SCORING & BENCHMARKS

75th percentile.
75% correct?

They answer different questions. Explore the difference before using a score to make a hiring decision.

EXPLORE THE IDEAIllustrative, not live norm data
75th

percentile
Above 75% of this hypothetical reference group.

1st percentile99th percentile

75 out of 100 dots are highlighted to explain relative position. This does not mean 75% of questions were answered correctly.

BEFORE YOU TRUST A PERCENTILE

Ask for its passport.

A benchmark should arrive with enough information to understand where it came from.

Reference population
Who is being compared?
Sample size
How much data supports the norm?
Test & norm version
Which assessment and scoring rules?
Limitations
Where should this comparison not be used?
HOW OUR BENCHMARKS WORK

Every percentile shows its passport.

HireValid benchmarks come from real candidates who took the same test version. Each percentile is labeled with its group and sample size, provisional until 200 candidates, and recorded with its benchmark version. The explorer above shows how percentiles work.

How we build our tests

Scoring methodology and responsible interpretation

What does a percentile mean?

A percentile describes relative position within a reference group. For example, a candidate at the 75th percentile scored above 75% of that documented group. It does not mean they answered 75% of questions correctly or have a 75% chance of succeeding in the job.

Compare results only when test versions, scoring rules and reference groups are appropriate. A score without its norm group can suggest more precision than the evidence supports.

Which reference group is used?

HireValid builds its benchmarks from its own candidates, separately for each test version. A test shows percentiles once 50 candidates have finished it, labeled as a provisional benchmark with its sample size until the sample reaches 200, and as a HireValid candidates benchmark with its sample size after that.

Benchmarks are recalculated daily and rebuilt as each sample grows. Every result records the benchmark version it was compared with, so an update never silently changes the meaning of an earlier result.

How are raw results scored?

Scoring uses classical test theory: correct responses, item difficulty and discrimination. A scaled score can summarize test performance. Overall assessment scores may combine tests with explicit weights.

Writing tasks require an explicit rubric and human review of AI-assisted scores. Employers should be able to inspect the rationale rather than receiving an unexplained number.

What makes a responsible benchmark?

Use a reference population that makes sense for the question you are asking. A broad working-population comparison is different from a role-specific applicant comparison. Neither should be treated as a universal pass mark.

Review item performance and group differences over time. Keep norm versions in reports so a later update does not silently change the meaning of an earlier hiring decision.

Start with what the role needs.

Find relevant testsCompare paid plans
READY WHEN YOU ARE

Scores you can explain to a hiring manager.

Every percentile shows its group and sample size. Try it free with three candidates.

Start freeSee pricing Free plan. No credit card.
Free plan. No credit card.Start free