Why a 92% AI Match Score Can Be Meaningless Without Context

"92% match" sounds precise. It isn't. A percentage on its own answers zero useful questions: matched on what? Missing what? Is the 8% gap a dealbreaker skill or something trivial? Without an answer, the number is a number you have to trust blindly.

The black-box problem

Most AI resume scoring tools run a resume and a job description through a model and hand back a single value. The model might be doing something reasonable internally — or it might be over-weighting a resume's job titles and under-weighting the actual skill overlap. From the outside, you can't tell the difference, and neither can the candidate you rejected.

It's also a compliance problem, not just a UX one

Jurisdictions increasingly require employers to be able to explain automated hiring decisions — not just claim they were fair. "The algorithm said 92%" is not an explanation a regulator, or a rejected candidate, will accept. If you can't show your work, you can't defend the decision.

What a score should come with

• Matched skills — explicitly stated in the resume and the job description
• Missing skills — required by the job description, absent from the resume
• Implied skills — not stated outright, but reasonably inferred from experience (and the evidence sentence that justifies the inference)
• The actual sentence from the resume that backs each of the above, not just a label

With that breakdown, a 63% match on a candidate missing one specific certification is obviously a different situation than a 63% match on a candidate missing half the required skill set — even though the number looks identical. The percentage becomes a summary of something you can inspect, not a verdict you have to accept on faith.