Most "AI job match" tools hand you one number and expect you to trust it. We don't. Every match in JobJab is broken into five weighted dimensions that sum to 100, and the score on each one is shown in your daily report — not buried.
This post explains the rubric, why we picked these five, and how to use the breakdown to decide which jobs are actually worth applying to.
The rubric
Five dimensions, fixed weights:
- Title alignment — 30 points
- Seniority match — 25 points
- Experience focus — 20 points
- Industry relevance — 15 points
- Location desirability — 10 points
Total: 100. Each dimension is scored 0–max independently against your profile and the posting, then summed.
Why title carries the most weight
Title mismatch is the single biggest cause of bad matches. A "Senior Product Manager" who applies to "Product Manager" is leveling down; one who applies to "Director of Product" is leveling up. Both are legitimate, but they're different decisions — and most job boards treat them as the same match.
By weighting title at 30 points, we let the rest of the rubric narrow the answer rather than override the title signal. A perfect-on-paper experience match for a wrong-title role still tops out around 70/100. That's intentional.
What "75+ = strong" actually means
The tier mapping is:
- Strong — 90 and above. Apply today.
- Good — 75 to 89. Apply this week.
- Stretch — under 75. Read the description; the model thinks you're outside the band.
These thresholds are calibrated against the rubric, not eyeballed. With dimension scores spread across 60–95 (Claude is told to reserve 90+ for "literally nothing to deduct"), a Strong match means the candidate is at or above expectation on every dimension and clearly above on title or seniority.
Reading your breakdown
When you click into a match, you'll see the five-bar chart. The colors aren't decorative — they're tier-mapped:
- Green: ≥75% of the dimension's max
- Cyan: ≥50%
- Amber: below 50%
If your overall score is 78 but your Location bar is amber, you can decide whether the location is a real blocker before you spend a Saturday tailoring your resume. That's the entire point of breaking the score apart.
What we deliberately do *not* score
- Salary fit. Different signal, different decision. Salary is shown but not factored.
- Company prestige. The model doesn't know which logos look good on a resume; you do.
- Recency of posting. Older postings aren't penalized — sometimes the right job sits open for weeks.
When the rubric disagrees with you
If you keep seeing 90+ scores for jobs you don't want, the model is reading your profile wrong. The fix is upstream: open Profile, re-read what's in How JobJab sees you, and either edit your résumé or refine your target titles. The summary feeds every score.
If you're seeing 60s on jobs you think are perfect, the model is right and you're optimistic. Read the dimension that's pulling the score down and ask whether the posting actually matches your seniority or industry — those two together account for 40 of the 100 points.
The one design choice we're least sure about
Location at 10 points. We've thought about pulling it out entirely (the hard-filter already removes geographically wrong matches) and we've thought about pushing it to 15. For now it's a small tiebreaker that distinguishes good locations from acceptable ones, but it's the dimension we're most likely to revisit.
If you have strong opinions, the founder's email is in the footer.