!)

So, what should you do?

Founders and CEOs should realign their strategies to emphasize unit economics, cash efficiency (which is challenging in AI), and sustainable growth models.

1. Reset Growth Expectations

   Plan for a lower-growth environment. Aggressive projections may no longer resonate with VCs or the public market without a strong distribution strategy and execution. (See Building Your PLG Company - the efficient growth playbook.)

2. Prioritize Margins and Cash Flow

   Public comps highlight that 76% gross margins and 20%+ EBITDA are the new bar for quality SaaS businesses. Founders should pivot product and pricing strategies to optimize contribution margins. (See The AI Pricing Supernova - the pricing playbook.)

3. Rethink Valuation Narratives

   If raising capital, align valuation expectations with the new reality (3–6x ARR multiples, not 8–10x). Strategic acquirers may be a better option for exits than IPOs in the near term.

4. Product-Led Efficiency

   The days of “grow at any cost” have been over for a while. CEOs need to think deeply about PLG strategies, retention levers, and cost-effective customer acquisition.

5. Prepare for a Longer Runway

   With revenue multiples compressed, capital efficiency and burn reduction are paramount. Extending runway may be more valuable than chasing expensive growth.

If PitchBook’s report reinforces anything, it’s that clarity and execution win in this market. I work with SaaS founders and CEOs to translate market realities into actionable growth strategies - sharpening unit economics, building valuation-ready narratives, and driving profitable expansion. Whether you’re planning your next fundraise or positioning for an exit, I can help you cut through the noise and move faster toward the outcomes that matter.

Let’s build your next phase of growth - book a call and let’s talk!

The Great SaaS Rebuild: AI’s Organizational Revolution

AI isn’t just transforming your product. It’s transforming your entire company, and the winners will look nothing like the SaaS companies of the last decade.

One of the many things I love about our industry is the fact that we are evolving and innovating the organization as quickly as we are developing and innovating the technology we build.

This article, Part 4 of the series, builds on the ideas explored in Part 1: The AI R&D Team Compression, Part 2: AI is Breaking SaaS Metrics, and Part 3: The AI Pricing Supernova. Together, these earlier pieces laid the groundwork for understanding how AI is reshaping software development, key performance metrics, and pricing models. Here, we turn our focus to the organizational structures of SaaS companies and how they are being rebuilt for the AI era.

Adaptability and agility in organizational structure are becoming cultural prerequisites for success. Companies that cling to rigid organizational charts, outdated roles, and entrenched responsibilities (often driven by empire building and ego) have always been at a disadvantage. In the AI era, this cultural flaw will be fatal. They will be outmaneuvered by competitors who adapt and reorganize as quickly as the market demands.

I’ve seen both models up close. Slack, for example, was insanely adaptive, with minor reorganizations happening every quarter and major ones every six months, ensuring the company was always aligned with its most pressing challenges and opportunities. On the other hand, I’ve also experienced organizations where massive protectionism reigned. A belief that history, hierarchy, and headcount were paramount. Those companies inevitably lost momentum and strategic flexibility.

The AI era is amplifying this reality. Teams are shrinking, not because companies are scaling back their ambitions, but because AI-native tooling is replacing repetitive execution with intelligent orchestration.

The traditional SaaS model, which relied on large engineering squads, sprawling sales teams, and siloed data functions, is giving way to smaller, high-leverage teams. These teams focus on orchestrating AI agents, design systems, and real-time insights rather than managing manual workflows. As a result, organizations are achieving three to five times the productivity with half or a third, of the headcount they once required.

Engineering squads that once needed ten-twenty people to deliver a product increment can now thrive with seven to ten engineers, supported by tools like GitHub Copilot, Replit Agents, and automated QA pipelines.

This is more than leaner operations. It represents a fundamental change in how software is imagined, built, and shipped, with human creativity and strategy layered on top of AI-driven execution.

Across every function, AI is not simply a productivity boost; it is liberating teams to focus on the work they always imagined they should be doing. The high-leverage, strategic, creative, and analytical work that was previously buried under mundane, repetitive tasks.

Teams this article covers

I’ll cover the impacts on all of these organizations and some of the tools they use (what’s your favorite tool?). And I’ll provide an approach you can take to lead, organize, or work with that team in the AI era.

