The personality test is not an assessment

Somewhere in most hiring pipelines there's a step where the candidate answers sixty forced-choice questions and comes back as a four-letter type or a colored quadrant. It produces a clean, shareable artifact. It feels like the objective part of the process. It's the part we'd cut first.

Where these instruments came from

The type indicators trace back to mid-century frameworks built for self-understanding and counseling, not personnel selection. They were designed to help people describe themselves to themselves. Nothing about that origin story is disqualifying, and plenty of good tools got repurposed, but it matters that selection was never the design goal, because selection is a much harder problem than description.

Two things go wrong when you repurpose them anyway.

Type sorting throws away the data. Human traits are continuous. Sorting people into buckets means someone one point either side of a cutoff lands in a different category and gets a different narrative attached to them. Retest a few weeks later, in a different mood, and a meaningful share of people come back a different type. An instrument that isn't stable across a month can't be telling you something durable about a career.

Self-report is the wrong sensor in a selection context. These questionnaires ask people to describe their own tendencies. That's plausible when there's nothing at stake. In a hiring process there is something enormous at stake, and every candidate can see which answer the job wants. You are not measuring the trait. You are measuring the candidate's read on the role, which is a real skill but not the one on the scorecard.

The Barnum problem

The reports land because they're written to land. "You value both collaboration and independent focus." "You are decisive but consider input." Nearly everyone reads their profile and recognizes themselves, which is evidence of good copywriting, not good measurement. A test where every result feels accurate to its subject has no discriminating power, and discriminating power is the entire job.

The part that should worry you more

A personality profile invites a manager to build a story about a person before that person has produced anything. Once the story exists, ambiguous behavior gets read to fit it. The quiet candidate becomes "low drive" instead of "thinking." The direct one becomes "abrasive" instead of "clear." You've handed the decision a lens, and the lens was calibrated by a questionnaire.

The same tools also drift, quietly, into proxies for background, language, and culture: categories that have nothing to do with whether someone can do the work, and that a self-report instrument has no way to hold constant.

What we use instead

A work sample, scored blind against a rubric written before submissions arrive. This is the least fashionable answer in hiring and the most durable one: if you want to know whether someone can do the job, watch them do a piece of the job.

Day One™ produces an artifact you can argue with. Not "high conscientiousness," but an actual deliverable, with an actual score, against criteria you can read. When a manager disagrees with the score, they can point at the work and say why, and the rubric gets sharper. There's no version of that conversation with a four-letter type.

Where preferences do belong

Everywhere except the decision.

We ask for POP, Personal Operating Preferences, at step three of the application, before anyone has done a Day One™. How someone works best. When they do their sharpest thinking. How they want feedback delivered. It's the same category of information a personality test claims to capture. The difference is what happens to it: POP is never scored, never ranked, and never part of a pass or fail. It exists to set someone up to win and to match them with a client whose operating style won't grind against theirs.

Preferences are for onboarding. Proof is for selection. Almost all of the damage comes from swapping those two.