The unit of research output. A structured template that turns a raw observation into an insight with a finding, an interpretation, and a “so what” — something a team can actually act on.
▸ Try the interactive toolThe Insight Statement Framework is a structured template for translating raw observations into actionable insight statements. An insight statement is the unit of output from any research effort — it's what goes into the Opportunity Solution Tree, the PRD, and the discovery read-out. The framework gives every insight a consistent, decision-ready shape.
The key distinction is between an observation (what you saw) and an insight (what it means and why it matters). “Four of five users hesitated at the pricing step” is an observation; the insight adds the interpretation (why they hesitated) and the implication (so what should change). A good insight statement bundles finding, meaning, and consequence so a reader can act on it without having to re-do the analysis.
The finding (what the data shows, with evidence) + the interpretation (what it means / why) + the implication (so what — the decision or opportunity it points to). Observation alone is not insight.
Research that stops at observations forces every reader to interpret the raw data themselves — so the insight either gets lost or gets re-invented inconsistently by whoever reads it.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Observations without meaning | “Users did X” leaves the reader to guess why it matters. | The framework adds interpretation and implication. |
| Inconsistent research output | Every researcher writes findings differently; nothing's comparable. | A consistent template makes insights reusable and stackable. |
| Insight lost in raw data | The decision-relevant point drowns in detail. | The statement distils the finding to its actionable core. |
| No link to a decision | Findings that don't point anywhere don't get acted on. | The ‘so what’ ties each insight to a decision or opportunity. |
What did the data actually show? Be specific and cite the evidence — how many users, which behaviour. This is the observation, the factual base.
What does it mean? Why did it happen? This is where observation becomes insight — grounded in the data, not invented, but going beyond mere description.
What decision or opportunity does this point to? An insight with no implication is trivia. The ‘so what’ is what makes it worth writing down.
Use the same shape every time so insights from different studies stack and compare. One crisp statement, not a paragraph of hedging.
Link the statement back to the raw observations (often from the affinity map) so anyone can verify it. An insight nobody can check is an opinion.
A research read-out listed observations: “users paused at the pricing step,” “several re-read the plan comparison,” “two abandoned at checkout.” Accurate, but inert — every reader had to work out for themselves what it meant and what to do, so mostly nothing happened.
Reframed as insight statements, the same data became actionable: finding (most users hesitated and re-read at pricing), interpretation (the plan differences weren't clear enough to choose confidently), implication (clarifying the comparison should reduce checkout abandonment — worth testing). Now the read-out pointed at a decision instead of leaving one to be reconstructed.
The deliverable is a set of consistent insight statements — finding, interpretation, implication — each traceable to evidence, ready for the OST and the read-out.
| Layer | Observation only | Full insight |
|---|---|---|
| Finding | “Users paused at pricing” | Same — with evidence |
| Interpretation | (missing) | “…because plan differences weren't clear” |
| Implication | (missing) | “…so clarifying it should cut abandonment” |
| Result | Reader must interpret | Reader can act |
The difference between an observation and an insight is the difference between data and a decision. Research that ships observations makes everyone else do the interpretive work; research that ships insights does it once, well.
The discipline that matters most is the ‘so what.’ It's tempting to stop at the finding — it's factual, defensible, and feels complete — but a finding without an implication is just trivia that sits in a folder. Forcing every insight to name the decision or opportunity it points to is what connects research to action; it's also a useful filter, because a finding you can't attach a ‘so what’ to may not be worth reporting. Consistent shape, evidence you can trace, and a clear implication: that's what turns a research effort into something the rest of the organisation can actually build on.
“Users did X” isn't an insight. Add what it means and what to do.
The ‘why’ must come from the evidence, not the team's theory. Keep it traceable.
An insight with no ‘so what’ is trivia. Tie it to a decision or opportunity.
If every insight looks different, they can't stack or compare. Use one shape.
Emergent themes (Tool 20) become the findings that insight statements build on.
Insight statements populate the opportunities branch of the OST (Tool 03).
Insights are the unit that flows into product specs and discovery presentations (Module 4).
Consistent insight statements are what make a repository (Tool 22) searchable and reusable.
AI helps turn observations into well-formed insight statements — and helps you check they're genuine insights, not restated data.
The judgment that stays yours: An insight is a non-obvious truth that changes what you build — recognising one takes domain judgment AI lacks. It can help you phrase and pressure-test, but you decide what's actually insightful.
Take a raw observation about a product (“people abandon the signup at step 3”). Add the two missing layers: what does it likely mean (interpretation), and what should change as a result (implication)?
Check the ‘so what’ — does your statement point to a specific decision or opportunity? If not, sharpen it.
If you struggle to attach a ‘so what,’ you've learned something useful: either the finding needs more interpretation, or it wasn't decision-relevant enough to report.