Match score
A profile's score against the sourcing criteria, the coverage, the threshold and how each one is calculated.
In Spanish it's also Match score. With it come coverage (cobertura) and the match score threshold (umbral de match score).
What it is
The Match score is a percentage that says how well a profile meets the job post's sourcing criteria. It answers a single question: is it worth spending time on them?
Each criterion gives one of four results:
| Result | What it means |
|---|---|
| Good fit | The profile has the data and meets it |
| Fit | You can infer from the profile that they meet it |
| Bad fit | The profile says they don't meet it |
| No data | The profile doesn't say enough to tell |
Next to the score goes coverage: how much of your criteria's weight had data in the profile.
What it's for
Ranking. Among hundreds of profiles, the score tells you who to start with, and coverage, how much to trust that number. It also decides who gets into the longlist.
Where you see it
| Where | What it shows |
|---|---|
| The longlist, in Sourcing | Each profile's score and coverage, and its result on each criterion |
| The candidate's Compatibility tab, in the ATS | The score, the coverage, each criterion with its result and the AI summary |
If you change the criteria after scoring, candidates' scores are marked stale until they're calculated again.
What it gets mixed up with
| Mixed up with | The difference |
|---|---|
| The scorecard | The Match score measures the profile, before talking to the candidate. The scorecard, from 1 to 5, measures what you saw in the interview. See Scorecard |
| Performance | It belongs to a search, not a profile: how much the profiles it brings are worth. See Ideal profile and searches |
| The ideal profile | It doesn't score. Only the criteria count |
Why it works this way
No data doesn't subtract. A profile not saying something isn't the same as not meeting it: punishing it would discard everyone who didn't write their profile well. That's why the criterion leaves the calculation, and what drops is coverage.
Coverage is separate from the score so a high number with little data doesn't mislead. A 100% on a single criterion says less than a 77% on all of them. That's why the longlist is sorted by both together, and with less than 30% coverage a profile doesn't qualify.
No data and Bad fit are two things. Bad fit is a problem with the candidate; No data, with a criterion the profile doesn't show. If a criterion stays as No data for almost everyone, it's worth asking it in screening.
The threshold starts at 50% because that's exactly the score of someone who's a Fit on everything: whoever can at least be inferred to meet every criterion gets in.
How it's calculated
Each result is worth points: Good fit, 2. Fit, 1. Bad fit, 0. No data doesn't count.
Match score = the sum of weight × points, divided by the sum of weight × 2, over the criteria that had data.
Coverage = the weight of the criteria with data, over the weight of all of them.
Each criterion's weight comes from its position. See Criteria.
Example with 5 criteria, weighing 30, 25, 20, 15 and 10:
| Criterion | Weight | Result | Points |
|---|---|---|---|
| 1 | 30 | Good fit | 2 |
| 2 | 25 | Fit | 1 |
| 3 | 20 | No data | doesn't count |
| 4 | 15 | Bad fit | 0 |
| 5 | 10 | Good fit | 2 |
- Match score = (30 × 2 + 25 × 1 + 15 × 0 + 10 × 2) ÷ (80 × 2) = 105 ÷ 160 = 65.6%
- Coverage = 80 ÷ 100 = 80%
With that score and coverage, the profile qualifies: it's above the 50% threshold and the 30% minimum coverage.
| If | It ends up |
|---|---|
| It fails a must-have criterion, with Bad fit or No data | Disqualified, whatever its score |
| It has less than 30% coverage | Below threshold |
| Its score reaches the threshold | Qualified |
| Its score doesn't | Below threshold |
The threshold is changed in Sourcing, between 0 and 100. The longlist is sorted by score × coverage.