How Many Keywords Should a Resume Have? What Postings Show
"How many keywords" sounds like it should have a number for an answer, and plenty of advice supplies one. The trouble is that no applicant tracking system publishes a keyword quota, and the postings you apply to do not ask for the same number of things. What you can measure is how concentrated postings are: how few terms most employers in a role actually repeat. That gives a more useful answer than any fixed count.
Why a fixed keyword count does not exist
The systems that read your resume do not count to a threshold.
In Greenhouse's Talent Filtering, a recruiter chooses the keywords, marks them as "Preferred" (only one must appear) or "Required" (all must appear), and the documentation states: "the keyword from your search must exactly match the keyword in the application." A resume either contains the term or it does not. Writing it ten times does not add anything to a match on that term.
Greenhouse's AI ranking, Talent Matching, is explicit that more is not better: "multiple terms can map to the same calibrated skill — so a longer list of matched terms doesn't always mean a higher match score." And when recruiters set up that calibration, the Talent Matching guide tells them: "we recommend focusing on 4–6 key skills."
That is a vendor telling recruiters to judge a role on a handful of skills. Your keyword work should aim at the same handful.
How concentrated postings really are
We measured which terms recur in each role's postings. Each percentage is the share of full-text postings that mention the term, from the ATS Resume AI analysis of all 503 S&P 500 careers sites and 9,260 live job postings, June 2026 (role pages).
| Role | Full-text postings | Most common term | Tracked terms in 30%+ of postings | In 20%+ | In 10%+ |
|---|---|---|---|---|---|
| Product Manager | 72 | Roadmap, 90.3% | 5 | 6 | 9 |
| Sales | 368 | Pipeline management, 55.7% | 9 | 11 | 15 |
| Customer Service | 34 | Customer success, 76.5% | 4 | 5 | 8 |
| Finance | 85 | Stakeholder management, 55.3% | 6 | 8 | 11 |
| Data Scientist | 101 | Cross-functional collaboration, 47.5% | 3 | 7 | 11 |
| Marketing | 84 | Cross-functional collaboration, 44.0% | 4 | 5 | 9 |
| Operations | 109 | Stakeholder management, 30.3% | 1 | 3 | 10 |
| Software Engineer | 628 | Python, 26.6% | 0 | 4 | 12 |
Two patterns stand out. Some roles have a dominant term: "roadmap" in 90.3% of product manager postings is close to universal. Others are fragmented: no tracked term reached 30% of software engineer postings, because the stacks differ by team. Python led at 26.6%, Java was at 21.2% and AWS at 18.5% (software engineer data).
What the numbers mean for your count
For a concentrated role, the core is short and predictable. A product manager resume that does not say "roadmap" is missing a term nine in ten postings use. A sales resume should show pipeline management, forecasting and SaaS, each of which appeared in more than half of full-text sales postings in our sample (sales data).
For a fragmented role, the generic core helps less and the specific posting matters more. With no tracked software engineering term above 30% in our analysis (software engineer data), the language that decides a match is whatever this particular team listed: their cloud, their languages, their tools.
Either way the working number is small. Between the role's recurring core and the posting's own required list, most resumes need somewhere around the 4–6 skills Greenhouse tells recruiters to calibrate on, plus the posting's exact job title and any named certification. Everything beyond that is ordinary description of your work, and it should read like description, not like a list assembled for a machine. If a term from the posting is not true of you, leave it out: a recruiter who searches for it and finds it will ask about it.
How many times to repeat each keyword
Twice is enough for most terms: once in a skills line, where a scan or search finds it quickly, and once in a bullet, where a reader sees evidence. Recruiters see where a term appears. A Greenhouse full-text search result shows a snippet of how the search terms are used in the profile (Greenhouse Support). A bullet that uses the term to describe a result reads better in that snippet than a bare list.
Repetition beyond that has a cost. In Ladders' 2018 eye-tracking study, "Evidence of keyword stuffing" was one of the traits of the worst-performing resumes, and its advice was that "keywords should be presented in context."
Build your keyword list in five steps
Copy the required section of the posting
Paste the requirements into a plain-text file. Ignore the company boilerplate and benefits.
Underline the hard nouns
Tools, methods, certifications, the job title, and domain terms. Soft phrases such as "self-starter" rarely work as search terms.
Check the role's core
Compare against the role's recurring terms on our resume keyword pages or with the free resume keyword generator. Add any core term the posting implies and you genuinely have.
Cut to what you can prove
Keep only terms you could talk about in an interview. A term you cannot back up costs you at the conversation stage.
Place each term twice
Once in the skills line, in the posting's spelling. Once in a bullet with a result attached. Then read the resume aloud; if a sentence sounds like a list, rewrite it.
A worked example
Posting excerpt for a product manager: "Own the product roadmap; partner cross-functionally with engineering and design; define KPIs; run user research."
Keyword list after the steps: product roadmap, cross-functional collaboration, KPIs, user research, product strategy. Five terms, matching the top of our product manager data, where roadmap was at 90.3%, cross-functional collaboration at 63.9% and KPIs / metrics at 31.9% of full-text postings (product manager data).
Before: "Managed product for the payments team."
After: "Owned the payments product roadmap and ran cross-functional planning with engineering and design; set the KPIs for checkout conversion and ran user research interviews every quarter."
The second version uses four of the five terms once, in a sentence a recruiter can check. That is the whole job. If you want a check against a specific posting, the free resume checker needs no account.
Sources
- Greenhouse Support, "Talent Filtering"
- Greenhouse Support, "Talent Matching"
- Greenhouse Support, "Talent Matching FAQ"
- Greenhouse Support, "Search resumes for keywords"
- Ladders, "Eye-Tracking Study" (2018), Internet Archive copy
Frequently asked questions
- Is there an ideal keyword density for a resume?
- No ATS vendor documents one. Greenhouse keyword filtering checks whether an exact term is present, and its AI matching says a longer list of matched terms does not always raise the score. Presence and context matter; density does not.
- How many skills should I list in my skills section?
- Enough to cover the posting's required tools and the role's recurring core, which our data puts at a handful of terms for most roles. A short skills line in which every item relates to the job reads better than a long one padded with loosely related terms.
- Should I include keywords from the preferred qualifications?
- Yes, when they are true. In Greenhouse Talent Filtering a recruiter can mark keywords as preferred, meaning any one of them qualifies a candidate for the results, so each true preferred term widens the searches you appear in.
- Do I need different keywords for every application?
- The role's core terms stay mostly the same, but the posting-specific tools change, especially in fragmented roles such as software engineering, where no single tracked term reached 30% of postings in our data.