Home/Blog/Product Discovery/How to Test Your Riskiest Assumption Bef…
Product Discovery

How to Test Your Riskiest Assumption Before You Build

How to test your riskiest assumption: list what must be true, rank each item by the damage if it is wrong, then run the cheapest test you would really obey.

Cover of the FIKR Space guide: How to Test Your Riskiest Assumption Before You Build

To test your riskiest assumption, list everything that must be true for the idea to work, rank those items by how damaging it would be to be wrong, and test the top one first with the cheapest test that could genuinely change your mind. The risky assumption is rarely the interesting one, and that gap is where most wasted work comes from.

This post is about that single step in discovery: choosing what to test and how to test it, so that you find out in days what you would otherwise find out at launch.

How to test your riskiest assumption, step by step

There are four steps, and the order matters.

  1. Write down what must be true.
  2. Rank by damage, not by curiosity.
  3. Choose the cheapest test that could move your belief.
  4. Agree in advance what you will do with each result.

The rest of this post takes them one at a time. Where a step opens a larger topic, there is a link to the post that covers it.

Step one: write down what must be true

An idea is a bundle of beliefs. Before testing any of them, make the bundle visible. For most ideas the beliefs fall into a few kinds.

Kind of beliefWritten as a question
The problemDo people actually have this problem?
BehaviourWill they change what they do?
UnderstandingWill they understand it?
Willingness to payWill they pay?
FeasibilityIs it technically possible to build?

Phrase every item as a question or a risk and never as a solution. This is the difference between a discovery item and a delivery item that has skipped the question. If a pre-decided solution is sitting on the list, take it off and write the doubt behind it instead.

Free guide, 34 pages: The Dual-Track Playbook. Send me the guide.

Step two: rank by damage, not by interest

This is the step teams get wrong. Asked what to test first, they reach for the assumption that is easiest or most enjoyable to look into. That is usually usability. It is comfortable, it is measurable, and it feels like progress.

Meanwhile the belief most likely to sink the idea, usually whether anyone cares, stays untested until launch. By then the cost of finding out is as high as it can possibly be.

A useful way to rank is to ask of each item: if this turned out to be false, would the rest of the idea still matter? The items for which the answer is no go to the top. Everything else can wait, however interesting it is.

A common trap

Usability questions feel rigorous because they produce clean results. But a perfectly usable thing that nobody wants is still a failure. If your list has a question about whether people care and a question about whether they can find a button, the first is the riskier one almost every time.

Step three: choose the cheapest test that could change your mind

The word "cheapest" does most of the work in discovery. Building is the most expensive way to test an assumption, so before reaching for it, ask what is the least expensive thing that would move your belief.

  • A conversation with people who match the target
  • A landing page that describes the idea
  • A prototype people can react to
  • A manual version of the service, done by hand
  • A concierge run with a handful of customers
  • A look at data you already have

The phrase "could change your mind" does the rest. A test is only worth running if one of its possible results would leave you thinking differently. If every result would leave your plan unchanged, you are not testing anything. You are decorating a decision already made.

The question that separates tests from theatre

Before running anything, ask: if this comes back negative, will we actually stop?

Your honest answerWhat to do
Yes, we would stop or change courseRun the test
No, we would build it anywaySkip the test and build, which at least is honest
We are not sureDecide now, before the result arrives, what a no would mean

A test you would ignore is worse than no test. It consumes time and it manufactures false confidence, so the team feels reassured by a result that was never allowed to count.

Step four: agree the verdict before the data

People are remarkably good at reading ambiguous results in favour of what they already wanted. The defence is to write the verdict down first. For each test, agree three things before it starts:

  • What result would count as a no.
  • What standard of evidence is enough to commit build capacity.
  • The date by which you will decide.

The standard does not have to be a high one. A light, reversible change can rest on much less evidence than a quarter of engineering. What matters is that it is stated, because an unstated bar always moves to suit what people wanted to do anyway. For more on setting that bar honestly, read what does validated mean in discovery.

When the test is a conversation

Many of the cheapest tests involve talking to people, and conversations have their own way of going wrong. If you describe the solution and ask whether people like it, you will receive politeness. To test the assumption that people have the problem at all, ask what they do about it today and when they last did it. That technique is the subject of how to interview users without leading them.

