Years of Experience Doesn't Predict Performance. Here Is What Does.

"5+ years of experience required" is asking for the wrong thing.

We've all written it. I've written it. It feels sensible. It's easy to check on a CV. Candidates know whether they have it or not. Recruiters can use it as a quick filter against a tall applicant stack. By every measure of convenience, "years of experience" is a great hiring criterion.

It just doesn't predict performance.

The most credible meta-analysis we have on this question – sixty years of data, eighty studies, eleven and a half thousand people – puts the corrected correlation between pre-hire work experience and job performance at 0.06. That is, statistically, not different from zero. The correlation with retention is even more damning: 0.00.

We are filtering candidates on a number that has no relationship to whether they will do the job well, or stay.

Why we keep using years anyway

The reason "5+ years" is everywhere is not that anyone has run the numbers and decided it predicts performance. It's that years are cheap to filter on. They're a single integer per candidate. They're already on every CV. They sort applicants in a way nobody will argue with. And – not least – they let a hiring manager feel they've done some screening before the work has actually started.

Easy isn't the same as effective. Life is just a little pesky like that.

Why years stops being useful at "1"

The intuition behind years is that more time on the job means more compounded learning. Sometimes it does. Often it doesn't. The problem is that "years of experience" doesn't distinguish between someone who learned for five years and someone who repeated year one five times.

On a CV, those two candidates look identical. The headline says "5 years in marketing" in both cases. But one of them has driven substantially better outcomes than the other – and the years-shaped filter cannot tell you which.

Worse: when you require "5+ years of experience" you are also actively filtering out the fastest learners. The candidate who achieved in three years what most people take five to deliver is, on paper, indistinguishable from the candidate who achieved nothing of note across either window. Both fail the filter at year three.

What sixty years of data says

The cleanest test we have is Van Iddekinge and colleagues' 2019 meta-analysis in Personnel Psychology: 81 independent samples, 11,577 people, sixty years of workplace studies. The corrected correlation between pre-hire work experience and subsequent job performance is 0.06. The correlation with retention is 0.00. Years of experience is, on the available evidence, a coin flip.

That is not a one-off finding. The same paper reviews six decades of preceding research and finds it consistent across industries, seniority levels, and economic eras. HBR's writeup of the same study puts it bluntly: experience is one of the most popular hiring criteria, and one of the least predictive.

Meanwhile, the screening methods that do predict performance – work samples, structured interviews, cognitive ability tests, and outcomes-based screening – sit considerably further up the validity scale. None of them ask "how many years have you been doing this?" as their primary question.

What to screen for instead

The answer isn't "ignore experience". The answer is to measure it better. Look for the outcomes that someone with the kind of five-year background you're imagining would actually have delivered.

For a marketing role, "5 years in marketing" can become a screen for commercial results: revenue driven, leads generated, campaigns shipped, customer acquisition costs reduced. For a supply chain role, the same translation gets you efficiencies, established processes, savings, and on-time delivery improvements. Same seniority signal. Different filter.

Now you're screening for outcomes, not for time served. And the whizz-kid who took three years to do what most people achieve in five is suddenly very obvious – because the bar is the work, not the calendar.

The skills-based hiring movement has been making this argument for a decade, and the data on adoption is finally catching up. 85% of companies report using some form of skills-based hiring in 2025; 82% report time-to-hire dropping after the shift; and skills-based hirers are about 60% more likely to report a successful hire than experience-based hirers. Job satisfaction follows: skills-based hires report being "very happy" in the role at 38%, versus 28% for experience-based hires.

The obvious push-back

The fairest critique of skills-based hiring is that the rhetoric runs miles ahead of the practice. One recent analysis found that while a clear majority of companies say they're doing skills-based hiring, only about 0.14% of actual hires are affected by degree-requirement removal. The label is widespread. The behaviour change is rare.