🤖 Engineering

🤖 Product Management

🤖 Design

🤖 Product Marketing

🤖 Data & Analytics

🤖 Customer Success

🤖 Sales

Submit Your Favorite Tool

Read more

Most companies claim that design matters, but unfortunately, few act like it. We can see it in the products they ship and those we use every day. As a product person, it is hard to engage with any subpar user experience and not feel frustration, annoyance, or even anger. (It drives my wife crazy.)

What I learned at Slack is that design isn’t the cherry on top or the final touch of polish. It’s the organizational principle and architecture at the heart of building products that define a category.

Design is not just a visual layer, but a way of thinking that forces clarity, empathy, and rigor across every function. When it’s missing, even the best strategy becomes vulnerable, and growth flattens. I saw this at Envoy, where five years of COVID-driven feature-factory production had stalled innovation and flattened growth.

This post isn’t about cleaner lines or sleeker buttons. It’s about treating design like a central nervous system, not a coat of paint. You’ll have to rewire habits, politics, and sometimes your own instincts. But the payoff isn’t aesthetic. It’s a compounding strategic advantage and sustainable growth.

💸 Why Do This?

What I'll walk you through below is a lot of work. And it’s hard work. So why invest this energy when you’ve got so much to do? So much to build. Simply put, it’s the return on investment. But it’s hidden or hard to see ROI that is the issue.

The greatest challenge to investing in design and user experience is the often indirect connection between it and revenue.

Add a feature a customer is asking for, and the account executive will tell you they can close a contract. This is alluring. It is also the genesis of feature-factory companies that peak and then plateau.

Design-led companies aren’t just building better products; they’re building bigger businesses.

Figma grew from zero to over $700 million in ARR within seven years, capturing more than 75 percent of the UI design market along the way.

Notion grew to about $300 million in revenue by 2024, and crossed the 100 million user mark in record time.

Even incumbents have proven the case. Airbnb, founded by a designer, has prioritized experience from the start. In 2024, it recently posted $11 billion in revenue.

These companies don’t treat design as polish. They treat it as infrastructure.

How do you do this? Read on.

✅ Build Design-First Teams

Post-Outlook and Slack, at every company I played a role at, this was the hardest shift to make. Orient your product development process to involve design from the start.

PMs need to think in terms of flows and not feature lists. Step back from the screen and understand the context of the user. What is their journey and what is the journey of their company? Where does your product fit in that journey?

Engineers need to understand that UX debt is real debt. It could be the most important debt for the longevity of the business. This is often the hardest concept to convince engineers to adopt. For them to make a change to a product that ultimately will “do the same thing” is often counterintuitive and not seen as “productive”. This is especially the case if you’re not getting support tickets or bug reports.

🛠️ Invest in Design Infrastructure

Invest time and resources into the tooling of the design team. This will pay dividends going forward. If you do everything else right, design will become your bottleneck to velocity. Design infrastructure introduces design scale.

Provide your design team with the best and latest tools. The tools designers use need to inspire and motivate. They want to learn from the best, and they can do that by using the best tools.

This means deploying Slack, not Teams. Notion, not Confluence. Asana, not Jira. Yes, you’ll need to figure out how to build a process that does not pool or isolate data across your process. But this isn’t hard. It’s just work.

Create and maintain a design system. This sounds basic, but so few companies do this. It is treated as a nice-to-have or something that is built over time on the side. This is a strategic mistake that will cost you in velocity moving forward.

Prioritize consistency and accessibility. I continue to hear from companies complaining that they have to invest in accessibility. Not only is that questionable from an ethical perspective, it’s short-sighted (excuse the pun) strategically.

In the Outlook Mobile team, we were initially forced to learn about accessibility. As an engineering-heavy team, it was hard for folks to invest in changes that would result in a product that would essentially “do the same thing” (see above).

However, through a deeply experiential approach to get the whole team to understand what using a product is like to use when you’re short-sighted or colorblind increased our collective empathy. This resulted in us building a much better product for those without accessibility challenges.

Last, but not least, you must budget for design research and iteration. It takes time for us to step out of our own world and get to understand that of people who are completely different to ourselves.

Different cultures, different languages, different challenges and constraints. It takes investment to gather and embody this learning. But it will pay dividends going forward.

Subscribe now

📊 Measure What Matters

Move past NPS. CEOs pay way too much attention to this metric. Yes, it’s a useful tool to get a baseline. But it is incredibly laggy, rarely actionable, and takes a long time to move. There is a better way.

