This comes up in almost every conversation we’re having with product leaders right now. Teams are investing in AI. They’re building features, shipping pilots, and showing demos that get leaders excited. Then someone asks the awkward but necessary question: “What are we actually getting from this?”
It’s not that they don’t care about ROI. They do. Many of them will tell you it’s the single biggest challenge they face as they scale AI initiatives. The problem is that AI has made ROI messier, not cleaner.
When prototyping is cheap and fast, teams can pursue more ideas than ever before. That sounds like a good thing, and it can be, if the prioritization discipline keeps pace. But in most organizations, it hasn’t. Teams now have more ideas than they know what to do with, and a much easier path to build them. Then you get AI work that ships beautifully, checks every delivery box, and still does very little for the business.
A recent Mind the Product article on ROI optimization in product development makes the underlying point well: product success has never been about output alone. The work has to connect to economic return. AI raises that bar because it makes output easier to produce, while making real value harder to isolate.
AI makes bad prioritization more expensive
Shiny objects are not new. Executive pet projects are not new. The loudest customer hijacking the roadmap is definitely not new. Product teams have been dealing with this stuff forever.
What’s changed is the cost of indulging them. When building a prototype took weeks, bad prioritization was expensive enough to create natural friction. Teams pushed back because the investment was visible. Now, when a prototype can materialize in hours, the friction disappears. A half-validated idea can become a shipped feature before anyone has asked whether the problem it solves is real, widespread, or valuable enough to justify ongoing investment.
McKinsey’s State of AI research captured how quickly generative AI moved into regular business use, boardroom agendas, and investment plans. But it also showed that adoption and impact were still uneven, with overall AI adoption holding steady and many organizations still early in managing risks and linking AI efforts to business value. The capability was moving fast. The operating discipline was still catching up.
AI creates more options, faster. That means product leaders have to raise the bar on what gets built. Customer impact and economic value need to do the filtering now, because “this will take too long to build” is no longer enough to slow bad ideas down.
Most teams still can’t connect product work to real economic outcomes
This is the part we hear the most. Teams have plenty of activity metrics: usage, engagement, time in product. The dashboards look good. But when leadership asks, “So what did this do for the business?” things get fuzzy fast.
The Mind the Product article includes examples that illustrate this clearly. In one case, a team optimized for user growth as their North Star, only to discover that roughly 80% of users tried the product once and never returned. The metric was moving in the right direction. The business was not. In another, teams celebrated strong engagement on new features without accounting for the full cost to acquire and support that usage, turning what looked like success into a long-term drag on profitability.
AI makes this dynamic worse because AI-powered features can lower the barrier to interaction. Users engage more easily, which inflates activity metrics, but that increased engagement doesn’t necessarily reflect real value being delivered. Without connecting those behaviors to retention, revenue, or cost impact, product leaders risk scaling features that look great in dashboards but fail to improve the business.
The trick is spotting the behaviors that predict the outcome before the outcome has already gone sideways. Once retention has cratered or margins have eroded, you are no longer managing ROI. You are cleaning up the mess. The work is identifying the customer and business outcomes that matter, then tracking the earlier signals that tell you whether you are on track.
ROI breaks down without system-level thinking
The Mind the Product article makes another point that’s particularly relevant for AI work: ROI is not a feature-level calculation. It’s a system problem involving dependencies, timing, and cost.
In traditional product development, teams have learned (sometimes painfully) that optimizing their piece of the work without accounting for upstream and downstream dependencies produces delays and missed value. AI introduces the same dynamic with additional complexity: data readiness, model performance, human-in-the-loop workflows, and operational integration all have to work together for an AI feature to deliver real business impact.
IBM’s Global AI Adoption Index points to the same system problem. Among enterprises exploring or deploying AI, the most common barriers included limited AI skills or expertise, too much data complexity, ethical concerns, difficulty integrating and scaling AI projects, lack of holistic AI strategy, and not having the use cases or end-user research defined. In other words, ROI does not break down only because the model is weak. It breaks down because the system around the model is not ready.
Microsoft’s 2026 Work Trend Index makes a similar argument from a workforce and operating model angle: employees are often ahead of the organizations around them. The report argues that leaders have to rearchitect work itself, including workflows, roles, decision rights, governance, and how people and AI work together. That is a much stronger bar than “add AI to the task and hope value shows up.”
For product leaders, AI ROI is not a feature-shipping problem. It is a system problem. The feature can work exactly as designed and still fail to matter if the workflow, data, handoffs, incentives, or measurement plan around it are broken.
