Skills-Based Hiring in the US in 2026: What Employers Actually Changed

The 53% that mostly didn't change anything

Roughly 70% of US employers now describe their hiring as skills-based, and 53% say they have removed degree requirements from at least some roles. Those are the numbers that make conference slides.

Here's the one that doesn't. Of the companies that dropped degree requirements, around 45% did so in name only: fewer than 1 in 700 hires actually went to a candidate without a degree. The requisition changed. The hiring behaviour didn't.

If you run talent acquisition at a US enterprise, that number is either an indictment or a relief depending on whether anyone has checked yours. My guess is that most organisations reading this have never pulled the report, because pulling it means finding out. It takes about an hour in Workday or Greenhouse to segment last year's hires by highest credential and compare it to the reqs that were supposed to be open to non-degree applicants. Do that before you buy another assessment platform.

Related reading: The US Corporate Training Market in 2026: Budgets, Pricing, and ROI · US Community College Employer Partnerships in 2026: Funding and Workforce Pell · AI Recruiting Tech for US Staffing Firms in 2026.

Stated preference versus revealed preference

The survey data is close to self-contradicting, which is itself the finding. A 2026 Lumina Foundation and Gallup survey found 75% of US employers expect a college degree to be as important or more important to their hiring decisions in five years as it is today. Among employers who formally removed degree requirements, 76% still say they prefer a four-year degree and 78% prefer at least a two-year one. And 79% of entry-level US postings still carry a bachelor's requirement.

So: most employers removed the requirement, kept the preference, and left the screening logic untouched. The req says "degree or equivalent experience". The recruiter screen still treats the degree as the cheap proxy, because it is a cheap proxy, and nobody replaced it with something equally cheap.

That's the actual problem. Skills-based hiring failed in most US organisations not because leaders didn't mean it, but because removing a filter without funding its replacement just moves the filtering somewhere less visible and less accountable.

What replaced the degree, ranked by how much signal it carries

The companies where this worked, and there are real ones (Google, Apple, Netflix, Walmart, Bank of America and Accenture have all pulled degree requirements from meaningful numbers of roles), replaced the credential with something concrete. Not with good intentions.

Notice the top of that table is the expensive end. That's the whole story. A degree filter costs nothing per candidate. Anything that predicts better costs something per candidate. Organisations that budgeted zero for the replacement got zero change in outcomes, and were then surprised.

AI competency is quietly becoming the new filter

More than a third of US entry-level postings now ask for some level of AI competency. Not a data science degree. Familiarity with the tools, some ability to prompt and verify, occasionally a named platform.

There's a real risk here that the sector walks straight into the same trap it just spent five years climbing out of. "AI competency" is undefined, unverified in most screens, and correlates hard with whether someone's last employer paid for a licence. Screen on it casually and you've built a new proxy filter with a fresh coat of paint, one that will disadvantage candidates from smaller employers and public-sector backgrounds for reasons that have nothing to do with capability.

The fix is unglamorous: define what AI competency means for the specific role, in one sentence, and test that. If a customer-operations analyst needs to be able to draft, check and correct an AI-generated summary against a source document, then give candidates a source document and an AI-generated summary with two errors in it. That takes twenty minutes to build and it beats any self-reported proficiency scale you'll find.

The cost of building a pipeline that actually works

Rough US planning numbers for standing up genuine skills-based hiring in a 5,000-person employer, first year:

• Job architecture and skills taxonomy work, typically $80,000 to $250,000 whether you use a consultancy or burn internal analyst time. Nobody escapes this step.
• Assessment tooling, roughly $25,000 to $120,000 per year at that headcount, depending on volume and how many role families you cover.
• Interviewer retraining, the line everyone cuts and shouldn't. Budget two half-days per hiring manager. At 200 managers that's real money and it's the difference between a rubric that gets used and one that gets ignored.
• Validation and adverse-impact analysis, $30,000 to $90,000, plus counsel. In the US this is not optional if you're testing at scale.

There's a fifth line most plans omit: the cost of a longer time-to-hire during the transition. Structured assessment adds steps, and steps add days. For the first two or three quarters your median time-to-fill will get worse before it gets better, and if nobody warned the business units in advance, that's the number they'll use to argue the whole thing was a mistake. Tell them the shape of the curve up front and the political cost drops to almost nothing.

Call it $200,000 to $500,000 in year one for a mid-size employer. Against that, the case is straightforward: a wider funnel in tight markets, faster fills in roles where degree-holders are scarce, and lower early attrition when the screen matches the work. Whether it pays back depends almost entirely on whether hiring managers use it, which is why the retraining line is the one to protect when the budget gets trimmed.

The legal layer US employers keep discovering late

Replacing a degree screen with a test is not a neutral swap in the United States, and this catches organisations out with some regularity. Any selection procedure that produces a substantially different pass rate across protected groups invites scrutiny under the Uniform Guidelines on Employee Selection Procedures, and a degree requirement, for all its faults, has decades of case law and habit around it. A new assessment does not.

What that means practically: validate before you scale, not after. Run the assessment alongside your existing process on a few hundred candidates, compare pass rates by group, and check that scores actually correlate with later performance ratings. If the assessment predicts nothing and screens out disproportionately, you've built a liability and paid a vendor for the privilege.

Several US states and cities have also moved on automated employment decision tools, with New York City's bias-audit rule the best known. If your assessment or ranking layer is doing anything a regulator would call automated decision-making, the audit and notice obligations are real and the vendor's assurance that they "handle compliance" is not a substitute for your own counsel reading the contract.

None of this is a reason to keep the degree filter. It is a reason to budget for validation as part of the programme rather than treating it as an optional extra, which is precisely what the organisations that got this right did and the ones now unwinding their assessment stack did not.

For training providers and EdTech vendors selling into this

The commercial opening in the US market right now is not another course catalogue. It's the verification layer. Employers have been told for five years to hire on skills and have been handed almost nothing that makes a skill as cheap to check as a degree is to check.

If you can hand a US employer a credential that (a) maps to a defined role, (b) reports what the holder can actually do rather than what they sat through, and (c) has placement data behind it, you're selling the thing that's missing. Enrolment numbers won't do it. The buyer question in 2026 is "how many of your completers were hired into the target role within six months, and did they stay twelve?" Vendors who can answer that will win procurement cycles against much bigger names.

Where this lands by 2028

My read is that the degree isn't coming back as a formal requirement and isn't going away as a preference. What changes is the middle. Sectors with dense credential infrastructure already (healthcare, IT, skilled trades, parts of financial services) will keep converting to skills-based screening because the alternative signals exist and are cheap. Sectors without that infrastructure will keep announcing it and keep hiring graduates.

So before you commit to a skills-based strategy, ask a narrower question than the one on the conference agenda: does a verified, low-cost signal exist for the roles I hire most? If yes, this works and you should move. If no, build or buy the signal first, or you'll join the 45% who changed the wording and nothing else.