Methodology

How job verdicts are scored.

Every verdict is Apply, Stretch or Skip, with a fit score from 0 to 100. Here's exactly how we get there.

Last updated 22 October 2026

Five dimensions

An AI model scores the job against your profile on five dimensions, each from 0 to 100. The fit score is their weighted average.

Skills & experience

35%

How well your skills and experience cover what the job needs. Must-haves count far more than nice-to-haves.

Seniority

15%

Whether the seniority matches: years, scope and title. Too junior and too senior both lower it.

Domain

15%

How familiar you are with the industry, the product and the kind of problem.

Logistics

15%

Location, remote, hybrid or on-site, employment type, visa and languages, against your preferences.

Pay & growth

20%

Pay against your floor when it's listed, room to grow, and how well the role fits your target roles.

The model works to fixed anchors: 90 and above is an exceptional fit, 75 strong, 60 a workable stretch, 40 weak and under 25 poor. Every score comes with its reason.

Requirements and evidence

Each requirement in the posting is matched to your profile as met, partial or missing. “Met” needs evidence: the facts from your profile that prove it, which the verdict shows you. Must-have coverage counts a met requirement as one and a partial one as half.

The decision

  • Apply: a fit score of 75 or more, with at least 70% of the must-haves covered.
  • Stretch: a fit score of 55 or more.
  • Skip: anything below that.

Two rules override the score. Any deal-breaker makes it Skip. And when the ghost-job risk is high, an Apply becomes a Stretch, because the role may not be open.

Deal-breakers

These come from the preferences you set, and each one rules a job out:

  • An on-site job when you don't want on-site work, or a hybrid one when you only want remote.
  • A location you can't work from, once you've set where you can work.
  • Pay where the top of the posted range is below your floor (in the same currency, compared as a yearly figure).
  • No visa sponsorship when you need it.
  • A required language that isn't in your profile.
  • A company or industry on your avoid list.
  • An employment type you don't want, such as contract when you want full-time.

These are flagged for you to check, but never rule a job out on their own:

  • The posting doesn't say whether it's remote, hybrid or on-site, or the location is unclear.
  • Pay isn't listed, or it's in a different currency from your floor. We don't guess exchange rates.
  • Visa sponsorship isn't mentioned and you need it.
  • A required language at a higher level than your profile shows.
  • The role looks more senior or more junior than you, or far from the level you're targeting.

Limits

The dimension scores are an AI model's judgement, not a measurement, and it can misread an unusual posting. That's why each one shows its reasoning and each met requirement shows its evidence. Every verdict asks whether you agree, and those answers help us find where it goes wrong. The thresholds and deal-breakers above are fixed rules, so they apply the same way every time.