The junior and mid rungs are merging

The statistic everyone quotes

There are several versions and they don't agree. TalentWorks' widely cited 2018 analysis found 61% of full-time postings labeled entry-level required three or more years of experience. LinkedIn's Economic Graph team put it at 35% in 2021, using a broader definition that counted postings asking for any years of prior experience. Zippia's 2026 figure is 38.4% requiring three-plus years. Indeed Hiring Lab has reported that in tech, the share of postings seeking five or more years rose from 37% to 42% between mid-2022 and mid-2025.

These are analyses of posting text with different samples, different definitions of entry-level and different thresholds. They agree on direction. They don't agree on size, and any one of them quoted as the number is really quoting a methodology.

They share another weakness. They measure what a posting says, and a posting doesn't tell you what a hiring manager will accept. A stated requirement can be a hard screen or a wish. The better evidence comes from somewhere else.

How the seniority mix is shifting

Indeed changed how it classifies seniority, and the change matters. Instead of reading titles or stated years, Indeed now classifies postings by the responsibilities and training the role involves. Under that method, as of early 2026, roughly 46% of U.S. postings were entry-level, 40% mid-level and 14% senior.

The movement inside that mix is the story. Entry-level postings had been falling since their 2022 peak and were down 7.5% year over year as of May 2026. Senior postings were up 14.7% over the same period. In software development, senior roles were 69.3% of postings in the first quarter of 2026, and when software postings rebounded between May 2025 and May 2026, 71% of the net increase came from senior roles.

Indeed is careful to call this correlational and to note that entry-level still dominates in absolute terms. The direction holds across their data, with weight moving to the top of the ladder and off the bottom, and it's measured by what the job involves rather than by what the posting calls itself.

Force one: fewer junior roles

The demand-side explanation has a name now. A 2025 paper by Harvard economists Seyed Mahdi Hosseini and Guy Lichtinger, using resume and posting data covering tens of millions of workers, found that at firms adopting generative AI, junior employment fell relative to non-adopting firms while senior employment at the same firms kept growing. They call it seniority-biased technological change. Indeed's software analysis uses the same phrase.

The mechanism is plausible. The tasks that used to define a first job, the routine and well-specified ones, are the tasks most easily absorbed by AI tools or by a more productive senior colleague using them. What's left of junior work tends to need judgment: knowing which output to trust, what to escalate, when the standard approach doesn't apply. Employers have no clean way to screen for judgment in someone with no track record, so they use the one proxy a resume offers. Years.

Three years is a stand-in. The work that remains needs something that three years tends to produce.

Force two: experienced applicants competing down

The supply side gets less attention and may matter as much. Indeed Hiring Lab reported in July 2026 that in May, 62% of applications started on Indeed by job seekers with six to nine years of experience were for entry-level roles. Only 6% were for senior roles. Nearly half of applications from workers with ten or more years of experience also went to entry-level postings, and just 12% went to senior ones.

Consider what that does to a hiring manager. A posting labeled entry-level draws applications from candidates with six years in the field, and some of them are excellent. The manager's picture of an entry-level candidate resets upward. The next posting's requirements follow the pool the last one attracted. The stated requirement gets negotiated, implicitly, with whoever applied last time.

Students can't see this part. From outside, three years required looks like an arbitrary bar. From inside, it often describes a pool in which three years is ordinary.

Why the two forces compound

Each one feeds the other. Fewer junior roles push experienced workers into competing for the roles that remain. Their presence raises the bar on those roles. The higher bar screens out true juniors, who then have fewer places to gain the experience the bar demands. What's left is a single merged rung, early-career but experienced, where there used to be two.

The label survives because it's useful. Entry-level signals a pay band and a level in the org chart. The population it describes has changed. It increasingly means junior compensation for work that assumes mid-level judgment, filled by people who've already done a version of it somewhere else.

This is the mechanism under most of the numbers a career center gets asked about. Placement rates for new graduates, underemployment among recent grads, the headlines about AI killing entry-level jobs, the years-required paradox itself. Each is real and each sits downstream of the merge. AI is one input to the demand-side force. Treating it as the whole story leads to the wrong response.

What this means for career services

Read the years line for what it stands in for. When a posting says three years, it usually means the employer needs someone who can be trusted with ambiguity. Advise on that rather than on the number. A capstone with real stakes, a work-study role with real responsibility, an internship where the student owned something that could fail. Those are evidence of judgment, and judgment is what the employer is actually screening for.

The internship now competes against work rather than against nothing. For years the pitch was that an internship puts a student ahead of peers with no experience. On a merged rung, the student's internship gets weighed against another applicant's four years of employment. That raises the bar on what an internship has to be. A summer of shadowing doesn't clear it. A paid role with ownership might.

Your listing standards are policy. What you allow to be labeled entry-level on your platform shapes what students apply to and how they read their chances. A posting that says entry-level and demands three years is either mislabeled or sending a signal about the pool. Either way you can ask the employer to clarify, and most will, because the mismatch costs them applicants they'd want.

Coach students to tell screens from artifacts. Some three-years lines are hard filters in an applicant tracking system. Others are expectation-setting language a strong application can get past. The difference usually shows in how the rest of the posting is written and in whether the employer runs a structured early-career program. Teaching students to tell them apart is worth more than telling them to apply anyway.

Leadership needs to hear about the merged rung too. Placement conversations still assume a distinct junior rung that graduates step onto. Increasingly there isn't one. The honest framing is that graduates are competing on a rung that has absorbed the one above it, and outcomes should be read against that market rather than the one that existed in 2019.

What would change this reading

If Indeed's entry-level share started rising again in task-classified data, the demand-side force would be weakening. If the share of experienced workers' applications going to entry-level roles fell substantially, so would the supply-side force. If the seniority-biased finding failed to replicate outside the firms and period the Harvard paper studied, the AI component would need to be scaled back. All three are worth watching. None has happened yet.