What product leaders can do tomorrow
- Define the signal chain from leading indicators to lagging outcomes
For each major outcome you expect from an AI investment (retention, revenue, cost reduction), define two or three leading indicators: the behaviors and activities that happen early enough to tell you whether you’re on track before the lagging metric reveals it.
Duolingo offers a useful reference here. Their investor reporting connects product engagement, paid subscriber growth, and revenue growth. But for product teams, the useful lesson is what has to happen before the financial results show up. Are people coming back? Are they completing more of the experience? Are they engaging deeply enough for the value to translate into retention or paid conversion? That is where the signal lives, before you’ve already hit the rocks.
Without a clear line of sight between leading indicators and major outcomes, teams can’t determine whether AI features are driving results or simply correlating with them.
- Raise the bar for prioritization before the work gets funded
AI makes it much easier to turn an idea into a prototype, demo, or feature. That is useful, but it also means weak ideas can travel much farther before anyone asks the hard questions.
So before an AI initiative gets prioritized, require three things: a clear customer problem, a clear business outcome, and a measurable signal that tells you whether the work is likely to create value.
The conversation should not be, “Can we build this?” In most cases, the answer is yes. The better conversation is, “What problem are we solving? How does solving that problem impact revenue, cost, risk, customer value, or another major outcome? And what would we need to see early to know we are on track?” This is the kind of discipline we help product teams build into their operating model: clearer prioritization criteria, stronger evidence requirements, and leading indicators that keep ROI from becoming a post-launch guessing game.
And if the team cannot measure those signals because the instrumentation does not exist or was not included in the requirements, it is not ready to move forward. Back to the drawing board. Even if the metrics are imperfect stakes in the ground, there needs to be a plan for how the team will prove value before the work gets too far down the road.
AI creates more options, faster. Prioritization has to get more disciplined as a result, not less.
- Make instrumentation part of the build, not an afterthought
This one is important enough to call out as its own principle, even though it’s connected to the first two. Look at your top three AI initiatives that have entered or are about to enter the build phase. Do the teams know what needs to be measured early and often to ensure ROI? Are those measurement requirements included in V1? If not, stop.
Microsoft’s Work Trend research reinforces the importance of measuring how AI changes the work, not just whether people use the tool. For product teams, that means instrumentation cannot be a future sprint item. It needs to be part of the initial build, so teams can evaluate value and guide iteration in real time.
AI systems are non-deterministic, and value emerges (or doesn’t) quickly through real usage. Without early instrumentation, teams can’t detect success or failure in time to act. And by the time they add tracking in a later iteration, they’ve lost the early usage data that would have told them the most.
The takeaway
Product success has never been about activity. It is about whether the work creates value. AI makes that standard harder to fake. When building and shipping get cheaper, speed is no longer the flex. The flex is knowing early whether what you built is changing behavior in a way that improves customer outcomes and business results.
The product leaders who win with AI won’t be the ones who build the most. They’ll be the ones who can connect signal to outcome faster than everyone else.
Want help putting this into practice?
At Product Rebels, we work with product leaders and teams navigating exactly this shift: using AI to accelerate delivery without losing product discipline or customer trust.
Is your team struggling to bring AI into your product operating model or establish the practices that lead to better customer and business outcomes?
Schedule 30 minutes with us to learn a little bit about you and explore how we can help.
About Product Rebels
Product Rebels helps product leaders bring their teams from good to great. We work with established product organizations that already know the basics of product management but want to operate at a higher level. Our focus is not Product Management 101. It’s helping teams build strong customer foundations and outcome-oriented ways of working that consistently translate into better results; for customers and for the business.
We partner with leaders and teams to change how product work actually happens day to day: how customer insight is gathered and shared, how problems are framed, how tradeoffs are made, and how learning compounds over time. AI is infused throughout these practices as an accelerator, helping teams synthesize learning faster, explore more options, and move with greater confidence without sacrificing judgment or customer connection.
The result is product teams that don’t just ship more, they build the right things, make better decisions under pressure, and deliver meaningful, sustained impact.
Sources and further reading
- Mind the Product, “The ROI Blueprint: Optimizing Design and Engineering for Economic Value” https://www.mindtheproduct.com/the-roi-blueprint-optimizing-design-and-engineering-for-economic-value/
- McKinsey & Company, “The State of AI in 2023” https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023
- IBM, “Global AI Adoption Index” https://www.ibm.com/reports/ai-adoption
- Microsoft, “2026 Work Trend Index Annual Report” https://www.microsoft.com/en-us/worklab/work-trend-index/copilot-will-fundamentally-change-work