What is assumption mapping?
Assumption mapping is a short exercise where you write down every belief your startup idea depends on, place each one on a grid by how important it is and how much evidence you have, and test the important, unproven ones first. David J. Bland, co-author of Testing Business Ideas with Alex Osterwalder, describes the grid in his assumptions mapping article for Strategyzer: importance on one axis, evidence on the other, and experiments aimed at the top right corner.
Here is why it matters. Most founders test what feels comfortable: they build a landing page because they know how, or they interview friends because it’s easy. Assumption mapping forces you to test the belief that would sink the business if it turned out wrong, even when that test is awkward, like asking strangers for money.
When should you use it?
Use it when:
- You have an idea and a rough plan, and you’re about to spend real time or money on it.
- You’ve done some research (Google Trends analysis, a few interviews) and can’t decide what to test next.
- You work with a co-founder and you disagree about what’s risky. The grid turns opinions into a shared list.
It works for every type of business: B2B and B2C, apps, physical products, services. Skip it only if you have one obvious, make-or-break question and a way to test it this week. Then just run the test.
How do you do it, step by step?
-
List assumptions in 5 groups. Spend 30 minutes and aim for 25 to 40 sticky notes or rows. Customer: who has the problem, how many, where you can reach them. Problem: how frequently it happens, what it costs them, what they do today. Solution: that your fix works better than the workaround, that you can deliver it. Price: that they’ll pay your price, and how many times a year. Channel: how they’ll find you and what that costs. Start each note with “We believe…”.
-
Make each one specific. “Parents want this” is useless. “Parents in their third trimester in Birmingham will pay £60 for an at-home car-seat check” can be tested.
-
Draw the grid. Vertical axis: important (the business fails if this is wrong) at the top, unimportant at the bottom. Horizontal axis: no evidence on the right, strong evidence on the left. Evidence means something you observed: a payment, a sign-up, repeated stories in interviews, public data. Your gut feeling and your mum’s enthusiasm don’t count.
-
Place every note, fast. Ask two questions per note: “If this is wrong, does the business die?” and “What have I SEEN that supports it?” If you’re unsure, place it more to the right.
-
Circle the top right corner. These are your leap-of-faith assumptions: important and unproven. Pick the top 3. Bland also groups assumptions by desirability (do they want it), feasibility (can you deliver it) and viability (does it make money), which helps you check that your top 3 aren’t all from one group.
-
Turn the riskiest into a hypothesis with a pass line. Use this format: We believe [customer] will [measurable action]. We’ll know we’re right if [number] out of [number] do it by [date]. The number has to come before the test, or you’ll move the goalposts after.
-
Choose the cheapest test that produces that action. Payment beliefs need a pre-sale offer or deposit. Demand beliefs fit a fake door test. Problem beliefs fit interviews. Run it, record the result, and move the note left (evidence found) or rewrite the idea (belief failed).
What does a worked example look like?
Say Tanya is a former paramedic in Birmingham who wants to run a mobile service checking and fitting baby car seats at people’s homes. This is a hypothetical case.
She writes 31 assumptions. Some land bottom left quickly: “new parents worry about fitting car seats” is backed by 8 interviews where it came up without her asking. Others sit top right:
- New parents will pay £60 for a home visit.
- Midwives and antenatal class leaders will recommend her.
- She can do 4 visits a day across the city.
The payment belief is the riskiest: everything else is irrelevant if nobody pays. Her hypothesis: We believe expecting parents in their third trimester will book a paid home car-seat check. We’ll know we’re right if at least 5 of the 40 parents we reach through two antenatal classes pay a £20 deposit within 14 days.
She asks two class leaders to hand out a flyer with a booking link and a payment button. Say 7 parents pay the deposit. She moves the note left, then picks the next one: the referral belief, tested by asking 10 class leaders whether they’d share her flyer every month.
If only 1 had paid, she’d have rewritten the offer (a cheaper check, or a package sold through baby shops) before spending a penny on a van.
Which free tools help?
| Tool | What you use it for | Free plan |
|---|---|---|
| Miro | An online whiteboard for the grid, good with a co-founder | Free plan available |
| Google Sheets | A list version with importance and evidence scores from 1 to 5 | Free with a Google account |
| Sticky notes and a wall | The fastest version if you work alone | A few euros |
| Google Forms | A quick sign-up or deposit form for the first test | Free with a Google account |
What mistakes should you avoid?
- Writing vague assumptions. “People want convenience” can’t fail, so it can’t teach you anything. Add who, what action and how much.
- Counting opinions as evidence. Compliments, likes and “I’d totally use that” go on the right side of the grid. Only behaviour moves a note left.
- Testing the comfortable thing first. Building a website feels productive. If the riskiest belief is about payment, ask for payment.
- Setting the pass line after the test. Decide the number before you start, write it down and share it with someone.
- Mapping once. Every test changes the map. Redo it after each result; it takes 20 minutes the second time.
How do you know it worked?
You have a grid, 3 circled leap-of-faith assumptions and one written hypothesis with a number and a date. Our rule of thumb: a go means your riskiest assumption passed its pre-set line, and you move straight on to the next one. A no-go means it failed by a wide margin, and you change the customer, the price or the offer before you test again. A narrow miss earns one rerun with a sharper offer, never three.
What should you do next?
Test your top assumption with problem interviews, a fake door test or a pre-sale offer, depending on what it’s about. If you want help listing assumptions, an AI co-founder checklist walks you through validation one step at a time. The Fe/male Switch startup game also has you map and test assumptions with an AI co-founder called a PlayPal.
FAQ
What is the difference between an assumption and a hypothesis?
An assumption is a belief you hold, like “parents will pay for this”. A hypothesis is that belief rewritten as a test with a measurable action, a number and a date, so it can clearly pass or fail.
How many assumptions should I test?
Start with the top 3 from the important, no-evidence corner, and test them one at a time, riskiest first. Expect the idea to change shape after the first or second test.
Who created assumption mapping?
The grid of importance against evidence is closely tied to David J. Bland, who wrote about it for Strategyzer and in Testing Business Ideas with Alex Osterwalder.
Updated 26 October 2026 · Written by the Fe/male Switch team