Instrument your product and within the user journey(s), calculate the time to first value. How long does it take for a new user to get value out of your product? What is the action that signifies that achievement?

Map key workflows and measure the percentage of them that are completed. For those that don’t complete, ask yourself and users - “Why did the user start that workflow?”, and then “why did they abandon?”. There is no point making it easier for a user to accomplish a goal they didn’t actually have.

Measure the rate of issues coming into customer support. Their severity and the time it takes to address the issues.

Before we had any instrumentation within Outlook Mobile, we measured the number of 1-star reviews we were getting on the app stores (a lot!!). We categorized them (manually!!) and then addressed the top ranking ones in every week-long sprint. This is what got us from a 2.2 star app to a 4.7 star app and winning the Verge’s best email product on the market in 2017.

🧭 Elevate Design Org-Wide

Design isn’t a service function waiting for instructions to be cascaded down via PMs. It needs a seat at the strategy table. Not only do we get incredible user-centric insights into discussions, but we get visualizations of the company and product vision that motivate and inspire. More importantly, we elevate design as a core company value and cultural touchstone. Everyone becomes a product person.

In practical terms, design should have a strong influence on roadmaps. And not just polish them and help with decks and mockups.

Inform and educate executives across the organization how design drives conversion, retention, and expansion. These conversations bring the subjective down to the objective and practical.

Lastly, work with your marketing team to help sales enablement prepare the sales organization to sell the experience. Not just the features. Move away from check-box comparison tables to deeper value and strategic problem solving.

As a product leader, you will look at all of the above advise and see a tremendous amount of work to be done. None of these are events. It is a process. Get buy-in from your stakeholders and treat the transformation as you would a product roadmap or strategy.

Your customers, users, and your business will thank you for it!

Why call it a “supernova”?

Because SaaS pricing isn’t evolving - it’s exploding. Like a supernova, the old model of flat-rate and seat-based pricing has burned through its fuel, collapsing under the weight of AI-driven costs and unpredictable usage patterns.

What follows is both destructive and creative: the end of predictable ARR, but also the birth of hybrid, usage-based, and outcome-oriented pricing models that better reflect how value (and cost) is delivered in the AI era. This shift is not incremental; it’s reshaping revenue models, investor metrics, and even product architecture at a fundamental level.

Summary: What You’ll Learn in This Article

SaaS pricing is undergoing a once-in-a-generation reset. In this article, you’ll discover:

• Why flat-rate and seat-based pricing are collapsing under AI-driven costs and unpredictable usage patterns.

• The Pricing Supernova Framework: six pricing models for the AI Era, from base subscriptions to agent-based pricing, that leading SaaS companies are adopting.

• Case studies from Snowflake, OpenAI, and Slack, showing how hybrid and usage-based models are reshaping revenue and margins.

• How investors now evaluate pricing strategies, focusing on contribution margin, usage momentum, and cohort-based growth efficiency.

• Why product leaders (not CMOs or CROs) must own pricing in the AI era, and the actionable steps PMs and CPOs can take today.

• How pricing and product design are converging, with usage metering, embedded monetization hooks, and transparent billing becoming product features.

If you want to understand why SaaS pricing is exploding like a supernova (and how to design for this new reality) this article is for you.

Key Takeaways

Some of what’s in this article.

1. Traditional SaaS pricing is breaking. Seat-based and flat-rate models are unsustainable for AI-native products due to rising variable costs from LLM queries, GPU inference, and infrastructure demands.

   What will replace it.

2. Usage-based and hybrid pricing are the new standard. Flexible pricing structures, ranging from metered tiers to workflow and outcome-based pricing, align revenue with value delivery and contribution margin.

   Examples and case studies.

3. Pricing now shapes product design and investor perception. Products must integrate usage metering, transparent billing, and monetization hooks, while investors prioritize contribution margins and efficiency metrics over predictable ARR growth.

   Founder and Product Leader Action plan.

The Pricing Supernova Framework: 6 Models for the AI Era

1. Base Subscription (Flat Fee)

Flat-rate pricing still serves as a familiar entry point for many early-stage SaaS companies. It provides simple, predictable revenue streams and is easy for buyers to understand. For AI products, flat fees often include a fixed amount of AI usage, say 1,000 API calls per month, to protect contribution margins.

