Home/Blog/Product Discovery/What Does Validated Mean in Discovery? S…
Product Discovery

What Does Validated Mean in Discovery? Set the Bar

What does validated mean in discovery? Agree in writing what would change your mind and what evidence is enough to commit build capacity, before any test runs.

Cover of the FIKR Space guide: What Does Validated Mean in Discovery? Set the Bar

What validated means in discovery is whatever your team agreed, in advance and in writing, would be enough evidence to commit build capacity. Without that agreement the word means whatever people already wanted to do.

It is one of the quiet decisions that decides whether a discovery track produces real filtering or just a stream of confirmations. This post is about setting the bar honestly, and about what you put in front of it.

What does validated mean in discovery, and why must you define it?

Teams routinely declare an assumption validated on evidence that would not survive a sceptical question. The guide gives three examples: five friendly users, an internal opinion, and a survey where everyone said yes. Each one feels like evidence. Each is mostly a reflection of the people who gave it.

The problem is not that these sources are worthless. The problem is that nobody decided beforehand whether they would be enough. Once a result is in, it is very easy to decide it was.

That is why the guide insists that the bar must be stated. An unstated bar always moves to accommodate what people already wanted to do. A stated bar cannot move without somebody noticing and having to say why.

Two things to agree in advance

The guide names two parts, and both need to be settled before the test starts.

AgreeThe question to answer
What would change your mindIn each direction, what result would make us commit, and what would make us stop?
What counts as enoughWhat standard of evidence is sufficient to commit build capacity?

Note the words "in each direction". It is easy to define what would convince you to proceed. It is harder, and more important, to say what would convince you to stop. A bar that can only be cleared one way is a ritual, not a filter. If the team cannot name a result that would count as a no, the test is not ready to run.

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

The bar does not have to be high

A common worry is that defining validation will make discovery heavy. It does not need to. The guide is explicit that different decisions warrant different rigour.

DecisionReasonable weight of evidence
A small, reversible changeFar less than a quarter of engineering would need
A large commitment of build capacityA higher standard, agreed and written down

So the aim is not a universal threshold. The aim is that for each decision, the threshold exists before the data does. A light bar that is stated beats a heavy bar that is implied.

Weak evidence and strong evidence

It helps to sort what you collect into what carries weight and what does not. The guide gives several signals that point in different directions.

  • Enthusiasm is weak evidence. People are generous in conversation.
  • Commitment is strong evidence. A small commitment beats a large compliment.
  • Past behaviour is evidence. Predicted behaviour is imagination.
  • Friendly or convenient respondents weaken a result. People who match the target strengthen it.

When you write the bar, borrow from this list. A reasonable bar might say that warm words alone do not count, that the people spoken to must match the target, and that some form of commitment is required. The technique for gathering that kind of evidence is covered in how to interview users without leading them.

Put assumptions into discovery, not features

The bar only means something if what you are validating is a genuine question. This is the second half of the decision, and it is easy to miss.

Discovery items should be phrased as questions or risks, not as solutions waiting to be validated. The guide contrasts a question about whether users can complete setup without help, which is a discovery item, with an instruction to build the new onboarding, which is a delivery item that has skipped the question.

The reason is practical. A discovery track full of pre-decided features is a delivery backlog with a longer approval process, and it will produce exactly the confirmations you went looking for. You cannot validate a feature. You can only validate or invalidate the beliefs on which it depends.

So the sequence runs as follows. Start from the risk. Phrase it as an assumption. Agree the bar. Then test. Reversing the order, so that the bar is fitted afterwards to the result you got, is the most common way a track becomes theatre. To pick which assumption to put at the front of that sequence, read how to test your riskiest assumption.

Writing the bar down

A written bar can be short. For each discovery question, a few lines are enough.

  1. The assumption, phrased as a question or risk.
  2. A result that would count as a yes, and what we will do then.
  3. A result that would count as a no, and what we will do then.
  4. The evidence that is enough to commit build capacity.
  5. The date by which we decide.

