The Worst Way to Use a Personality Test (Avoid This)

A mid-sized tech company put every new hire through an MBTI workshop in their first week. By month six, two patterns had set in. New ENTJs were assumed to want the team-lead track whether they wanted it or not. New INFPs were quietly routed away from client-facing work. Neither group had been asked. The test had decided for them, and the assumption had hardened into a soft career ceiling.

This is the textbook failure mode of personality testing at work. The test does not break anything by itself. The way the test gets used breaks things. Pittenger's review of the MBTI's psychometric track record is worth a read for anyone running these programs ¹. But the deeper issue is not which test you pick. It is what people do with the result.

These are the four ways teams quietly hurt themselves with personality tests, and what to do instead.

Failure mode one: using results to sort people

The most common misuse is also the most expensive. A team takes a test. Results get shared. Within a week, people start sorting each other into types and assigning jobs accordingly. The "details person" gets the details work. The "people person" gets the people work. Everyone is locked in.

What the research says about this: type-based sorting is unreliable because most personality dimensions are continuous, not categorical. The MBTI in particular has well-documented test-retest issues — a meaningful share of people who retake it within a few weeks come out a different "type" ¹. Sorting humans into 16 boxes when the boxes themselves are unstable is a recipe for putting the right person in the wrong slot.

What to do instead: use results as information, not assignment. A trait score is a hypothesis about what kind of work may feel sustainable for someone. The actual job conversation should still involve the person.

Failure mode two: hiring on personality

A second pattern shows up in hiring. A team decides certain types are right for certain roles, and starts screening candidates against the profile. This feels rigorous. It is usually not.

Two problems. First, the validity evidence for personality tests in hiring is much weaker than vendors imply. Job performance correlates with Conscientiousness at around 0.2 across most roles — real, but not strong enough to base a hire on ². Second, personality-based hiring tends to homogenize teams toward whatever profile the existing leadership already has. The "culture fit" problem is mostly this problem.

What to do instead: use personality data as one input into structured interviewing, never as a gate. The signal is too noisy at the individual level. Roberts and colleagues' work on personality as a predictor of life outcomes is honest about how much variance the traits explain and how much they do not ³.

Failure mode three: making the test public inside a team

The third failure mode is the one that feels most generous and does the most damage. A leader runs a workshop, everyone shares their results, and the team puts profile printouts on their desks. Transparency, the thinking goes, builds trust.

What it actually does, more often, is build a permanent shorthand for ignoring each other. "Oh, that is just Sarah being a high-Neuroticism person." "We can't ask Marco that, he is an introvert." The team has stopped seeing individuals and started seeing labels.

Personality is private information. Choosing to share it is reasonable. Being expected to share it as part of work is closer to a soft confidentiality breach. The healthier pattern is one-on-one: a manager and a report comparing notes on what each of them needs to work well, with the test as a vocabulary, not a poster.

Failure mode four: treating change as betrayal

The fourth pattern is subtle and shows up later. Someone took the test three years ago, the team built expectations around the result, and now the person has grown. The team keeps treating them as the old version. The person stops bringing the new version to work.

Roberts and Mroczek's work on adult personality change is unambiguous: people do shift, particularly in their 20s and 30s, and often by enough to matter in a workplace ⁴. A test taken at 26 is not a description of the person at 31.

What to do instead: retest periodically, or just let people self-report changes without making it weird. The test result is a snapshot. Treating it as a permanent ID card is what causes the damage.

Why MBTI in particular keeps causing problems

Most of the cases described above use the MBTI as the testing instrument. There is a reason for that. The MBTI was designed for self-reflection, not for prediction. Its categorical type system implies sharper differences between people than the data supports. Pittenger and others have shown that the four-letter type changes for a non-trivial fraction of test-takers across short periods, which is a problem if you are using the type to make decisions ¹.

The Big Five and HEXACO models are more measurement-honest. They report continuous scores, acknowledge measurement error, and do not push people into discrete tribes. They are also less fun to put on a coffee mug, which is part of why MBTI keeps winning in corporate culture.

If a workplace is going to use personality testing at all, picking a model with continuous scoring and known measurement properties is a small choice with large downstream effects.

What good use of a personality test looks like

A short positive list, since the failure modes do not say what to do.

The best use of a personality test at work is private, voluntary, and one-on-one. A manager and a report use the language to figure out what conditions help the report do their best work. The result does not leave that conversation unless the report wants it to.

The second-best use is self-directed. A person takes the test, reads the result, and uses it to think about their own pattern. Nothing about it has to involve anyone else.

The worst uses are the ones above: sorting, hiring, public broadcast, freezing people in old versions of themselves. All four are common, all four are correctable.

The test is a mirror. Used as a mirror, it helps. Used as a label-maker, it hurts.

Take the Big Five test in private — your results stay yours →

References