Fit Scoring Methodology

How we calculate fit

Every time we draft outreach for you to a company, we compute a fit score between 0 and 100. The score has four axes:

• Skills— overlap between your skills/experience and the role's requirements.
• Compensation— alignment between your stated comp floor and the company's published comp band, when both are available.
• Stage— match between your stated stage preference (e.g. seed, series B, public) and the company's funding / maturity stage.
• Work mode — alignment of remote / hybrid / on-site preference and geography (timezone, locations).

Each axis is scored 0–100, or "not enough signal" when neither side has data. The overall score weights skills (0.45), work mode (0.25), compensation (0.15), and stage (0.15); when an axis is missing, the remaining weights renormalize so the active axes sum to 1.0.

Labels:75+ Strong · 55–74 Good · 35–54 Weak · <35 Poor.

What model

We use Claude Sonnet 4.6(Anthropic, PBC — United States), accessed via our LiteLLM proxy. The scorer reads your positioning narrative, structured resume, skills, work mode preference, and (if you've provided them) your "why leaving / why next" notes and comp floor. It reads the company's name, domain, industry, size, description, locations, stage, and comp band when known.

What we do with it

The score appears as a badge on each company in your dashboard. It does not affect which companies we surface to you — those are filtered by your hard criteria (industry, role, work mode, geography). The score is a triage aid, not a gate.

Rescoring

If a score looks wrong, click Request rescoring on the company's detail page to trigger a fresh evaluation against your current profile and the company's current signal.

Regulatory posture

This score is a covered "Automated Employment Decision Tool" under NYC Local Law 144 to the extent it influences your decisions. It does not substantially assist or replace hiring decisions made by employers; outreach is sent from your own Gmail account and employers respond to you, not to us. We provide this transparency surface and the rescoring tool as standard practice regardless of jurisdiction.

Other things to know

• Where your data goes: the inputs above are sent to Anthropic in the United States each time a score is computed. Anthropic does not retain them for training (Anthropic Commercial Terms). Transfers from the EEA / UK / Switzerland are made under the Standard Contractual Clauses. Full sub-processor list and retention terms in the Privacy Policy.
• How long we keep it: the score, axes, strengths, gaps, and rationale are kept against the (you, company) pairing until you request a rescore (the prior is overwritten), delete your account, or remove the pairing.
• Bias audit: we plan an independent bias audit at least annually starting Q4 2026; the public summary will appear on this page.
• Contact: hiresling@42labs.io · 42labs OÜ (Estonia, EU).