Challenge: Without usage caps or careful monitoring, high-value customers can become margin-draining power users. Early-stage companies often overestimate the profitability of flat pricing because they fail to account for variable AI costs.

2. Usage-Based Tiering

Usage-based pricing is rapidly becoming the default for AI-native products. By metering high-cost features, such as AI queries or compute minutes, companies ensure that heavy usage correlates with revenue. For example, a baseline plan might include 1,000 API calls, with $0.05 charged for each additional call.

Microsoft’s Copilot has a complex billing rate manifest that converts the type of answer provided by the agent into an equivalent measure of value in terms of the number of messages. The price of a message ($0.01/message) is actually buried in a blog post rather than on their top-level pricing table.

• Generative Answer = 2 messages

• Agent Action = 5 messages

Impact: This model not only protects margins but also encourages thoughtful usage, aligning customer behavior with underlying infrastructure costs. However, that thought about usage creates friction in adoption. The old adage, “Don’t make me think!”, is still a founding principle in product design.

3. Credit Bundles & Prepaid Packs

Credit bundles offer budget predictability for customers while improving cash flow for vendors. A company might sell 10,000 AI requests for $400/month, smoothing out consumption spikes and creating upfront revenue recognition. Microsoft also has a prepaid pack for messages - $200 for 25,000 messages per month.

Challenge: The complexity of tracking credits and expiration dates can lead to poor customer experience if not well-integrated with in-product analytics and dashboards. Building this process into your sales motion, with close collaboration with Customer Success, will be key.

4. Workflow-Based Pricing

Instead of charging for raw inputs (like tokens or compute time), companies price around completed workflows. Packaged tasks that deliver a clear outcome. This shifts the narrative from technical metrics to business impact. It brings the Jobs-to-be-Done framework of building products directly in line with the tangible needs of the customer and the conversation your account executive will have when pitching the value proposition.

Example: A customer relationship platform, such as ServiceDirect, might charge for “qualified leads generated” rather than the number of AI calls processed.

5. Outcome-Based Pricing

Outcome-based pricing represents the ultimate alignment of vendor and the customer’s success. Companies charge when a specific business outcome is achieved, such as increased revenue, reduced churn, or verified cost savings. For example, Zendesk AI and Intercom Fin charge based upon the number of support tickets resolved autonomously. This moves the discussion toward measurable value, something that the best account executives love to talk about. Value-based selling rather than features.

Challenge: This model requires transparent metrics, strong customer trust, and clear contractual definitions. But when executed well, it cements the vendor as a true partner rather than a utility.

6. Agent-Based Pricing (The Frontier)

Looking forward, AI-native companies will monetize AI agents like human labor. Pricing will be based on the value of tasks performed or roles replaced. This “digital labor” pricing model is still nascent but signals a shift to value alignment that transcends seat or workflow-based pricing. There are varying models being tested, and even though Salesforce is marketing Agentforce as a labor replacement, their pricing model is still based on credits or conversations at $2/agent/conversation.

How Snowflake, OpenAI, and Slack Are Winning With Hybrid Models

Snowflake: Consumption-Based Predictability?

Snowflake’s rise to $4B+ ARR is anchored in its consumption-based pricing model. Customers pay for actual compute and storage usage, but Snowflake pairs this with transparent usage dashboards and tiered discounts that make costs predictable. By combining pay-as-you-go with enterprise pre-commit options, Snowflake provides both flexibility and financial planning certainty.

Investor Appeal: Contribution margin helps us understand how much control the company has over its revenue generation engine and the cost model associated with that revenue generation.

Beware: It’s important to note, however. From their Q3 earnings note, we can see “gross profits for Q3 FY2025 was $621.2 million, representing a gross margin of 66%, down from 69% in Q3 FY2024. The decline in gross margin was attributed to higher costs associated with product revenue, which grew by 46% year-over-year, outpacing the growth in product revenue itself.”

This points to the challenge we’ve been discussion. AI drives COGS (Cost of Goods Sold) up while the company struggles to increase their ACV (Average Contract Value) to compensate for the increase in value AI delivers.

Sources: Snowflake Pricing Overview | Snowflake Q3 Earnings

OpenAI: Tiered API Pricing with Guardrails

OpenAI’s API pricing demonstrates hybrid monetization: a base subscription (e.g., ChatGPT Plus) paired with metered API usage (tokens) for developers and access to different reasoning models (OpenAI o3, o4-mini, and o4-mini-high). All with user experience expectation setting guardrails.