FIKR OS, strategy meets execution: OKRs, roadmaps and kanbans in one system. See FIKR OS.

What to do with the result

Every test ends in one of three outcomes, and each should lead to an action.

OutcomeAction
The assumption holdsMove to the next most dangerous assumption, or let the work graduate if the bar is met
The assumption failsKill or reshape the idea, and record what you learned
You still do not knowName the one thing you would test next and timebox it

A failed assumption is not a failed test. It is the most valuable kind of output, because it returns a quarter of capacity to better uses. Make sure it is written down, so the same idea does not return in six months with nobody remembering why it was dropped. Practical advice on that is in how to keep a discovery record.

Why the order is so hard to keep

If ranking by damage is so simple, why do teams keep ranking by comfort? The reasons are human, and naming them helps.

  • Comfortable questions feel like progress. A measurable usability result arrives quickly and looks like work done.
  • Decisive questions are uncomfortable. Asking whether anyone cares risks an answer nobody wants.
  • The solution is already loved. When a feature is pre-decided, the team is hunting for confirmation, not for the weak point.
  • Building feels safer than asking. It produces visible output, and delivery always has a visible cost to stopping.

None of these is a flaw of character. They are the default pull, and the only reliable counter is a written ranking that the team agreed before the first test began.

Using the guide's own example

The guide gives a clean illustration of the difference between a testable assumption and a disguised feature. "Will users complete setup without help" is a discovery item: it is a question, it can be answered cheaply, and a negative answer would change what you do. "Build the new onboarding" is not a question at all. It is a decision wearing the costume of research.

Apply the same move to your own list. Take each line that begins with a verb such as build, add or launch, and ask what doubt sits behind it. The doubt is the assumption. The line itself is a feature, and features belong on the delivery side only once they have earned their evidence.

Who should be there

Testing the riskiest assumption is not a job for a separate research group. The engineers and designers who will build the thing should be present in the discovery that shapes it. Not for every interview, but enough that the reasoning is shared rather than handed over as a document later. They are also often the people best placed to say which assumption is the most dangerous, especially the feasibility one that the rest of the team cannot rank alone.

It also helps to link every question to a goal. Discovery attached to no objective wanders. Attached to one, it has a direction, and finding a cheaper route to the same outcome counts as success. For the bigger picture of how the tracks fit together, start with what is dual track agile.

Free guide, 34 pages: The Dual-Track Playbook. Send me the guide.

A short checklist

  • For each idea, the assumptions that must be true are listed.
  • They are ranked by how damaging it would be to be wrong.
  • We are testing the most dangerous one, not the most convenient one.
  • We chose the cheapest test that could genuinely change our mind.
  • If it comes back negative, we have agreed we will actually stop.
  • The test has a date to decide.
  • We have said which result counts as a no.
FIKR OS, strategy meets execution: OKRs, roadmaps and kanbans in one system. See FIKR OS.

Frequently asked questions

How do I know which assumption is riskiest?

Ask which one would do the most damage if it were wrong. If the whole idea would stop mattering, that assumption is at the top. Questions about whether anyone cares usually outrank questions about how something is used.

What if the cheapest test feels too small to trust?

Match the test to the decision. Small reversible changes warrant light evidence. Larger commitments need a higher bar. State the bar before you start so it cannot move afterwards.

Should I test several assumptions at once?

The guide recommends starting with the most dangerous one. Testing one question at a time keeps the result readable and the effort small, and the next question is easier to choose once you know the answer to the first.

What if I am sure the assumption is true?

Then be honest about whether a negative result would change anything. If it would not, the guide's advice is to skip the test and build. If it would, run it, because certainty is not the same as evidence.

The full method, with the checklists, is in The Dual-Track Playbook, a free 34-page playbook.

LG

Luis Goncalves

// Founder & CEO at FIKR Space

Three-time founder. Built and exited Evolution4All before this. Now building FIKR Space — the operating infrastructure underneath every innovation ecosystem (startups, accelerators, governments, investors). Lisbon-based, works global.

Claim your founding member spot.

Every founding member is hand-reviewed. Tell us who you are and how you want to co-create — we get back inside 48 hours.

Role
Job Title
Country
Cookies

We use cookies.

Essential ones keep the site working. The rest only run if you say so. No dark patterns, no pre-ticked boxes.Privacy policy

Co-create with us. It's about building it together.Co-Create With Us