You just deleted "Bachelor's degree required" from the JD. Reposted it with the line everyone's using: "We hire for skills, not credentials."

Feels like progress. Feels current.

Then you open the first 30 CVs and do exactly what you always did. Scan the last employer. Count the years. Nod at anyone who worked somewhere you recognize.

You didn't drop the degree filter. You swapped it for one that's just as lazy.

Skills-based hiring doesn't die when you write the JD. It dies when you read the CV.

Why "hiring for skills" mostly lives on the JD and nowhere else

Deleting the degree line takes five seconds. Changing how you judge a human takes a lot more. That's the part almost everyone skips.

You drop one proxy and grab another

A degree is a proxy — a label standing in for the ability you'd rather not measure directly.

The trap is that you replace it with proxies that are just as thin:

• The last company's name.
• Years of experience.
• The right job title.
• "The CV looks strong."

Drop the degree but reject someone because they've never worked at a name you know, and you haven't dropped the proxy. You've just changed the label.

Real skill is hard to score. A proxy is easy to skim

You need one Senior Backend Java engineer. 200 CVs land in three days.

Scoring real skill means reading each project, understanding how they handled concurrency, why they chose that architecture, what they traded away. Fifteen minutes a CV.

Skimming a proxy takes fifteen seconds: has "5 years," has a familiar logo, next.

Deadline's EOD. You pick the fifteen-second move. Everyone picks the fifteen-second move. That's not laziness — it's not having a tool that scores the thing that matters as fast as the thing that doesn't.

Nobody defines "skill" before the CVs arrive

Here's the real gap. "Hire for skills" sounds great, but which skills? What evidence counts as a pass?

If you can't answer those two questions before applications land, then at screening time your brain falls back to the easiest thing to grab: the label. The proxy isn't a villain you chose. It's what rushes in to fill the blank when you never defined what you were actually looking for.

The "Proxy → Proof" framework: score the skill, not the label

A three-step routine you can run today with nothing but a spreadsheet. For every line in the JD, run it through three passes:

Step 1 — Strip the proxy

Look at each requirement and ask: is this a skill, or just a label?

"CS degree" — label. "5 years of experience" — label. "From a top-tier company" — label.

Cross them out. Not because they're meaningless, but because they're pretending to be skills.

Step 2 — Name the real skill

For every label you crossed out, ask what it was standing in for.

"5 years of experience" is usually shorthand for "can design a service that holds under load and keep it alive when it breaks." That's the actual skill. Write it in terms of behavior, not time served.

• Label: 5 years of Java.
• Real skill: owns a production service end to end, from design to the 2 a.m. incident.

Step 3 — Define the proof

For each real skill, ask what in a CV would actually demonstrate it.

A project that spells out an architecture trade-off. A production incident they can narrate start to finish. An open-source contribution with code you can read.

Now you have a checklist scored on evidence, not on labels. The self-taught candidate with no degree but all three pieces of proof — passes. The "perfect on paper" candidate whose paper is empty of evidence — doesn't.

That's skills-based hiring. Not the line you removed from the JD. The thing you're willing to read closely in the CV.

The Knoot Angle

Dropping the degree is a statement. Scoring for proof is the work. And that work doesn't scale on human eyes when 200 CVs land in three days.

This is where Knoot's AI Screening comes in. You never screen blind: you feed it the skill criteria you defined above, and it compares every candidate against those exact criteria. It doesn't see "5 years" and wave them through — it hunts for evidence of each skill, ranks by how well it matches, and explains why. Where a candidate claims a skill the CV never backs up, it raises a flag for you to look closer.

AI flags. AI doesn't reject. You're still the one who decides — the difference is you're now deciding on proof, not on the label that happened to be easiest to skim.

Dropping the degree is what you announce. Scoring for proof is what you prove.