State of the Resume 2026: What Thousands of AI-Scored Resumes Reveal

Since we launched RoastMyResume, our AI has scored thousands of resumes: assigning letter grades, calculating Buzzword Density Scores, and, since September, counting the evidence in every experience bullet. That's a lot of resumes. When you look at that many, patterns emerge that no single review would reveal.

This is our attempt to make sense of those patterns: the State of the Resume in 2026.

A note on methodology, up front. This is not a peer-reviewed study. The sample is self-selected (people who chose to run their resume through an AI roast tool) and it's analysed by a consistent AI scorer, not human reviewers. Every number below is computed from stored scoring results over a rolling window of the most recent resumes, and refreshed on the research page. We don't keep resume text, so we only report what the scorer records. Where we couldn't measure something, we say so rather than estimate. Treat the numbers as directional, not gospel.

With that established, here's what we found.

Finding 1: The biggest buzzword problem isn't adjectives. It's claims with no numbers.

Our Buzzword Density Score rates resumes 0 to 100 on how much fluff they contain, and sorts every flagged phrase into five categories. Across the 460 most recent scored resumes (September 29 to October 6, 2026):

• The median resume scores 22. The middle half of resumes sit between 12 and 28. Almost none score above 75 under the current calibration.
• Vague metrics are the largest category of flagged phrases, at 28%: "improved efficiency," "drove results," "supported strong team performance." Empty adjectives and corporate clichés are close behind at 26% each, inflated verbs 14%, title puffery 6%.
• The most-quoted single phrase is "cross-functional," on about one resume in ten, roughly twice as often as anything else. "Strategic," "team player," "spearheaded," "detail-oriented," and "passionate" follow.

We broke the categories down phrase by phrase in The 2026 Resume Buzzword Report. The top-line finding is simple: the average resume would improve more from deleting words than from adding them.

Finding 2: Half of experience bullets carry a number, and that one habit predicts the grade

Since September 2026 the scorer counts, for every resume, how many experience bullets contain a number, how many state a result, and how many only describe a duty. The counts cover a few hundred resumes at a time (391 at this update, September 29 to October 6) and refresh on the research page, but the shape has held steady:

• About half of experience bullets contain a number. A percentage, a dollar figure, a count, a timeframe, a team size.
• About a third of bullets only describe a duty, with no result stated at all. On roughly a third of resumes, duty-only bullets are the majority.
• Quantification predicts the grade more than anything else we measure. Resumes that put a number in at least six of every ten bullets land in the B range or better most of the time. Resumes with no numbered bullet at all have not yet produced a single B.
• Unsupported self-descriptions are on nearly half of resumes: "strong communicator," "proven track record," "detail-oriented," with no bullet that proves the claim.
• Only about half of resumes open with a specific summary. The rest have a generic one anyone in the field could paste, or no summary at all.

This is the single most actionable finding in the report. Adding a real metric to a vague bullet is the highest-leverage edit most people can make, and most haven't made it. The full method is in 'Responsible For' Is Destroying Your Resume.

Finding 3: Almost nobody tailors

Our Job Match tool scores a resume against a specific posting and lists the keywords the posting wants that the resume isn't showing. We don't retain those results, so we can't put a statistic on them, and we won't invent one. What we can say from reading a great many of them is that the most common gap isn't experience. It's language.

Candidates routinely have the relevant experience and describe it in different words than the posting uses. The resume says "worked with other teams"; the posting wants "cross-functional collaboration." To a system matching on the posting's terms, those don't connect.

Finding 4: AI-written resumes are surging, and they flag for the same phrases

The roast can't tell with certainty whether a resume was written by a person or a model, so we don't have a clean number on AI-written resumes. What we can measure is which phrases get flagged, and the list reads like a language model's favourites: "cross-functional" (the single most-quoted phrase in the dataset), "spearheaded," "leveraged," "strategic," "dynamic." Ask an AI to "make my resume sound more impressive" and it reaches for exactly the words that trigger recruiter fatigue.

This creates a trap: candidates use AI to improve a resume, the AI raises the buzzword density, and the resume ends up sounding like every other AI-assisted resume in the pile. We covered the tells in Can Recruiters Tell If Your Resume Was Written by AI?. The short version: the problem isn't AI, it's unedited AI.

Finding 5: Length doesn't move the grade

Nothing in the grade is a page count. The scorer reads the content, counts the evidence, and ranks the resume; a second page of real results costs nothing, and a padded single page earns nothing. When the roast comments on length, it's commenting on padding: old roles with full bullet lists, the same duties restated under every job.

The takeaway that surprised us: people worry far too much about length and far too little about content. Nobody gets rejected for being one page instead of two. They get rejected for the buzzwords, the missing metrics, and the untailored language in Findings 1 to 3. The length-by-industry breakdown is in How Long Should Your Resume Be?.

Finding 6: The grade distribution is a bell curve tilted toward mediocre

Grades are cut from a blind rank plus the counted evidence, so the distribution reflects the pile rather than the scorer's mood. Under the grading scale in place since late September 2026:

• C+ is the single most common grade, on nearly a third of resumes, and the C range as a whole takes close to two thirds. Competent, unremarkable, indistinguishable from the pile.
• About a quarter land in the B range, most of them B-minus.
• A grades are rare, about one resume in a hundred: genuinely strong, quantified, tailored resumes that would stand out in any stack.
• About one in ten falls to the D range. Usually not because the person is unqualified, but because the resume actively works against them: buzzword soup, zero metrics, format problems, or all three.

The encouraging read: most resumes are C's that could be B's or A's with a few hours of focused editing. The gap between a mediocre resume and a strong one is smaller than most people think, and it's almost always the same handful of fixes.

What it all means: the 2026 resume playbook

Pull the six findings together and the state of the resume in 2026 is remarkably consistent. The average resume fails in the same predictable ways, which means it improves in the same predictable ways. Do these five things and you're ahead of most of the applicant pool:

1. Cut the buzzwords. Delete "team player," "passionate," "detail-oriented," "cross-functional," and their cousins unless you can immediately follow them with evidence. (Finding 1)
2. Add real numbers. Every bullet with a verb-plus-outcome shape needs a metric. If you don't have the exact number, estimate honestly. (Finding 2)
3. Tailor to the specific posting. Mirror its language. Fifteen minutes per application you actually want. (Finding 3)
4. If you use AI, edit what it gives you. Use it as a draft tool, then cut the buzzwords it adds and restore your own voice. (Finding 4)
5. Stop obsessing over length. One page or two, it doesn't matter. Substance density does. (Finding 5)

Do those five things and you move from the C+ plurality into the B-and-above minority. Not because you became a better candidate, but because your resume finally represents the candidate you already are.

The bigger picture

The most striking thing about scoring thousands of resumes isn't any single statistic. It's the consistency. The same mistakes appear across every industry, every experience level, and every career stage. A senior engineering manager and a recent marketing graduate make many of the same errors: too many buzzwords, too few numbers, no tailoring, and increasingly, unedited AI output.

That consistency is good news. It means the fixes are universal too. You don't need insider knowledge or a professional resume writer. You need to add metrics, cut fluff, tailor your language, and, if you're using AI, edit its output. That's the whole game.

If you want to see where your resume falls on every dimension in this report, run it through the free roast: you'll get a letter grade, the evidence counts behind it, a Buzzword Density Score, and the specific phrases dragging you down, in about 30 seconds.