A match score is only useful if you can see what produced it. Ours is built from seven factors, each scored separately, each with a written reason, and each visible on the role itself.
The seven factors
- Skills — how much of what the role asks for you can already evidence, including skills backed by your public repositories.
- Experience — seniority and years against what the posting actually requires, not against its job title.
- Education — relevant where a role or a visa route genuinely requires it, and weighted low where it does not.
- Eligibility — whether you can legally take the role. This one can cap the whole score.
- Location — the role's location against where you are and where you would move.
- Sponsorship — the employer's sponsorship history against whether you need it.
- Preferences — the things you told us you want, which stop a technically perfect match from being a role you would hate.
Why eligibility caps the score
A role you cannot legally take is not a 70% match with one problem. It is not a match. When eligibility is blocked the score is capped rather than averaged, because averaging it would let a strong skills fit hide a hard blocker — which is exactly the failure that wastes a month of someone's search.
When to ignore the score
Two cases. First, when you know something the model does not: a referral, a prior conversation, a side project that is not in your profile yet. Second, when the score is high but the written reasons are thin — a score built mostly from keyword overlap on a vague posting deserves less weight than its number suggests.
The score is there to order your week, not to make your decisions. If the reasons do not convince you, the number should not either.