Article: 1 OF 3
A year ago, AI adoption among product managers was still a question mark. Would it be a short-lived novelty or a lasting tool?
Our 2025 Survey of the Product Management Profession found 49% of product people using AI frequently. Now in 2026, that number is 69%. In twelve months, AI has gone from something half the profession was experimenting with to something the majority relies on daily. It’s no longer a novelty. It’s part of the infrastructure of modern product work.
And by most measures, it’s delivering. When we surveyed 677 product managers across 40 countries for our 2026 Survey, 97% reported that AI has improved their personal productivity. Faster research and analysis. Quicker document drafts. Less time wrestling with the repetitive tasks that used to eat into the week. More time — at least in theory — for the high-value work that matters.
If you’d stopped there, you’d have a straightforward adoption success story. A new technology arrives, people use it, it makes them faster. Case closed. The efficiency case for AI in product management is essentially settled.
The twist
But here is what caught our attention.
When we asked those same product managers whether AI had improved their product outcomes — better decisions, better products, better results for the business — the number dropped to 64%.
The 33-point gap
Let’s pause here for a moment. 97% say they’re more productive. 64% say it’s making a difference where it counts. That’s a 33-point gap. A third of the productivity gain is evaporating before it reaches the business.
This isn’t what you’d expect. If product managers are faster, better-informed, and spending less time on administrative work, they should be driving better product outcomes. The strategy should be sharper, roadmaps better, launches more frequent, customer problems more cleanly solved. But for roughly a third of PMs, the speed is real and the outcomes are not keeping up.
You’re not imagining it
If this pattern feels familiar beyond product management, it should. McKinsey’s State of AI research found a similar dynamic at enterprise level: 88% of organisations are now using AI, but only 6% qualify as high performers seeing significant business value. Almost everyone has adopted the technology. A remarkably small minority is capturing its full potential.
Economists have a name for this. Brynjolfsson, Rock, and Syverson call it the Productivity J-Curve — every transformative technology, from electrification to the personal computer, shows a lag between adoption and results. The value comes from the complementary investments organisations make around the technology, not from the technology itself. Without those investments, you get adoption without impact.
Our data suggests product management may be in that phase. The AI tools are in place. The complementary investments have not caught up.
The symptoms
The survey gave us texture on where the gap shows up in practice. Product managers told us about AI-generated outputs that went unverified — plausible-sounding analysis that nobody checked against reality. About less experienced PMs leaning on AI as a substitute for the judgement that comes from years of customer conversations and failed launches. About speed without direction: doing more, faster, without anyone pausing to ask whether it was the right thing to do.
These feel like symptoms, not root causes. Over-trust, lack of verification, inexperience — they point to something deeper. The question that interested us was:
What separates the teams where AI productivity does translate into better product outcomes from the teams where it doesn’t?
What we went looking for
We went into the data expecting the answer to be about AI maturity. Better prompting. More sophisticated tooling. More experience with the technology itself.
We found an organisational story.
The gap had very little to do with AI itself, and almost everything to do with the organisational conditions surrounding it. Five conditions, specifically, kept showing up in the data. Product managers who had all five in place were over five times more likely to report that AI was significantly improving their product outcomes. 44% versus 9%. The difference was not in the tool. It was in the system the tool was applied to.
And only 8% of respondents had all five conditions in place.
The answer wasn’t better AI. It was what surrounded it.
Next in this series: It’s Not Just the AI. It’s the Organisation.
Based on the 2026 Survey of the Product Management Profession (677 respondents, 40 countries). Read the full survey report
Leave a comment