What Founder Self-Assessments Can and Cannot Tell You

A test that flatters you is entertainment. A test that changes next week's calendar is a tool.

Founder self-assessments are stuck between two extreme positions: sold as deep insight, or dismissed as astrology. Both are wrong, and the difference lies in knowing exactly what these instruments measure.

The honest scope

A self-assessment measures your stated preference and your self-perception — what you think you prefer and what you think you are. Not capability, and not real behaviour under pressure.

This is not a limitation a better test could remove. It is inherent to the method: the only data source is your own answers.

Which also causes people to undervalue it. Stated preference is real data. That you believe you decide quickly, even when wrong, says something important about how you interpret your own behaviour.

What they measure reliably

• Preference between equally valid options. Which you say you choose when speed and certainty conflict.
• Internal inconsistency. When your answers to two related questions do not agree, that inconsistency is itself a finding.
• The gap between self-perception and role expectation. Between what you think you are and what your current role demands.
• Change over time. Retake the same instrument six months later and the direction of change is meaningful, even when the absolute number is not.

The fourth has the most value and the least use, because almost nobody takes the test twice.

What they cannot

Row three is the important one. Behaviour under pressure is frequently the inverse of stated preference, and no questionnaire captures it. The only source for that is a log of your real decisions.

Turning a result into one changed decision

A result that produces no change gets read and forgotten. A three-step method that works:

1. Pick one statement in the result that made you uncomfortable. Not the ones that validated you — those add nothing.
2. Find a real instance from the last three months that confirms or refutes it. If you cannot find one, the statement is meaningless for you, however correct it may be.
3. Name one specific decision you will make differently next week. One decision, not a general behaviour change.

Step two is the real filter. Most test results are general enough that any instance confirms them. If you cannot find a specific case, that statement is unusable.

Reading values, attitude and decision style together

Any single instrument is limited. Where information actually gets produced is in the contradictions between them.

If your stated values rank stability and predictability highly while your decision style is fast and experimental, that contradiction is a genuine finding. It probably means you work in an environment demanding behaviour that does not match your preference — and that gap is where burnout is manufactured.

Similarly, if your attitude toward risk is conservative while your decision style says you decide fast, one of those two self-perceptions is inaccurate. Only a real log tells you which.

Why these instruments have a bad reputation

Scepticism about them is justified, and it has three roots.

The Barnum effect. Statements general enough to apply to everyone — "you are sometimes introverted and sometimes outgoing". When half a report reads like that, the whole report loses credibility, even where the other half is accurate.

Categories instead of spectra. Converting a continuous trait into four types destroys information and becomes identity. People start explaining their behaviour by their type, which is precisely the opposite of the intent.

Predictive claims. An instrument claiming to predict job performance has left its own scope. Stated preference is a weak predictor of performance.

A sound instrument abandons all three: it makes specific statements, preserves the spectrum, and states plainly what it does not measure.

The honest alternative: a behaviour log

If you want to know how you actually decide, no questionnaire beats a two-week log.

The method is simple: for two weeks, record every decision in one line — topic, time spent, reversible or not, and whether you decided alone or consulted. Write no interpretation in the moment.

At the end the patterns present themselves, and unlike a test, this data is behaviour rather than perception.

Combining the two is strongest: the test generates the hypothesis, the log tests it. Neither is sufficient alone.

A worked example: a result that was rejected and was correct

A founder rejected their decision-style result. The report said they postponed important decisions. Their reaction was that this was the opposite of reality and they were known for deciding quickly.

We ran step two: find instances from the last three months. They named three fast decisions, all three genuine.

Then we looked at the calendar log. All three fast decisions were reversible ones. The two irreversible decisions of that period — a senior hire and a price change — had taken eleven and seven weeks respectively.

Both pictures were correct. They were fast on reversible decisions and slow on irreversible ones. The test did not see that distinction and reported only the average.

The real finding did not come from the test. It came from the contradiction between the result and the real log — and that contradiction would never have been examined without the test.

So what are they actually worth

A self-assessment is a starting point for investigation, not a conclusion. Its value is in the hypotheses it generates, not the answers it gives.

And one under-discussed value: shared language. When two co-founders each hold their own result, the conversation about their differing styles moves from the personal level to the structural one, which is considerably more manageable.

Are free tests as valid as paid ones?

Validity depends on length and question structure, not price. Anything taking under five minutes usually produces nothing but validation.

How often should they be retaken?

Every six to twelve months. Sooner and you are looking at response noise rather than real change.

What if the result contradicts my self-image?

That is the best possible outcome. A result that matches your self-image entirely carries no new information.

Closing

A self-assessment does not tell you who you are. It tells you who you think you are — and the gap between those two is the only part genuinely worth examining.

This site's three instruments are on the self-discovery page, with exactly these limits.