/
Candidate ranking
Candidate ranking and AI scoring
Results don't arrive as a pile of raw scores. Every candidate is scored and ranked against everyone who took the same test, summarized by AI, and compared so the strongest candidate rises to the top.
Scored and ranked automatically
As candidates finish, they're scored and ranked against everyone who took the same test, with medals for top performers. Ties are broken deterministically (by correct count, then who finished first), so the order is consistent and fair. Filter by test, favorite standouts, and hide the noise.
An AI summary for every candidate
Each completed assessment gets a concise AI summary - the score, overall performance, and notable patterns like spending unusually long on a question or a clear weak area. It's the fastest way to triage a stack of candidates before reading the detail.
The strongest candidate, surfaced
Talent Approved compares everyone who took the test and surfaces the single strongest candidate to interview next, with a short rationale weighing the top contenders. It's allowed to recommend nobody when no candidate clearly stands out - so the recommendation stays trustworthy rather than always crowning a winner.
Spot patterns across the whole cohort
Switch to the matrix view to see candidates down the side and questions across the top, each cell marked correct or incorrect. It's the quickest way to spot a question everyone misses (maybe it's ambiguous) or a skill the whole pool is weak on. Hover a cell to see the correct answer.
Rankings you can trust
Fair, consistent, and comparable
- Candidates are ranked only against others who took the same test, so comparisons are like-for-like.
- Tie-breaking is deterministic, so the order never shifts arbitrarily.
- Synthetic "AI Review" validation runs are excluded from human rankings and the recommendation.
- The AI recommendation can decline to pick anyone, keeping it meaningful.
Frequently asked questions
How is the ranking calculated?
Candidates are ranked by their score against everyone who completed the same test. Ties are broken by the raw number of correct answers, then by who finished first, so the order is consistent and fair.
Can the AI choose not to recommend anyone?
Yes. When no candidate clearly stands out, the AI-recommended-candidate feature can recommend nobody, which keeps the recommendation trustworthy.
What's in the AI summary?
A short, plain-language verdict for each completed assessment: the score, overall performance, and notable patterns such as time spent or weak areas - generated from the candidate's answers.