Hiring Based on Proof: Beyond the CV

• Hiring based on proof means you judge a candidate on evidence of what they can do, such as work samples, structured tasks and demonstrated skills, instead of the keywords, titles and pedigree printed on a CV. You rank people by capability and fit, and how well their document plays a filter stops mattering. That is evidence-based hiring.

• Keyword screening matches strings. A candidate who has done the exact job well for a decade can be filtered out because they wrote "shipped" where your template wanted "delivered", while a keyword-stuffed CV with none of the skill gets through. Keyword screening is designed to work that way.

Why the CV is weak evidence

• A CV is a claim. The candidate writes it, in language they chose, arranged to look good and to survive filters. It tells you what someone says they did.

• Research on selection methods points the same way. SHRM's guide to assessment methods says [unstructured interviews are not particularly useful for predicting job performance](https://www.shrm.org/content/dam/en/shrm/topics-tools/news/hr-magazine/assessment_methods.pdf). It describes the alternatives: structured interviews with standard rating criteria, and work sample tests that mirror the job's tasks and are scored by trained evaluators.

• Interviews have their own gaps. Harvard Business Review argues that [job interviews often are not evaluating the right skills](https://hbr.org/2025/08/job-interviews-arent-evaluating-the-right-skills). If you rank people on the wrong signal, you get the wrong hire, only faster.

Signals and proof

• Traditional screening relies on proxy signals. Evidence-based hiring looks at the work directly.

• Everything in the left column can be inflated, borrowed or gamed. The right column is much harder to fake.

What proof-based hiring looks like

• You don't need to rebuild your process. You need to move the point where evidence comes in. Most teams screen on pedigree first and only look at capability once a candidate clears the informal filters. Put the evidence first instead.

• A practical evidence-based evaluation runs like this:

1. Define the job. Write down the three or four things the person must be able to do. Leave out "5+ years" and the degree, and list the tasks.
2. Collect evidence against those tasks: a short work sample, a structured exercise, or a portfolio tied to the role. Keep it small enough to fit in real working time.
3. Score everyone the same way, with one rubric. Using the same questions and rating criteria for every candidate is what makes the result reliable ([SHRM assessment methods](https://www.shrm.org/content/dam/en/shrm/topics-tools/news/hr-magazine/assessment_methods.pdf)).
4. Rank people on the evidence. Order the pool by demonstrated fit.
5. Interview to confirm. By the time you talk, you already know they can do the work, so the conversation can test fit and motivation.

• You end up with a shortlist you can defend. Every name on it got there on evidence a hiring manager can see.

Is this slower?

• It feels as if measuring evidence must add work, but the slow, expensive parts of hiring are manual CV screening, rereading, second-guessing proxies, and the mis-hires that pedigree screening lets through. Cut the proxy stage and you cut the slowest stage, so ranking on evidence from the start shortens the process.

• Unframed [builds shortlists in hours](/spaces/build-shortlist-in-hours) by ranking every applicant on evidence of fit.

The cost of getting it wrong

• Every qualified person your filter rejects is a hire a competitor can make. Every unqualified person it lets through is a mis-hire you pay for in ramp time, in the load on the rest of the team, and in the replacement search six months later. Both mistakes come from ranking people on the wrong evidence.

• For the mechanics of why the filter fails, read [why an ATS rejects qualified candidates](/spaces/why-ats-rejects-qualified-candidates) and the difference between [keyword matching and skills ranking](/spaces/keyword-matching-vs-skills-ranking). Both come down to whether you match strings or measure capability.

Hire on what candidates can show you

• Ask whether the candidate can do the job, and have them show you. Define the work, collect the evidence, and score it the same way for everyone. If you want to rank applicants on evidence instead of keywords, see how Unframed works for [companies hiring on proof](/companies).