“To ensure a smooth experience for all users, Plus subscriptions may include usage limits such as message caps, especially during high demand. These limits may vary based on system conditions.”

Lesson for Startups: Pricing-aware product design, such as tiered access to different model efficiencies, keeps contribution margins positive even as usage scales. Set expectations with users in terms of experience and provide yourself space to maneuver.

Source: OpenAI Pricing | What is ChatGPT Plus

Slack: Freemium + Usage-Driven Expansion in the Age of AI

Slack’s early success hinged on a freemium product-led growth (PLG) motion (Something my teams loved to drive when I was there). The free tier was deliberately designed to showcase core collaboration value, unlimited messages for the first 90 days, basic integrations, and team messaging. However, feature gating, like restricting searchable message history and limiting advanced integrations, naturally pushed teams to upgrade once they experienced Slack’s full potential. This strategy enabled us to build viral adoption loops while keeping Customer Acquisition Costs (CAC) remarkably low.

As AI reshapes workplace productivity, Slack is evolving its pricing model to reflect new value layers. With Slack AI (launched in 2024), the company now offers AI-driven features such as intelligent summarization of conversations, semantic search across channels, and automated action items. These features sit on top of Slack’s collaboration graph and generate incremental compute costs - making usage-based pricing for AI components a necessity.

How Slack prices AI:

• AI features are not bundled into the base free or paid tiers. Instead, Slack charges an add-on fee per user per month for AI functionality, currently priced at $18/user/month for Slack AI (as of 2025).

• Large enterprise contracts include usage-based provisions for certain AI-powered workflows, ensuring costs scale in alignment with heavy adoption of AI-powered search and summarization.

• Slack’s AI strategy mirrors a hybrid pricing approach, with a predictable subscription fee for core messaging and value-based pricing for high-cost AI features.

Impact on metrics:

• The AI add-on provides a new expansion revenue stream, boosting Net Dollar Retention (NDR) among enterprise accounts.

• By separating AI usage from its standard plans, Slack protects its contribution margin while encouraging teams to experiment with AI tools without eroding profitability.

• Slack’s enterprise pricing tiers (Business+ and Enterprise Grid) now highlight AI-powered features as differentiators, effectively upselling customers to higher-value tiers.

Result: Slack has maintained its PLG DNA while adapting to AI’s economic reality, turning AI-powered collaboration into a monetizable premium layer on top of its already sticky product experience.

Sources: Slack Pricing | Slack AI Overview

Lovable: From Freemium Friction to Hybrid Pricing for Scale

Lovable, a no-code AI platform launched in 2024, exploded to $10M ARR in just 2 months with a freemium model built for viral adoption. It now boasts to be the fastest-growing SaaS company in history, today announced reaching $100M ARR.

However, heavy AI usage from superusers began to erode margins due to high compute costs. In response, Lovable pivoted to a hybrid pricing structure with monthly subscriptions ($25 for Pro, $50 for Business) and credit allowances (e.g., 100 credits/month). Power users pay overage fees, while teams and enterprises get flexible billing with add-ons.

Impact: The hybrid model preserved predictable revenue while ensuring high-usage customers paid proportionally for infrastructure consumption. Lovable’s approach aligns perfectly with margin-aware product-led growth, a model increasingly critical for AI SaaS.

Source: Lovable Pricing | Growth Unhinged | Case Study: Lovable AI

Read more

Who should read this: If you’re a SaaS founder, product leader, operator, or investor, this article is for you. Whether you're running AI-native teams, building usage-based products, or evaluating growth-stage companies, you need new metrics that reflect how modern software businesses actually work. This post breaks down what’s no longer working, and suggests what can emerge in its place.

This is the second installment in the series, “AI - The End of Software As We Know It” exploring how AI is dismantling and reshaping the SaaS model.

In Part 1, “The Compression,” we looked at how software teams are shrinking while output is rising. This article builds on that shift by examining why the SaaS metrics we’ve used for the last 15 years no longer reflect how modern software businesses operate.

I also suggest a set of new metrics that enable operators and investors better understand the health of their company and business in the AI era.

The Metrics Are Still on the Dashboard. But They Don’t Reflect the Road Ahead.

