The Resume Keyword Gap Report 2026

1. 42% of resumes do not contain the job title they are applying for

The job title is the one keyword every posting contains. It is also the first thing a recruiter’s search and most applicant tracking systems look for. In 42.5% of scans the exact title of the posting, or a close variant, appeared nowhere in the resume. Candidates applying for a “Programme Manager” role describe themselves as a “Delivery Lead”; candidates for “Backend Developer” write “Software Engineer”. The fix costs one line in the summary.

2. Half of resumes match at least three quarters of the keywords, and a quarter match fewer than half

The median resume matches 76 of 100 keyword points. The distribution is two-humped: 53% of resumes score 75 or higher, while 26% score below 50. The scan data cannot say who is in the low group. One possible explanation, consistent with the patterns in the rest of this report but not established by it, is a single general resume sent to every posting.

3. A typical posting asks for 15 keywords. The typical resume misses 6.

Across all scans, 41% of the keywords a posting asks for are absent from the resume. Candidates do prioritise: keywords the posting marks as critical are missed 31% of the time, against 47% for the important-but-not-critical tier. Still, nearly a third of the must-have terms are missing from the average application.

4. The biggest gap is vocabulary, not hard skills

Only 25% of hard-skill keywords such as Python, SQL or Excel are missing. Tools are missed 33% of the time, methodologies 40%, certifications 44%, and domain-specific language such as “FMCG”, “clinical trials” or “trade marketing” 53%. The pattern is consistent across job families. People list what they can do and skip the words that describe the industry they would do it in, which is precisely the language a recruiter searches for.

5. Matched keywords live in the Experience section, not the skills list

37% of matched keywords sit in the Experience section and 27% in a Skills list. Education contributes 1%. Keyword placement is also the weakest of the five formatting components we score: resumes reach 51% of the available placement points, against 77% for formatting safety. A keyword mentioned once in a skills list is easy to add and easy for a recruiter to discount. The same keyword inside a bullet that describes what was built with it is what both software and humans reward.

6. Longer resumes score higher

Resumes under 300 words have a median score of 71; resumes over 1,000 words reach 82, and 66% of them score 75 or higher against 48% of the shortest group. This is not an argument for padding. Nearly half of all resumes fall between 300 and 700 words, where the score barely moves. The gain appears when a resume has enough detail to describe work in the posting’s own terms.

7. Python is the most demanded skill. Kubernetes and TypeScript are the most missed.

Python appears in 507 postings, more than any other skill, and is missing from only 8% of the resumes that target them. The skills candidates leave out are the operational ones: Kubernetes is missing from 46% of resumes targeting postings that ask for it, TypeScript 43%, AWS 34%, Docker 28%. Foundational languages and tools such as Git, HTML, CSS and SQL are almost never missed. One possible explanation is that candidates list what they learned and leave out what they now use every day; the scan data cannot confirm it.

8. Project managers score lowest. Healthcare, sales and HR score highest.

Healthcare resumes match best with a median of 83; project and program management resumes match worst at 62. Project postings are dense with role-specific nouns such as dependencies, budgets, automation, which candidates describe in their own words instead. Software engineering, the largest family with 810 scans, sits at 75; its most-missed terms are TypeScript, Docker, Kubernetes, AWS. Marketing resumes most often miss FMCG, trade marketing, cross-functional collaboration.

9. Nearly half of resumes have no numbers in their bullet points

46% of resumes contain bullet points with no figure in them: no team size, budget, percentage or count. 8.6% have no detectable email address and 8% no recognisable Experience section, usually because of an unconventional heading or a two-column layout. Only 3.3% use tables or columns, the layout most likely to be misread by older parsers. On formatting, then, most resumes are already safe. The remaining gap is content.

Methodology

Limitations

• The sample is self-selected: people who chose to check a resume against a posting with a free tool. It over-represents technology roles and under-represents candidates who do not tailor at all.
• Keyword extraction and scoring are 22 Skills’ own models, not the behaviour of any named applicant tracking system. Different systems weigh keywords differently.
• The length finding is a correlation. Longer resumes have more room for a posting’s terms and usually belong to more experienced candidates.
• Job families are inferred from posting titles, so unusual titles land in an “other” group that is not shown.
• Repeat scans of the same resume or posting are included. Anonymous scans carry no identifier, so the number of unique people is unknown. Deduplicated figures move by three points or less (see Methodology, repeated scans).

