How recruiters use ATS scoring (and why it differs from yours)

Most candidates picture the ATS as a binary gate. It's not. It's a sorting layer that hands the recruiter a ranked list. The recruiter then re-sorts it manually about 40% of the time. The mental model matters because it tells you which signals are worth fighting for.

What the ATS scores you on

Most modern ATSes compute a per-candidate match score, exposed to the recruiter as a percentage or a bucket (Strong / Moderate / Weak). The score blends:

• Keyword overlap between résumé and JD (TF-IDF or embedding-based, depending on tenant).
• Years of experience parsed from work history vs. JD requirement.
• Education level vs. JD requirement.
• Location radius from job location.
• Authorisation / sponsorship answers.

The first one — keyword overlap — is the dominant signal in most tenants. It's why JD-tailored résumés outperform generic ones by such large margins.

What the recruiter does with the score

Two patterns I've heard repeatedly from recruiters:

1. Sort by score, scan top 50. The recruiter pulls the top 50 by ATS score and reads each résumé for ~6 seconds. The score got you onto the list; the résumé itself decides the next step.
2. Manual override on bucket boundary. A candidate with a 78% score (Moderate) and a strong cover letter or referral often gets pulled up over an 89% score (Strong) with no narrative. The override is most common at bucket boundaries.

This is why a 92% score isn't always better than an 80% with a strong cover line. The cover line operates on the human, not the machine.

What you can change

Three things move the score the most:

1. The skills section.Listed skills are the densest keyword target. If the JD lists 12 skills, your résumé should reference 8-10 of them by name. The phrasing should match the JD's phrasing — “PostgreSQL” not “Postgres” if the JD writes it out.
2. The first three bullets of your most-recent role. ATSes weight position-relevance — bullets near the top of the most-recent role count for more than bullets at the bottom of older roles. Front-load.
3. The headline.The role title in your headline (or summary, if you have one) directly matches the JD's title. “Senior Software Engineer” specifically; not “Software Engineer / Tech Lead / Architect.”

What doesn't move the score (and isn't worth optimising)

• Format choices (single column vs two column, fonts). The ATS strips formatting before scoring.
• Hobbies, volunteer work, extracurriculars unless they're directly relevant.
• The objective / summary statement, except as a keyword surface.
• GPAs from school more than 5 years ago.

The recruiter override is the lever

If the ATS gives you 80% and the human gives you the time of day, you're competitive. The cover letter, the referral, the cold outreach message — these are all interventions on the human, not the machine. They're why you can't game the ATS to a 99% score and call it a day.

Practical: tailor the résumé to clear the ATS gate (75%+ score), then put the bulk of your effort into the narrative that moves the recruiter. Going from 75% to 95% on the ATS score is mostly diminishing returns; the recruiter doesn't care once you're in the top pool.

What if the company doesn't use a scoring ATS?

Smaller companies (under 200 employees) often use Lever or Ashby with no scoring layer enabled. The recruiter sees applications in submission order. In that case, the early bird wins — apply within 24 hours of the posting going up, ideally in the first 6 hours.

How to spot a no-score ATS: check if the company posts on Y Combinator's Work at a Startup, Wellfound, or their own careers page directly. Those routes typically don't score.