In board and investor meetings across the software industry, a key set of metrics (aka “Unit Economics”) is used for those outside of the business to get a succinct and objective understanding of how well the business is performing.

I touched on some key metrics to measure Product Market Fit in an earlier article. Here’s a quick primer on the five core SaaS metrics that shaped the last era…the ones that need to evolve to meet the need of the AI era.

• CAC Payback: Measures how long it takes to recoup customer acquisition costs using gross profit. Commonly used to assess marketing efficiency and sales ROI. From a product perspective, this is influenced by the core value proposition and therefore price.

• LTV/CAC: Compares the lifetime value of a customer (based on expected revenue and churn) to the cost of acquiring them. Used to evaluate the return on growth investment. From a product perspective, this is commonly influenced by driving PLG motions to increase usage.

• Net Dollar Retention (NDR): Reflects revenue expansion, contraction, and churn over time within a customer base. Serves as a signal for product-market fit and upsell success. From a product perspective, this is driven by retention rates through product delight; and upsells and cross-sells via PLG and PLS motions.

• Gross Margin: Shows how much revenue remains after direct costs (COGS) are subtracted. Historically used to indicate software profitability and business model efficiency. From a product perspective, this is your core value proposition - how big is the customer problem you’re solving, and how often do you solve it.

• The Rule of 40: Combines growth rate and profitability (typically EBITDA margin) to offer a composite signal of SaaS health. A Rule of 40 score over 40 was considered strong. From a product perspective, the better your product experience and monetization hooks, the faster your ARR grows. This directly lifts the growth side of the Rule of 40.

Founders, operators, and investors built entire operating models and fundraises on these KPIs. But in the AI era of usage-based pricing, AI-native products, and infra-heavy cost structures, these metrics are breaking down (check out - Why AI Is Breaking Your SaaS Pricing Model).

These unit economics are still useful. But because how a company makes money is changing, they’re no longer sufficient.

Why These Metrics Are Breaking

Traditional SaaS metrics were designed around a simple, scalable model: fixed pricing, predictable revenue, and minimal marginal cost. But the AI-native cost model breaks these assumptions.

In today’s stack, revenue is volatile, cost of goods sold includes real-time compute, and customer value ramps unpredictably. And so pricing is no longer fixed; it flexes by engagement tier, output, or agent cost.

As a result, old metrics like CAC Payback and Gross Margin become misleading. They flatten out variability, obscure expansion dynamics, and mask infrastructure costs that scale nonlinearly. That’s why these metrics don’t need to be discarded, but they do need to be reinterpreted and augmented.

What’s Breaking, Why, and What to Do

These metrics didn’t fail. They’re just being asked to explain a reality they weren’t designed for. When cost structures become elastic, pricing becomes nonlinear, and usage patterns define value, traditional SaaS KPIs start to distort instead of clarify.

This section breaks down exactly where each legacy metric is falling short, and what we can use to augment or replace it.

Read more

AI capability is accelerating. But adoption is fragile. According to reports, many AI tools and agents will be abandoned by the end of the year.

Not because companies don’t see the potential, but because most SaaS vendors and their customers are still figuring out how to embed that potential into real workflows.

The bottleneck isn’t the model. It’s the system. And transforming the system means translating between technology and how work actually gets done.

This approach was the one we took at Microsoft when I was leading enterprise adoption against stiff competition against the incumbents - Google and Apple. It’s even more important now with the AI transformation, as everything is going to change.

To make that leap, we need two roles to step up:

• SaaS vendors who understand and redesign customer workflows

• IT and Operations leaders inside customer orgs who deploy and adapt those workflows

  And where they meet? That’s where AI transformation happens.

  Beyond the Build is a reader-supported publication about AI and scaling SaaS companies.

The SaaS Vendor's Role: Build AI That Supports Real Work

The next wave of AI-native software won’t win on novelty, it’ll win on deep understanding. SaaS companies must go deeper than feature development and start owning workflow transformation.

This means investing in the discovery work that’s often overlooked or underfunded—understanding not just what users say they want, but how their jobs actually unfold in messy, human systems.

PMs → Redesign the System, Not Just the Feature

Product managers must evolve from feature shippers to workflow designers. That means:

• Mapping end-to-end user tasks, expanding upon simple Jobs-to-be-Done

• Identifying manual, repetitive, or error-prone steps

• Testing where agents, automation, or intelligence can slot in without breaking trust

PMs fluent in both product strategy and the language of AI orchestration will unlock the biggest gains.