Keep this with the question itself, so that anyone looking later can see what was promised before the data arrived. That also makes it easy to review in a retrospective, which the guide recommends doing on the reasoning and not only the output. It lets you ask which tests actually changed a decision and which were theatre.

Who decides when the bar is met

A bar needs a judge. The guide asks teams to decide who says "we have learned enough, this moves to delivery". Without a clear answer, one of two things happens. Either everything graduates and the discovery track becomes a formality, or nothing does and the team collects well-researched ideas that never ship.

Name the person or small group, and give them permission to say both "not yet" and "we know enough, stop researching". Permission to call time matters just as much as permission to ask for more, and it is far less common. It is also what stops the bar from ratcheting upward until nothing can ever meet it, which is the way discovery becomes precious about certainty. Once something clears the bar, it should travel onward with its reasoning attached, as explained in how to hand off discovery to delivery.

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

Reading results against the bar

Once the bar exists, results become much easier to read. Lay each one beside what you wrote down beforehand.

What came backWhat the written bar saysWhat happens
Meets the standard of evidenceEnough to commit build capacityThe item graduates, with its reasoning
Meets the "no" conditionWe agreed we would stopKill it, and record what was learned
Falls between the twoNeither condition was metName the one thing to test next and set a new date

Because the conditions were agreed first, nobody has to argue about whether a mixed result is "good enough". The argument already took place, calmly, before there was anything to be defensive about.

Validated is not the same as finished

An assumption that clears the bar is validated for the decision it was attached to. It is not proven for ever. Markets move, and the confidence attached to a finding drops with every month it goes untested. This is why the guide treats delivery as the mechanism that validates discovery: a conclusion nobody shipped is a hypothesis with better manners.

The loop therefore continues after shipping. The prediction attached to the work should be checked against what actually happened, which is the subject of how to close the loop after shipping. If the outcome did not move, your bar was too low, and the retrospective is the place to say so.

Calibrating the bar over time

A bar is a best guess, and the way to improve a guess is to compare it with reality. The guide suggests asking four questions in the retrospective, on the reasoning and not only the output.

  • Which assumptions turned out to be wrong?
  • Where were we overconfident?
  • Which tests changed a decision, and which were only for show?
  • How long did items really take compared with our estimates?

If several assumptions that cleared the bar later proved wrong, raise the standard for that kind of decision. If a bar demanded more than the decision deserved, lower it. Half an hour a cycle is enough, and it is what makes the whole practice compound.

Signs your bar is not honest

  • It was written after the results came in.
  • Nobody can name a result that would have counted as a no.
  • Everything that has been tested has been validated.
  • The evidence cited is friendly users, internal opinion or a survey in which everyone agreed.
  • Negative results are explained away rather than recorded.

The second to last sign is the most telling. A track that has never produced a decision not to build is not acting as a filter at all.

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

A checklist for the bar

  • We have agreed what validated means, and what evidence is enough to commit capacity.
  • The bar is written down, so it cannot quietly move to suit what we wanted.
  • It sets out what would change our mind in each direction.
  • The item being tested is an assumption, not a feature.
  • One named person or small group decides when it is met.
  • They can say "not yet" and they can say "we know enough".
FIKR OS, strategy meets execution: OKRs, roadmaps and kanbans in one system. See FIKR OS.

Frequently asked questions

Does defining validated slow discovery down?

Only a little, and it saves more than it costs. The bar can be a few lines long, and it stops arguments about whether a result was good enough, which would otherwise happen after the data arrives.

Can the bar be different for different questions?

Yes, and it should be. Different decisions warrant different rigour. A small reversible change needs far less evidence than a quarter of engineering.

What if the result lands between yes and no?

Decide with what you have. If the honest answer is that you still do not know, write down the one thing you would test next, and set a new date.

Who should set the bar?

The team, together, before the test begins, with the person who decides what graduates having the final word. The point is that it exists in writing before anyone knows the result.

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