“Data-driven” is one of the most abused phrases in product. Data should inform judgement, not replace it — and the difference between the two is where good and bad decisions diverge.
▸ Try the interactive tool“Data-driven” is one of the most abused phrases in product management. It gets stitched onto roadmaps, repeated in standups, and printed on slides — usually to lend a decision the authority of evidence it never actually had. The honest stance is data-informed: data is a crucial input to judgement, not a substitute for it.
The distinction matters because data and judgement each have failure modes. Pure “data-driven” decision-making treats numbers as if they speak for themselves — but data is always partial, often lagging, and silent on causation and on everything it doesn't measure. Pure intuition ignores evidence. The data-informed PM holds both: they take the data seriously, understand its limits, and then apply judgement to decide. Data narrows the uncertainty; it rarely eliminates it.
Data-driven — the numbers decide; judgement is abdicated.
Data-informed — data is a serious input, its limits understood, and judgement makes the call.
Data narrows uncertainty; it doesn't remove the need to decide.
Both extremes fail: worshipping data abdicates judgement to numbers that can't see causation or context, while ignoring data flies blind. Data-informed is the disciplined middle.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Worshipping the numbers | Treating data as objective truth ignores its blind spots and biases. | Data-informed takes data seriously while knowing its limits. |
| Decisions by gut alone | Ignoring evidence flies blind and rationalises bias. | Data-informed grounds judgement in real evidence. |
| ‘Data-driven’ as cover | Citing data to lend false authority to a pre-made decision. | Honest data use informs the decision, not decorates it. |
| Mistaking correlation for cause | Acting on a number that shows association, not causation. | Judgement interprets what the data can and can't establish. |
Gather and weigh the relevant evidence honestly — don't cherry-pick the numbers that fit your hope. The data-informed PM starts by genuinely listening to what the data says.
Ask what this data can't tell you: causation, context, the unmeasured, the lag. Every metric has blind spots, and knowing them is what separates informed use from worship.
Numbers don't speak for themselves — someone has to decide what they mean and what to do. This is where experience, context, and qualitative insight come in alongside the data.
Make the decision as a human judgement informed by evidence — not a calculation the data made for you. Own it, rather than hiding behind ‘the data said so.’
The data tells you what; qualitative research tells you why. The strongest decisions combine both, which is the whole spirit of being informed rather than driven.
A team faced a decision and pulled the relevant numbers, which pointed fairly clearly in one direction. A purely ‘data-driven’ team would have stopped there and executed. But the data, on inspection, had real limits — it captured behaviour but not the reason behind it, and a recent change meant the trend might not hold.
Being data-informed, the team took the numbers seriously, named what the data couldn't tell them, added qualitative insight about why users behaved that way, and then made a judgement call — which differed slightly from what the raw numbers alone would have dictated, and proved better for it. They used the data as a serious input, not an autopilot.
The deliverable is a decision that takes the evidence seriously, names its limits, and applies explicit judgement — not a number presented as a verdict.
| Data-driven (trap) | Data-informed (goal) |
|---|---|
| Numbers decide | Numbers inform; judgement decides |
| Data treated as truth | Data's limits understood |
| Quant only | Quant (what) + qual (why) |
| Hide behind 'the data' | Own the human call |
“Data-driven” is often a way to avoid the responsibility of deciding — to let a number take the blame. “Data-informed” keeps the human in the loop, which is exactly where the hard, valuable judgement lives.
The deeper point is that data and judgement are complements, not competitors. Data is indispensable precisely because gut-alone decisions rationalise bias and ignore evidence — but data alone is dangerous because it's partial, often lagging, silent on causation, and blind to everything outside what's measured. The PM's skill is holding both: taking the numbers seriously enough to be challenged by them, while understanding their limits well enough not to be ruled by them. That's why pairing quantitative (what happened) with qualitative (why) is the recurring move — it's the practical form of being informed rather than driven, and it's the foundation everything else in this module builds on.
All data is partial and has blind spots. Take it seriously, but know its limits.
Citing data to justify a pre-made decision isn't analysis. Let evidence genuinely inform.
Numbers show what, not why. Pair them with qualitative insight.
The data can't make the decision. Own the human call.
Every metrics tool that follows is meant to inform judgement, not replace it.
Knowing which metrics matter (Tool 02) is the first step in using data well.
The 'why' behind the data comes from Module 3's research methods.
Treating data as truth is the root of many anti-patterns (Tool 27).
Recall a decision justified as 'data-driven.' Ask honestly: did the data actually make the call, or was it cited to lend authority to a decision already made?
Then ask what the data couldn't tell you in that case — the causation, context, or 'why' it was silent on.
If the data was decorating a decision rather than informing it, you've found the abuse the phrase 'data-driven' so often hides — and why 'data-informed' is the more honest standard.