Leaderboard fallacy
First place is just an opinion in a hat.
A ranked list without per-criterion breakdown forces every decision to be a vote of confidence in the model. There is no debate, only deference.
AI candidate ranking software
AI candidate ranking derived from evidence, not opinion.
Most AI ranking tools hand you a sorted leaderboard backed by a single composite score. ShortlistTable computes review priority from per-criterion AI verdicts you can audit — every candidate’s position is explained by a column-by-column breakdown the recruiter can override in one click.
Review priority, not leaderboard · Recruiter override · No auto-reject
shortlisttable.app / tables / ranking-q2Review priority
| Candidate | Must-haves met | Nice-to-haves | Years senior | Review priority | |
|---|---|---|---|---|---|
Alice Chen Ledger · 7y | 5 / 5 | 3 / 4 | 7 | Call first | 1st |
Tomás Almeida SRE · 9y | 4 / 5 | 2 / 4 | 9 | Review fit | 2nd |
Marco Silva CloudCo · 12y | 3 / 5 | 3 / 4 | 12 | Review fit | 3rd |
Citation · Alice Chen · “Five-year span owning the payments event bus on-call rotation” · resume p.1Source-backed
Review priority
Not a leaderboard
Per-column
Sortable, filterable
0
Auto-rejections
1-click
Override the ranking
Why ranking ≠ judgement
A ranked list without evidence is just sorted opinion.
Sorting candidates is easy. Sorting them in a way you can defend is hard. A good ranking tool tells you not just who is at the top — but why, criterion by criterion, with the evidence to back each judgment.
We use “review priority” language instead of “best candidate” language because the recruiter still makes the call. The engine orders the review work; humans decide the hire.
Auto-reject creep
Cutting at rank #25 is a hidden screen-out.
If recruiters only ever look at the top 25 of a ranked list, the bottom 100 are silently rejected. We avoid this by separating priority from rejection.
Tie-breaking by composite score
The number doesn’t mean what you think.
Composite fit scores combine criteria with hidden weights. Two candidates with the same score can be wildly different fits, and you cannot tell which until you read both resumes.
How priority works
Ranking that comes from transparent inputs.
01 · Feature
Review priority, not score
We bucket candidates into Call first / Review fit / Hold. Each bucket is derived from the per-column verdicts, not a hidden weighting.
Call firstReview fitHold
02 · Feature
Per-criterion explanation
Open any candidate and see exactly which must-haves passed, which need review, and which failed. No mystery composite.
5 / 5
must-haves
verified
03 · Feature
Sort by any column
Sometimes you need to see the top candidates for one specific must-have, not the overall ranking. Every column is sortable.
Sort: must-havesSort: years seniorSort: stack match
04 · Feature
No auto-reject
The hold queue is not a delete queue. Low-priority candidates stay in the workspace with their evidence intact, ready to be revisited.
Hold queueRe-runnableAudit trail
05 · Feature
Recruiter override the ranking
Don’t agree with a priority? Move the candidate up or down in one click — with a note that survives in the audit history.
Engine: Review fit
Override: Call first
06 · Feature
Re-rank as columns evolve
Edit the column definitions and the ranking updates. The audit log keeps the previous ranking so you can compare.
Edit column→Re-run→Compare ranking
How it compares
Candidate ranking vs the alternatives.
| Property | Black-box AI ranker | ATS rank column | Spreadsheet | ShortlistTable |
|---|---|---|---|---|
| Ranking derives from explicit criteria | ✕Composite score | ✕Keyword match | ✓Manual | ✓Per-column verdicts |
| Per-candidate explanation | ✕Score only | ✕Score only | –Manual notes | ✓Column-by-column |
| Sort by any criterion | ✕By score | –Limited fields | ✓Yes | ✓Yes — typed columns |
| Recruiter override the ranking | ✕No | –Limited | ✓Yes | ✓1-click + audit |
| No auto-reject | ✕Common | ✕Common | ✓Manual | ✓Hold queue only |
✓ Supported– Partial / manual✕ Not supported
FAQ
Common questions on candidate ranking.
Why “review priority” instead of “best candidate”?+
Because the engine doesn’t decide — the recruiter does. The engine gathers evidence and orders the work; the recruiter decides who gets hired. We keep the language honest so the workflow stays honest.
Can I sort by a single criterion instead of overall priority?+
Yes. Every column is sortable. Sometimes the right view is “rank by must-have #3 only” — the table supports that with one click.
What happens if I change a column mid-pass?+
The affected cells re-run, the per-candidate verdicts update, and the priority re-derives. The previous ranking is in the audit log so you can compare.
Is there a threshold for hold queue?+
Yes — configurable per workspace. By default, candidates with two or more must-have failures land in the hold queue. The hold queue is never deleted automatically.
Can hiring managers re-rank candidates themselves?+
Yes — invite them as a viewer with override permissions. Their overrides are logged separately from the recruiter’s, so the trail stays clean.
Does the ranking explain itself in the export?+
Yes. CSV exports include the priority column plus the per-criterion verdicts and evidence behind each, so the hiring manager opening the CSV sees the reasoning, not just the order.
Ranking with receipts
A ranking the hiring manager can audit.
Try ShortlistTable on a batch of up to 25 resumes — see the priority bucket and the evidence behind it.