Every role you see on Jobrods comes with a single number: a match score from 0 to 100%. It is the most important thing on the page, so it is worth being precise about what it means and how we calculate it. The short version is that a match score is a weighted blend of four signals — skills, experience, location, and salary — measured against the specific role in front of you. The long version is below.

The four factors

We deliberately kept the model small and legible. Instead of dozens of opaque features, we lean on four that candidates actually care about and can reason about themselves:

  • Skills (40%). We extract the skills implied by your CV and compare them to the skills a role requires. Related skills count partially — if a posting asks for React and you have years of Vue, you are not treated as a zero.
  • Experience (30%). Seniority and years in relevant roles. A mid-level engineer applying to a staff role will see this factor pull the score down, and we tell you so rather than hiding it.
  • Location (20%). Commute distance, remote eligibility, and relocation preferences. A fully remote role you are eligible for scores full marks here.
  • Salary (10%). How the advertised band overlaps with the range you told us you are looking for.

Why these weights

The weights are not arbitrary. We started with equal weighting, then tuned against anonymized feedback on which matches people actually pursued. Skills consistently turned out to be the strongest predictor of a candidate saying "yes, this is relevant," so it carries the most weight. Salary matters but tends to be a filter rather than a ranking signal, so it carries the least.

Explainability is the point

A score on its own is just a number you have to trust blindly. That is why every Jobrods match expands into a breakdown showing how each factor contributed. If a role scores 74%, you can see that it might be 95% on skills and location but only 40% on experience — which tells you exactly what to address in your application, or whether to skip it.

We will never show you a recommendation we cannot explain. If our model cannot justify a score, that is a bug, not a feature.

What is next

We are experimenting with letting you adjust the weights yourself — if salary matters more to you than location right now, you should be able to say so. The principle stays the same: your search, your priorities, fully transparent maths.