ResumeAdapter
Passes 01 and 03
Updated 2026-08-09

A match percentage
hides the only thing you needed to know.

The problem

Every requirement lands in one of three states, and they need three different responses. One is already fine. One is a twenty-minute rewording of work you genuinely did. One cannot be fixed by writing at all. A single percentage makes all three look the same.

Required vs preferredPer-criterion verdictsRecheck after editing
The three outcomes
Only one is a gap
01
Met and findable

The posting asks for a thing, you have done that thing, and the resume says so in words a reader looking for that requirement would search for.

What to do

Nothing. Leave it alone. The most common self-inflicted wound in tailoring is rewriting the lines that were already working.

02
Met, worded differently

You have genuinely done the work, but the resume describes it in your last employer's internal vocabulary, or in a synonym the posting never uses.

What to do

Restate the same real experience in the posting's language. This is the highest-yield edit on most resumes, and it adds nothing that was not already true.

03
Genuinely absent

There is no experience behind the requirement. No rewording will produce it, because there is nothing to reword.

What to do

Decide honestly: apply anyway if it is a preferred qualification or an adjacent skill you can evidence partially, or accept that this posting is not the one. Never manufacture it.

The quick answer

How do you compare a resume to a job description?

ResumeAdapter Editorial

Per the ResumeAdapter Editorial review of how mismatches present, wording mismatch is consistently the largest single category on resumes from qualified candidates, ahead of genuine skill gaps. The candidate names the work after the tool, the team, or the internal project, and the posting names it after the discipline. See the underlying ATS statistics.

A low result on a requirement you meet is usually vocabulary or parsing. Neither is a reason to stop applying.

Separate the posting into required criteria and preferred qualifications before reading the resume, then take each requirement one at a time and sort it into one of three outcomes. Met and findable means the resume already states it in language a reader searching for that requirement would find, and you leave it alone. Met but worded differently means you did the work and described it in an internal project name or a synonym the posting never uses, and the fix is to restate the same real experience in the posting's vocabulary. Genuinely absent means there is no experience behind it, and no rewording will produce it. Fix the middle category first: it is the highest-yield edit available and it adds nothing untrue. Then confirm the file parses, because a section that does not extract was never in the comparison at all. .

The arithmetic problem

Two resumes, the same score, opposite problems.

Take two candidates against the same posting, both scoring in the same band. A single percentage puts them in the same place. They are not in the same place.

One is twenty minutes of editing. The other is a different job.

The first candidate meets every requirement and describes three of them in a previous employer's internal vocabulary. The second genuinely lacks three requirements and writes beautifully about the rest. The first is one editing session from being a strong applicant, and every fix is a restatement of something true. The second is looking at either a different posting or a longer timeline. Telling both of them the same number serves neither.

This is also why comparing scores across different postings tells you almost nothing. Requirements change from one posting to the next, so a lower number on a harder posting is not a worse resume. The comparison that carries information is your revision against your own earlier version, on the same posting, with everything else held still.

One more failure mode looks exactly like a content problem and is not. If the file did not parse cleanly, the section carrying your strongest evidence may never have entered the comparison. Before you rewrite anything in response to a low result, check what actually extracted.

Worked example

Four criteria, four different verdicts.

An illustrative comparison for a mid-level growth marketing posting, constructed for this page rather than taken from a live posting or a real candidate. Read the right-hand column first: three of these four would look like partial credit in a percentage, and only one is a real gap.

Criterion in the postingLine on the resumeVerdict
Required
5+ years managing paid acquisition budgets

Owned paid acquisition across search and social, 2019 to present, scaling monthly spend from 40k to 300k.

Outcome 01

Stated in the posting's own terms, with a scope and a date range a reader can check.

Required
Experience with marketing attribution modelling

Built the internal Lighthouse dashboard so each channel could be credited for revenue it actually drove.

Outcome 02

This is attribution modelling, described entirely in one company's internal project name. A reader searching the posting's phrase finds nothing.

Required
Managed a team of 3 or more

Mentored two junior marketers and ran the weekly channel review.

Outcome 03

Adjacent but short of the stated bar, and the resume is already accurate about it. Do not round two up to three.

Preferred
Familiarity with SQL

Pulled campaign cohorts directly from the warehouse for weekly reporting.

Outcome 02

Almost certainly SQL, but the resume never names it. A preferred qualification met in substance and invisible in wording.

Two of these four are outcome 02, and both are fixed by naming the discipline alongside the internal description: attribution modelling next to the dashboard, SQL next to pulling cohorts from the warehouse. Neither edit invents anything. The team-size row is the one to leave alone, because rounding two reports up to three is the point at which tailoring becomes a claim you cannot defend in the interview.

