Skills With Evidence: Why 'Python on a CV' Tells You Nothing

Two candidates apply for the same role. Both list Python on their CV.

One used Python for four years at BP, building REST APIs for an internal platform. The other completed a Python course last month and added it to their skills section.

Same word. Different candidates. If you match on the word, you can't tell them apart. If you read for evidence, the difference is obvious in thirty seconds — and it's the thirty seconds that decide whether your client interviews a developer or a hopeful.

Why claims seduce us

A CV's skills section is the candidate's marketing department. It's written to be scanned, to tick boxes, to survive a keyword search. Every word in it is a claim: "I can do this."

There's nothing dishonest about claims. A candidate who's used Python once, on a course, in a lab, isn't lying when they list it — they just mean something different by it than your client does.

The problem is what happens in your head when you scan: a listed skill feels like a verified skill. Your brain does the keyword tool's job — word seen, box ticked, next CV. The evidence check takes seconds. The shortcut costs interviews.

The evidence trail

Every real skill leaves a trail across a CV: an employer, a date range, a description of work, and — in the best cases — an outcome. Reading for evidence means following that trail for the three or four skills the role actually depends on.

Four checkpoints. For each one, weak and strong answers look like this:

The grading isn't binary. "Adequate" is often fine — for a nice-to-have skill. The question is whether the trail matches the weight the role places on the skill.

The red flags — instant detectors

Some CVs can be graded in ten seconds with these flags:

• The summary-only skill. Listed in the skills section, never appears inside a role. Claim, not evidence. Detector: search the skill word in the job entries — if it isn't there, it isn't there.
• The outcome-less bullet. "Responsible for the API platform." Responsible for what happened to it? No outcome, no evidence of effect. Detector: look for the verb — led, built, migrated, cut, shipped — followed by a result.
• The one-off mention. "Used Kubernetes" appears once, in a project from 2019, never again. Real, but thin. Detector: count the mentions across the career.
• The decade-old skill. Python everywhere until 2017, nothing since. Skills decay. Detector: check the most recent role for each critical skill.

The CV audit, worked end-to-end

Our running example: Senior Cloud Security Engineer, energy-sector client. The ad's three non-negotiables: Azure security, SIEM experience, regulated-environment experience. One CV, run through the trail:

Azure security — listed prominently. Where? A fintech, 2021–2024. How long? Three years. Scale? "Managed Azure infrastructure across 30+ services." Outcome? "Cut infrastructure cost 38% during migration." Every checkpoint hit, at a recognisable name. Strong — shortlist on this alone.

SIEM — listed in the skills section. Where? Nowhere in any role description. The word appears once, in the summary. Claim without evidence. The keyword matcher's favourite. This CV would have ranked highly on a "SIEM" search — and the client would have found out in the interview.

Regulated environment — not listed at all. But read the roles: "4 years at a UK energy trading firm." The candidate never wrote "regulated" — they lived it. Evidence without the keyword. This is the inverse case: the trail exists, the label doesn't.

The audit takes two minutes. The result is a shortlist decision the recruiter can defend: "Strong on Azure, genuinely regulated-sector, but no SIEM evidence — worth an interview to probe it, or pass depending on how hard a requirement SIEM is."

That last clause is the point. Evidence reading isn't about accepting or rejecting. It's about knowing exactly what you're shortlisting on.

Set the bar before you read

Not every skill needs a full trail. The bar comes from the job ad, not from habit:

• Non-negotiable skills need the full trail — where, how long, scale, outcome. Weak trail means probe in the interview or pass.
• Nice-to-haves need any trail at all. One real mention is enough to tick them.
• Skills that don't matter to this role don't get audited. You're not grading the CV, you're testing it against this role.

Ten minutes before you touch the CV pile: write down the three skills that decide this role, and what each one's trail has to show. Then read. The bar keeps you fast and keeps you honest.

Monday morning

• Take your last five shortlisted candidates. Mark every skill in their summary as claim or evidence — regrade anyone you got wrong
• For your current role, write down the three deciding skills and the trail each one must show before you read another CV
• Find one candidate you advanced on a summary-only skill. Reassess before the interview

---

Claim-checking takes seconds per CV and is the highest-leverage habit in screening. The recruiters who do it aren't working harder than the ones who don't — they're just spending the same minutes on the decision that matters: is this real experience? Because that question, answered honestly, is what separates a shortlist from a pile.