Design Researchers → Find the Latent Friction

Great researchers surface not just pain points, but process. They uncover:

• Shadow workflows

• Workarounds hidden from dashboards

• Tacit knowledge that can be codified or supported by AI

Design research is the lens that shows product teams where and how AI can provide leverage.

Customer Success → Your Customer’s Workflow Guide

CS isn’t just about retention anymore, it’s about transformation. Success teams sit closest to customer reality and are uniquely positioned to:

• Identify which customers are AI-ready

• Educate stakeholders on new capabilities

• Partner with internal ops teams to rewire old habits

In the AI-native era, CS becomes the bridge between product intent and customer transformation.

Subscribe now

The Customer’s Role: Operationalize the Change, Don’t Just Buy the Tool

Inside customer organizations, the role of IT and Operations becomes more strategic than ever. No matter how good the AI product, value only gets unlocked when it’s adopted into actual work. This has always been the case, but it’s even more important now.

This is not about installing software. It’s about building confidence, reshaping processes, and learning to collaborate with software that thinks.

IT & Ops → From System Maintainers to Change Agents

These teams have a front-row seat to their company’s:

• Process inefficiencies

• Tooling gaps

• Integration points

• Security and compliance constraints

• Communication and education channels and processes

They know how information flows across the org, and where it breaks. That makes them ideal partners in helping teams pilot AI workflows, validate outcomes, and scale usage. They’re not gatekeepers. They are enablers.

But they can’t do it alone. They need partners.

Read more

This is the first installment in a new series - AI - The End of Software as we Know It. How software is being rebuilt from the inside out - by AI, by infrastructure, and by teams that no longer look like the ones we scaled just five years ago.

Subscribe now

The Illusion of the Tech Rebound

It would be comforting to think the 2022-2024 wave of tech layoffs was a temporary correction. That once macro conditions stabilize, the hiring engines will roar back to life.

But the data says otherwise.

This isn’t a pause.

It’s a reset.

Engineering teams are getting smaller, permanently. Not because demand is shrinking, but because the mechanics of software creation are changing.

Layoffs.fyi reports that over 645,000 tech jobs have been cut since 2022, with engineering bearing the brunt. Synthesizing this data with other public announcements of layoffs and productivity gains, I estimate that by 2030 engineering teams could be 35%-50% of their 2022 size.

The Productivity Shock

ICONIQ Analytics: The AI Builder's Playbook 2025 outlines how leading teams are shifting R&D investments. Internal AI productivity budgets are set to double in 2025 across all startup revenue tiers. According to respondents, we’re starting to see headcount budgets being used for those internal productivity investments. Across the ICONIQ portfolio, an average of 33% of a company’s total code is being written by AI.

As anticipated, engineering output has increased. According to Sundar Pitchai (Alphabet CEO), their Goose coding assistant has increased engineering productivity by 10%. GenAI tools like Goose are generating boilerplate code, auto-documenting endpoints, writing unit tests, and performing debug and code performance optimization tasks. And it’s expected that there will be a multiplier effect. These productivity gains will compound over time.

Increased Engineering Velocity

Investments in AI productivity tools are compressing the time-to-build. That’s exciting news for companies and customers. But in what way are software companies accelerating? Both GitHub and outside researchers have observed a positive impact in controlled experiments and field studies where Copilot was used:

• 55% faster task completion using predictive text

• Quality improvements across 8 dimensions (e.g. readability, error-free, maintainability)

• 50% faster time-to-merge

What does that mean to the overall engineering organization?

Even though 25% of code is reportedly AI-produced within Google, they maintain that AI is not taking away employees’ jobs. Instead, AI enhances their capabilities and enables them to focus on higher-level tasks.

But the public numbers don’t support this.

The Hiring Cliff

As noted above, Layoffs.fyi tracked over 645,000 tech layoffs from 2022. These cuts have been across Meta, Google, Amazon, Stripe, Microsoft, etc. All of which cut deeply into engineering. What began as overcorrection to the hiring bonanza during COVID and ZIRP is now strategy: leaner teams, higher leverage. And that’s showing up in the macro-numbers, according to Indeed and the Federal Reserve.

The Squad Restructured

Starting now, everything needs to change. How squads are structured, run, and operated going forward.

Read more