That's a real critique – of how skills-based hiring is being implemented, not of whether the underlying logic holds. The Van Iddekinge data on years versus performance is the data on years versus performance regardless of how many companies have actually changed their job descriptions in response. If the adoption gap means anything, it means the teams that do rewrite their filters honestly are competing against opponents who are still, in practice, screening on years.

That's an opportunity, not a refutation.

How to actually do it

Three concrete adjustments do most of the work. None of them require an HR transformation.

1. Translate every "X years of Y" line into outcomes. Go through each of your job descriptions and replace experience-shaped language with outcome-shaped language. "5 years in marketing" becomes something like "demonstrable ownership of growth-stage marketing results – revenue, pipeline, or customer-acquisition outcomes you can speak to in detail". The signal you're after stays the same. The filter stops excluding the right people.
2. Screen the CVs for the numbers, not the dates. When you scan a CV, you're looking for quantified outcomes – dollars, percentages, customer counts, projects shipped. CVs full of titles and dates and zero numbers are, on average, less useful than CVs that name specific results, regardless of which one has more years on it.
3. Ask outcome-shaped interview questions. "Tell me about a specific marketing campaign you led, what the target was, what you delivered, and what you'd change next time" beats "tell me about your background in marketing" by almost any predictive measure. Structured outcome-based questions are among the most predictive instruments in selection psychology.

Where lemonly fits

We built lemonly partly because translating "X years of Y" into "specific outcomes that look like X years of Y" is exactly the kind of pattern-matching that's slow to do by hand on hundreds of CVs and quick to do well with the right tooling. Our candidate screening product evaluates each applicant against the outcomes you've defined for the role, not against a years-of-experience integer that the last sixty years of research says won't predict whether they succeed. The result is a shortlist sorted by what the candidate has actually delivered – which is the bit that matters when they walk in for the interview.

Find the talent you're looking for

lemonly delivers qualified candidates on-demand. Find out how.

Book A Demo

Pick a time for your demo on the next page.

Common objections, answered

Doesn't experience still matter for senior roles?

Seniority signals matter; the question is whether years of experience is the best proxy for them. The Van Iddekinge data covers senior and junior roles alike, and the 0.06 correlation holds across them. What you actually want for senior roles is evidence of having owned outcomes at scale. That's a screen on outcomes, not a screen on tenure.

How do we screen for outcomes at high volume?

This is the implementation problem the skills-based hiring movement runs into in practice. The honest answer is that doing it well at high volume is hard without tooling – which is why the gap between rhetoric and practice exists. Either tighten your sourcing so the volume is lower (which makes manual outcomes-screening tractable), or use automated screening that scores CVs against your defined outcomes rather than against keyword presence.

Won't candidates just learn to write outcome-shaped CVs even if they don't have outcomes?

Some will. Most won't. And the ones who try will fall over at the first outcome-shaped follow-up question in an interview, which is information you couldn't have gotten from a years filter at all. The screen still works because the second stage of the process is doing real evaluation.

What about regulatory roles where years are required by law?

If a regulator says you need a particular certification or a minimum number of years, that's a hard floor and you screen on it. The argument here is about the soft floors that nobody is forcing you to use. "5+ years" on a marketing JD is a choice, not a regulation.

Key takeaways

• The corrected correlation between pre-hire experience and job performance is 0.06; with retention, 0.00. Years of experience is, statistically, a coin flip (Van Iddekinge et al., Personnel Psychology, 2019).
• We use it anyway because it's cheap to filter on, not because it works.
• "5+ years" can't tell apart someone who learned for five years from someone who repeated year one five times. It also actively filters out the fastest learners.
• The fix is to translate every "X years of Y" line into the outcomes someone with that profile would actually have delivered – revenue, growth, deliverables, savings.
• 85% of companies say they do skills-based hiring; 82% see time-to-hire drop, and skills-based hirers are 60% more likely to make successful hires. The gap between rhetoric and practice is large – which is your opportunity.