Frequently asked questions

What percentage of resumes do not include the job title?

42% of resumes do not contain the job title of the posting they are matched against, based on 3,182 scans analysed by 22 Skills between October 2025 and September 2026.

How many keywords does a job posting typically ask for?

A typical posting asks for about 15 skill keywords, and the typical resume is missing 6 of them.

Which resume keywords are missed most often?

Domain-specific language is missed most (53%), followed by certifications (44%) and methodologies (40%); hard skills are missed least (25%). Among in-demand technical skills, Kubernetes, TypeScript, AWS and Docker are the ones most often left out.

Does resume length affect ATS keyword match?

In this sample, yes: resumes over 1,000 words have a median keyword match of 82, against 71 for resumes under 300 words. The relationship is a correlation; longer resumes have more room for a posting’s terms and usually belong to more experienced candidates.

Where in a resume should keywords go?

37% of matched keywords sit in the Experience section and 27% in a skills list. Keyword placement is the weakest formatting component in the sample, so a keyword inside a bullet that describes real work carries more weight than the same word in a list.

What is the average ATS keyword match score?

The median keyword match score across 4,040 scans is 76 of 100; 53% of resumes score 75 or higher and 26% score below 50.

How to cite

This report and its charts may be quoted and reproduced under a CC BY 4.0 licence with attribution and a link to this page.

22 Skills (2026). The Resume Keyword Gap Report 2026: What 4,040 resume-to-job scans reveal about the keywords job postings ask for and resumes leave out. Version 1.1. https://www.22skills.com/research/resume-keyword-gap-2026

BibTeX

@techreport{22skills2026keywordgap,
  title     = {The Resume Keyword Gap Report 2026: What 4,040 resume-to-job scans reveal about the keywords job postings ask for and resumes leave out},
  author    = {{22 Skills}},
  institution = {22 Skills},
  year      = {2026},
  month     = {9},
  version   = {1.1},
  url       = {https://www.22skills.com/research/resume-keyword-gap-2026},
  note      = {4,040 scans, 2025-10-05 to 2026-09-10. CC BY 4.0.}
}

Data

Every published figure is available as a downloadable table. The CSV has one row per figure with the columns table, label, metric, value, unit and n, and opens directly in Excel, Google Sheets, R or pandas: /research/resume-keyword-gap-2026/data.csv. The same figures are available as structured JSON: /research/resume-keyword-gap-2026/data.json. Both include the publication and modification dates, sample sizes and the citation string.

Charts as images

Every chart is available as a PNG with the source line embedded, for articles, slides and social posts: score distribution, miss rate by priority, miss rate by keyword type, where keywords sit, length and score, most demanded skills, most missed skills, least missed skills, score by job family, formatting gaps.

Quotable sentences

• 42% of resumes do not contain the job title of the posting they are applying for (22 Skills, 4,040 scans, 2026).
• A typical job posting asks for 15 keywords; the typical resume is missing 6 of them.
• Only 25% of hard-skill keywords are missing from resumes, against 53% of domain-specific terms.
• 46% of resumes have bullet points without a single number.
• Resumes over 1,000 words have a median keyword match of 82, against 71 for resumes under 300 words.

Changes

Version 1.1 · 2026-09-10

• Formatting table: the job-title row now uses the same basis as finding 1 (42.5% of 3,182 scans with a title check). Version 1.0 showed 35.1%, the same missing count divided by all 3,853 scans with keyword data, which counted scans without a title check as having the title.
• Methodology: added a repeated-scans section with unique resume, posting and pair counts, and headline figures recomputed on first-scan and last-scan bases.
• Wording: "ATS scans" replaced with "resume-to-job scans" in the subtitle, description and citation; finding 4 retitled "The biggest gap is vocabulary, not hard skills"; two explanatory sentences in findings 2 and 7 labelled as hypotheses.
• Data: figures published as JSON and CSV; citation and Dublin Core metadata added.

Version 1.0 · 2026-09-10

• First publication.

Version 1.1, published 2026-09-10, last updated 2026-09-10. Figures computed on 2026-09-10. Next scheduled update: March 2027. Aggregated data tables are available on request via the press page.