Why entry-level jobs require experience: what 2,059 job postings show
Notch Research · September 2026 cohort · Related: the first study · method
They mostly don't. Of the 2,059 job postings we collected from company career sites in September 2026, postings labelled entry level almost never asked for more than two years of experience: one did, out of 150 that stated a minimum at all. What the complaint is really about is scarcity. Only 16% of postings were entry level, and in marketing, design, and product the share of postings open to someone with two years or less was 7%, 7%, and 9%. The entry-level job has gone missing from the roles people most want to enter.
The numbers, in one table
Across all 2,059 postings, 74% stated a minimum years-of-experience figure. Of those, 24% accepted two years or less, 22% asked for eight years or more, and the median floor was five years. By role family, of postings that stated a floor:
| Role family | Postings | Accept 2 years or less | Want 8 or more | Median floor (years) |
|---|---|---|---|---|
| Marketing | 108 | 7% | 49% | 7 |
| Design | 63 | 7% | 28% | 6 |
| Product | 87 | 9% | 36% | 6 |
| Software engineering | 257 | 14% | 21% | 5 |
| People and HR | 99 | 15% | 29% | 5 |
| Data and analytics | 134 | 15% | 17% | 4 |
| Finance | 115 | 21% | 25% | 5 |
| Sales | 349 | 29% | 22% | 5 |
| Operations | 419 | 31% | 16% | 3 |
| Customer success | 127 | 33% | 9% | 3 |
Sorted by the entry-level share. Marketing is the striking row: 7% of postings with a floor accept two years or less, 49% want eight or more, and the median floor is seven years. A marketing job opening in this corpus is, in the typical case, a senior job.
By employer type
The same cut by the five panels we collect, with the share of postings that disclose a salary alongside, because the two move together:
| Employer panel | Postings | Accept 2 years or less | Want 8 or more | Post a salary |
|---|---|---|---|---|
| Startups | 339 | 10% | 22% | 12% |
| Tech companies | 851 | 14% | 32% | 71% |
| Consumer brands | 237 | 32% | 18% | 68% |
| Global enterprises | 322 | 38% | 5% | 15% |
| Large US employers | 310 | 47% | 7% | 59% |
Large US employers (hospitals, banks, retailers, utilities, universities, state agencies, 26 in the panel) are the entry-level job market: 47% of their stated floors accept two years or less and 7% want eight or more. Startups are the opposite, 10% and 22%, and they disclose salary least often (12%). Tech companies sit in between on floors and highest on salary disclosure (71%), which is the pay-transparency law showing up, since the tech panel is the most US-based.
Why employers write it this way
Three mechanisms account for most of what the table shows, and none of them is a conspiracy against new graduates.
- Requisitions are copied from the last hire. A posting inherits the years figure from the person who left or the one hired two years ago. The floor describes the previous incumbent, which is why floors cluster at round numbers and why "preferred" and "required" are used loosely.
- "Entry level" is a pay band, not an experience band. The extractor labels seniority from the posting's own language. A posting at the bottom pay grade can still say "2+ years" because the grade below it doesn't exist at that company.
- The junior rung is thinnest where AI demand is highest. The role families with the fewest entry-level postings (marketing, design, product, software engineering) are the ones where the share of postings naming an AI-related competency is highest: 56%, 71%, 59%, and 54% in September. Operations and customer success, with the most entry-level openings, are at 24% and 37%. We report the correlation; we don't know the mechanism, and employers citing AI as the reason are rare in the postings themselves.
What a stated floor means in practice
A years figure in a posting does three different jobs, and you can usually tell which from the wording. Where it sits under "required" or "minimum qualifications," it may back a screening question that filters you out before a recruiter looks; our guides to Greenhouse and Taleo show where those questions appear. Where it sits under "preferred" or inside a paragraph, it's shorthand for the level of judgment the hiring manager expects. Where it appears only in the title ("Senior"), it's a pay grade. Apply to the second and third kinds when the rest of the posting matches you.
How to apply when you're under the floor
- Lead with the two or three requirements you meet outright, in the posting's words. Recruiter searches run on those terms, whatever the years figure says. How to tailor the resume to the posting.
- Translate adjacent experience into the posting's vocabulary rather than apologizing for it. The career-change guide covers the translation.
- Target the role families and employer types where the floors are lowest if you're early in a career: operations, customer success, sales, and large US employers. The table above is the map.
- Check whether the years figure is a knockout question in the application form before spending an hour on it. If the form asks "Do you have 5+ years," answering honestly below the line ends the application; the posting's other requirements are then the ones to pursue elsewhere.
Method
Postings are collected weekly from the public career sites of five employer panels: 39 tech companies' Greenhouse boards; consumer brands; nine multinationals via SmartRecruiters; 14 venture-backed startups via Ashby and Lever; and 26 large US employers via Workday. September 2026: 2,059 postings (tech 851, startups 339, enterprise 322, large US employers 310, consumer 237). Each posting is read by the competency-extraction engine that powers the Notch tool, which records the stated minimum years of experience, the seniority label implied by the posting's language, salary disclosure, degree requirement, named AI tools, and work arrangement. "Entry level" in this piece means the extractor's seniority label; "accept two years or less" means a stated minimum of 0, 1, or 2 years among postings that stated one. Panels are never blended into one number because their composition differs by design. Role families with fewer than 60 postings are omitted from the table. Monthly samples are small enough that differences under about eight points should be read as noise until several months agree; the August cohort put tech at 18% entry level and consumer at 22%, within that band of September's 14% and 32%.
Writers and researchers: we can cut this by role, panel, or month. Email hi@notchresume.com. Free for working journalists, with attribution.
Frequently asked questions
Do entry-level jobs require experience?
In our September 2026 corpus, postings labelled entry level almost never do: of 150 entry-level postings that stated a minimum, one asked for more than two years. The problem is how few postings are entry level at all, 331 of 2,059, or 16%.
What percentage of jobs are entry level?
16% of the 2,059 postings we collected in September 2026 across five employer panels. It ranges from about 10% at startups to 47% at large US employers such as hospitals, banks, and retailers.
Which jobs have the most entry-level openings?
Customer success (33% of postings with a stated floor accept two years or less), operations (31%), and sales (29%). Marketing (7%), design (7%), and product (9%) have the fewest.
Should I apply if I don't meet the years requirement?
Yes, when the posting treats the number as preferred rather than required, or when your experience is close and the rest of the requirements match. A stated floor is a screening rule only where the employer turns it into a knockout question; most floors are a recruiter's shorthand for the level they expect.
Where does this data come from?
Real job postings pulled weekly from company career sites across five panels (tech, consumer, enterprise, startups, large US employers), with the requirements extracted by the same engine that powers the Notch tool. Method and limitations are on this page and on the About page.
Published October 15, 2026. Data: Notch job-posting observatory, September 2026 cohort. Every number on this page is computed directly from the corpus.