Know what any HR tech vendor is really selling you — before you sign.

Make the vendor prove it

The sales deck is built to survive your questions. This is built to outlast their answers. The evaluation hands you the exact disclosures a serious vendor can produce and a weak one can't: the validity coefficient against measured performance, on what sample, against what base rate; the auto-rejection rate, and whether anyone ever measured how the rejected candidates performed; every path where the tool rejects a human with no human in the loop.

Ask those questions and one of two things happens. The vendor produces the data — and your evaluation just got stronger. Or the vendor can't — and the silence is the finding. Either way, the burden of proof has moved to the side of the table that should have been carrying it all along.

Judged by the science — not the sales deck

"4× more accurate." "Reduces bias." "AI-powered." Every claim gets checked against the published selection-validity literature instead of repeated back to you. The benchmark isn't the vendor's white paper — it's the same meta-analytic evidence I/O psychologists use to rank selection methods, anchored to Sackett, Zhang, Berry & Lievens (2022).

Compliant everywhere you hire

A hiring tool that's defensible in one country can be a lawsuit in the next. The evaluation reads each vendor against the employment law that actually governs your markets — and tells you where your exposure sits before you sign, not after a complaint lands. It also checks current case law, where courts now treat screening vendors as the employer's agent — so the vendor's risk becomes yours.

This surfaces where the risk is and what to put in front of your counsel — it isn't legal advice or a compliance certificate. The framework's legal logic was independently reviewed by Littler Mendelson. It also flags the gap most audits miss: passing a disparate-impact test is not the same as being legally safe, and neither one measures the strong candidates a tool quietly rejects.

Two ways a tool fails you — most buyer's guides check one

Every evaluation asks whether a tool picks badly — the bad hire you can see. Almost none ask the harder question: whose best candidates is it throwing away before anyone looks? That's the Type II error, and it's invisible by design — rejected candidates leave no performance data, and the vendor's own bias audit won't catch it.

A keyword screen barely better than a coin flip isn't just weakly diagnostic — at a typical cutoff it also misses roughly two-thirds of the strong candidates in the pool. Stack three such filters and the misses compound: career changers, returners, fast learners with fewer years on paper, anyone who doesn't match the historical pattern — screened out before a single interview. Adding screening tools to a résumé-based process barely improves who you pick, and steadily worsens who you never see.

The evaluation makes both error types explicit for every vendor you run — the hire you'd regret, and the hire you'll never know you missed.

One evaluation, three depths

Every conclusion is on the first screen. It opens by stating exactly what was evaluated, then gives you the verdict — and holds up when the vendor pushes back.

Whether you're running the demo or signing the contract, you read only as far as you need.