UX1: Thinking like a product designer
Evidence, assumptions and how sure you are
🎯You'll be able to
Sort design beliefs into facts, evidence-backed beliefs and assumptions, and choose which assumptions to test first.
🏠The everyday picture
A shopkeeper who "knows" customers want a new brand can be right or can be stuck with boxes of unsold stock. A small trial order first costs little and settles the question.
📘What it really is
- Kinds of evidence:
- What people do (observed in sessions, or counted in analytics) is the strongest for behaviour.
- What people say (interviews, surveys) is good for goals, feelings and reasons, but people predict their own behaviour poorly.
-
Expert opinion (heuristics, guidelines) is useful and quick, but it's still an educated guess about a specific product.
-
Assumptions are beliefs nobody has checked. Write them as testable statements: "Most patients book for someone else" rather than "patients are busy".
-
Risk and uncertainty: test first the assumptions that are both uncertain and important. A wrong guess about button colour costs little; a wrong guess about who books appointments can sink the product.
-
Small samples: five sessions can show that a problem exists and roughly how bad it is. They can't show what share of all patients have it. Surveys and analytics answer "how many?".
-
Say how sure you are. "4 of 5 participants couldn't tell which time they'd chosen" is honest. "80% of users can't use the picker" is not: five people aren't a measure of all users.
📝A worked example
Input: four beliefs the SehatSlot team holds.
| Belief | Kind | Important? | Uncertain? | Next step |
|---|---|---|---|---|
| Most people book on a phone | Evidence: survey, 79% of 120 | Yes | Low | Design phone-first |
| People can't tell which slot is chosen | Evidence: 4 of 5 sessions | Yes | Low for "it happens"; unknown how often | Fix, then re-test |
| Older patients prefer to call | Assumption | Yes | High | Ask in the next round; check call logs |
| Patients want a dark theme | Assumption | Low | High | Wait |
Expected output: the "older patients prefer to call" assumption goes into the next study plan, because it's important and uncertain.
🌍Where it's used
- Product teams keep an assumption list and review it at each planning meeting.
- Charities and public services pilot a new form in one district before rolling it out.
- Start-ups test the riskiest assumption with the cheapest experiment, sometimes a paper prototype.
📝Case scenario
A tutoring app assumed parents would pay monthly. Interviews with 12 parents showed most preferred a term fee they could plan for. The team tested both options on the pricing page with a small group before changing the app.
🖱️Try it
List five beliefs you hold about how people book doctors. Mark each as fact, evidence or assumption, and as important or not. Success: at least two important assumptions, each written as a testable statement.
⚠️Common mistakes
- Reporting small samples as percentages. Sign: "60% of users". Fix: "3 of 5 participants".
- Trusting what people say they'd do. Sign: "they said they'd use it daily". Fix: watch what they do, or test with a prototype.
- Testing what's easy, not what's risky. Sign: weeks on colours, nothing on the core journey. Fix: rank assumptions by importance.
🛠️For your project
Project 1's test 6 checks that you count task results exactly as the notes show; reporting "x/5" keeps the sample size visible.
✅Quick check
Which is stronger evidence about behaviour: what people say or what they do?
Show the answer
What they do.
How should you report a problem seen by 3 of 5 participants?
Show the answer
As "3 of 5 participants", not as a percentage of all users.