Resources/Lifestyle/Inside the build: why we didn't write our own personality questions
Lifestyle · product-update

Inside the build: why we didn't write our own personality questions

The tempting version of this product invents its own framework. We used a public-domain instrument with three decades of validation behind it instead.

2 min readUpdated Jul 2026

Every personality product eventually faces the same fork: build a proprietary framework from scratch, or build on top of something that already exists and has been tested. The first option is more marketable — a proprietary model means a distinctive vocabulary, something that can't be directly compared to a competitor's. It's also, for a young product, a bet against several decades of psychometric validation that a scrappier approach doesn't have time to replicate.

We used the IPIP-NEO-120: four items per facet, thirty facets across five domains, drawn entirely from the International Personality Item Pool — a public-domain item bank released specifically so products and researchers wouldn't have to either license a commercial instrument or invent their own. It's not the same as the proprietary NEO-PI-R it was built to approximate, but it's been independently validated against it, and it measures the same underlying structure.

What that decision actually costs and buys

The honest tradeoff: a public-domain instrument means Bearing's assessment isn't structurally distinct from anyone else's implementation of the same item pool. There's no proprietary moat in the questions themselves. What it buys instead is that every one of the 120 questions has a documented history, published psychometric properties, and — because the item pool is public — nothing in the assessment is a black box we're asking people to trust on faith.

The alternative, writing 120 original questions and validating them from scratch, is a multi-year research undertaking on its own, before any product gets built around it. Using an existing, public-domain, well-studied item pool meant the actual product work could focus on what happens after the assessment — the report, the writing, the way results get shown back to someone — rather than re-deriving basic psychometrics that already existed.

Where the real product work went instead

Thirty facets and five domains is the scoring layer, and it's mostly not ours — it's the IPIP-NEO-120 doing what it was built to do. The part that's actually Bearing's own work sits on top of that: how a raw facet score becomes a trait word, how six related facets get shown together instead of as thirty disconnected numbers, when the report suggests going deeper into a specific branch instead of stopping at the surface level. That's where a genuinely different product gets built — not in reinventing the measurement, but in what happens to the measurement once it exists.

Is the IPIP-NEO-120 the same thing as the official NEO-PI-R?
No — the NEO-PI-R is a commercial, copyrighted instrument. The IPIP-NEO-120 is a separate, public-domain item set built to measure the same five-domain, thirty-facet structure, developed independently and validated against it.
Why 120 questions specifically?
Four items per facet, thirty facets, is the standard IPIP-NEO-120 structure — enough per facet for a reasonably reliable score without the length of the full 300-item version.