Journal / Product
When AI becomes your product team's sixth sense
There is a moment every experienced product manager knows. You are in a meeting, someone asks why a metric moved, and the most senior person in the room offers an answer before the data is pulled. Confident. Immediate. Often wrong.
The problem with intuition
Product intuition is pattern recognition built on a limited, personal sample. When a senior PM looks at a funnel and immediately suspects the drop-off is a copy problem rather than a UX problem, they are drawing on their own history — a handful of past products, a particular user base, a specific era. That history is real, but it is also narrow, and it carries every bias the person has accumulated along with it.
The deeper problem is that intuition is invisible. It cannot be challenged, audited, or improved systematically. When an intuition-driven decision fails, the lesson is absorbed privately and rarely surfaces the structural issue. When it succeeds, it reinforces confidence in a method that does not scale.
Teams that run on the intuition of their most experienced members are one resignation away from starting over. They are also one blind spot away from a costly mistake that better data would have prevented.
The signal was always there
Most product teams are sitting on more customer signal than they know what to do with. Support tickets. Session recordings. In-app surveys. App store reviews. Sales call transcripts. Churned customer interviews. NPS verbatims.
The problem has never been a lack of signal. It has been the cost of processing it. A researcher can synthesise fifty interviews into themes over a week. Nobody has the bandwidth to synthesise five thousand support tickets into actionable patterns every sprint. So the signal sits unread, or gets sampled, or gets summarised once a quarter in a report that is already out of date. In the gap, intuition fills in.
AI removes the processing bottleneck. What previously required a team of researchers and analysts working for weeks can now run continuously in the background. Customer language shifts get flagged in real time. A new complaint theme emerging in support tickets surfaces before it shows up in churn. A segment of users who are quietly succeeding with the product in an unexpected way becomes visible before anyone thought to look for them.
The signal was always there. The question is whether you build a culture that reads it, or one that guesses instead.
Decisions grounded in evidence
The practical change is not just speed. It is the nature of the questions teams feel entitled to ask — and the expectation that those questions get answered with data before a decision is made.
In a traditional setup, pulling insight has a cost: writing a brief, waiting for a researcher, waiting for analysis, scheduling a readout. That cost creates a filter. Teams only ask the questions that feel important enough to justify the overhead. Small hypotheses go untested. Weak signals get ignored because chasing them does not seem worth a two-week research cycle. Intuition steps in to cover the gaps.
When insight is cheap and fast, the filter disappears. A PM can ask "are the users who complete onboarding in under five minutes retaining better at 30 days?" before a planning meeting and have a real answer by the time the meeting starts. A designer can ask "what do users say about this screen in support tickets?" before a redesign and build on actual language rather than assumptions.
Small questions, answered quickly, compound. Teams that ask more questions make better decisions — not because any single answer is transformative, but because the cumulative effect of testing more hypotheses is a much more accurate model of what is actually happening. No intuition required.
What data-driven product work looks like in practice
Customer feedback becomes a live input rather than a periodic report. Instead of a monthly NPS summary, the team has a dashboard that surfaces emerging themes across all feedback channels, updated continuously. New issues appear within days of emerging, not weeks. Decisions are made on current data, not a snapshot from last quarter.
Hypothesis generation becomes a team sport. Because everyone has access to the same customer signal, the designer and the developer are forming hypotheses alongside the PM, all grounded in the same evidence base. No single person's gut read dominates the room.
Pre-mortems get sharper. Before shipping a feature, the team surfaces every piece of customer feedback related to the problem being solved, every similar feature that has been tried, every segment that is likely to respond differently. "What could go wrong" gets answered with evidence rather than speculation.
Onboarding accelerates. A new PM or designer can query the product's entire customer feedback history on day one. The institutional knowledge that previously lived only in the heads of people who had been around for three years is now accessible to anyone who knows how to ask. No years of accumulated gut feel required.
What this does not replace
None of this removes the judgement required to act on insight. Data can tell you that users are confused at a particular step. It cannot tell you whether the right fix is better copy, a different flow, or removing the step entirely. That still requires a human who understands the customer, the product constraints, and the strategy well enough to make a call.
And it does not replace qualitative research conducted by skilled researchers, which still surfaces nuanced, contextual understanding that quantitative signal alone cannot produce.
What it replaces is the gap between those moments of structured understanding — the weeks when the team is operating on assumptions because the last research cycle was two months ago and the next one is not scheduled yet. That gap is where intuition does its damage.
The team that knows, wins
Product is fundamentally an information problem. The team that understands its customers most accurately, most completely, and most quickly will make better decisions than the team that does not. That has always been true.
What has changed is the cost of that understanding. For most of product management's history, deep customer insight was expensive, slow, and available only to teams with dedicated research functions. AI is making it cheap, fast, and available to everyone.
There is no longer a good reason to guess. Build a team that knows.