What to expect back

Five things a comparison should hand you.

  1. 01
    Output

    Required criteria, listed separately from preferred

    Why

    Postings mix the two on purpose. A comparison that pools them lets a long preferred list drag down a resume that answers every requirement, and hides a missing requirement behind five met preferences.

  2. 02
    Output

    A verdict per criterion, not a total

    Why

    The three outcomes need three different responses. A total tells you the temperature and nothing about what to do next.

  3. 03
    Output

    The posting's own phrasing for anything you should restate

    Why

    For a met-but-worded-differently criterion, the exact phrase is the fix. Being told you are missing something you have done is worse than useless.

  4. 04
    Output

    What survived the parse

    Why

    If a section did not extract, its content was never in the comparison at all, and a low result there is a formatting problem wearing a content problem's clothes.

  5. 05
    Output

    A rerun against the same posting

    Why

    The only honest measure of an edit is the same comparison, on the same posting, before and after. Comparing against a different posting measures the posting, not the edit.

If all you get back is a number, the comparison has done the easy part and left you the hard part.

FAQ

Comparing a resume to a job description: FAQ

The questions people ask once they have a result in front of them and have to decide what to change. Answers are byte-identical to the FAQPage JSON-LD on this page.

How do I compare my resume to a job description?

Split the posting into required criteria and preferred qualifications first, before you open the resume. Then take each required criterion one at a time and decide which of three things is true: it is met and stated in language a reader searching for that requirement would find, it is met but described in different words, or there is genuinely no experience behind it. Do the same for preferred qualifications, then check that the bullets carrying your strongest evidence survive when the file is extracted as text. Fix the second category first, because restating real experience in the posting's vocabulary is the highest-yield edit available and it adds nothing untrue.

What is a good resume match percentage?

There is no threshold worth chasing, and treating one as a target produces worse resumes. A percentage compresses three different situations into one number: requirements you meet and state clearly, requirements you meet but describe in other words, and requirements you do not meet at all. Two resumes can score identically with completely different problems, and the one with a wording mismatch is a twenty-minute fix while the other one may not be fixable at all. Use the number only to compare a revision against its own earlier version on the same posting.

Why does my resume score low when I am qualified for the job?

Usually one of two reasons, and they are both mechanical rather than a judgment about you. The first is vocabulary: you did the work under a different name, whether an internal project title, a previous employer's jargon, or a synonym the posting never uses, and a comparison that matches literal terms cannot see through that. The second is parsing: a multi-column layout, a table, a text box, or a header can drop an entire section during extraction, so the strongest evidence on the page was never read. Both are fixable without changing a single fact about your history.

Should I copy keywords from the job description into my resume?

Copy the phrasing for things you have actually done. That is not gaming anything, it is answering the question in the language it was asked. What you must not do is add a keyword for work you have not done, which is the line between tailoring and fabricating, and it fails at the interview even when it passes the screen. The practical rule: every phrase you lift from the posting has to attach to a specific thing you can describe for two minutes if asked.

Required versus preferred qualifications: which should I focus on?

Required first, always, and treat them as the pass or fail layer. Preferred qualifications are where postings put their wish list, and they are frequently long precisely because they are optional. The useful move on preferred qualifications is not to acquire them but to check whether you already meet some without saying so, which is common: people under-report the adjacent tools and secondary responsibilities they use weekly. Meeting most requirements and a few preferences beats meeting all preferences and missing a requirement.

How often should I re-compare my resume as I edit it?

Once after each meaningful pass of edits, against the same posting, rather than after every line. Comparing too often turns into writing for the comparison instead of for the reader, which is how resumes end up keyword-stuffed and unreadable to the human at the end of the funnel. The check that matters is whether a specific criterion moved from worded-differently to findable. If the wording changed and the criterion did not move, the edit did not do what you thought.

Continue
The rest of the review
Pillar

The five passes of a resume review

Where this comparison sits in the full method, and why parse integrity has to run before any of it.

Read
Rubric

Score it against a published rubric

Five weighted dimensions with band descriptions, so two reviewers reach the same reading.

Read
Checklist

Work through it on paper

Every pass as a pass or fail question, in the two-part order the Department of Labor teaches.

Read
Run the comparison

Paste the posting.
Get a verdict per criterion, not a number.

Required criteria separated from preferred, each one sorted into met, worded differently, or absent, and the posting's own phrasing for anything worth restating. Edit, then run it again on the same posting to see whether it moved. Free to run, and you sign in first.

Read the full method →