# The Missing Skill Score, Six Dimensions to Claim, Qualify, or Leave Off

*Score any posting term your resume lacks on six dimensions and reach the same claim, qualify, or leave-off call, wording, and placement a second scorer would.*

- Canonical URL: https://www.refolk.ai/candidates/guides/missing-skill-score-claim-qualify
- Pillar: Applying at volume
- Format: Framework
- Published: 2026-11-09
- Last reviewed: 2026-11-09
- Reading time: 20 min
- Keywords: should I put a skill on my resume I barely know, add job description keywords I don't have, how to list beginner skills on resume, working knowledge vs proficient on resume, resume keywords without lying ATS

## Key takeaways

- Greenhouse keyword filtering requires an exact string match, so a skill you hold as SEO counts as missing against a search for Search Engine Optimization unless both forms appear on the page.
- In the Express-Harris survey of 1,002 US hiring decision-makers, 34% said resumes fail to match real skills often or all the time, while only 22% of job seekers in a companion sample admitted listing skills they lack.
- OPM's proficiency scale gives a checkable line for any qualifier: Basic means you can perform simple tasks with guidance, Intermediate means you can perform tasks independently.
- Because working knowledge reads as Basic on one published scale and as intermediate on another, write it only when you can use the skill without guidance.
- In TestGorilla's survey, 76% of employers use skills tests and 36% report testing before resume screening, so an inflated tool claim can fail before a human reads the resume.
- In Refolk's index, US Data Analysts listing Tableau outnumber those listing Power BI 10.6 to 1, against 6.7 to 1 in the UK, so a tool's exact name and market change how much a single line covers.

This guide is the scoring standard for one call you make again and again in a high-volume search. A posting names a skill, tool, or method your base resume doesn't contain, and you have only partly used it. It is written for job seekers tailoring dozens of applications who need the same answer every time. You score the term on six dimensions, get one of three verdicts (claim it, qualify it, or leave it off), and write it on a set line so the claim holds up wherever it gets checked.

## What is a gap term, and why does the call cost more now?

A gap term is any skill, tool, or method a posting names that your base resume does not contain in the same words. The call costs more than it used to because hiring managers now say resumes overstate skills, and more of them are building in ways to check.

Two definitions used throughout:

- **Base resume:** the untailored master version you start each application from.
- **Verdict:** one of three outcomes for a gap term. **Claim** means listing it plainly. **Qualify** means listing it with a word that states the depth or source of your exposure. **Leave off** means not listing it at all.

The employer-side numbers come from four surveys. They agree on the direction even though they measure different things.

| Survey | Figure | Sample | Fieldwork / release |
|---|---|---|---|
| Express-Harris US | 80% sometimes / 34% often | 1,002 hiring decision-makers | Nov 3-19, 2025 / Feb 11, 2026 |
| Express-Harris Canada | 82% / 29% | not established | Feb 2026 |
| Robert Half US | 65% harder to verify | 2,000+ hiring managers | not stated / Mar 10, 2026 |
| TestGorilla US+UK | 76% use skills tests | 1,084 employers + 1,076 seekers | 2025 |

The Express-Harris figures count hiring managers who said candidates' resumes don't match their real-world skills at least sometimes (80%) or all the time or often (34%). In the same release, 86% said AI makes it too easy to embellish a resume. Robert Half found that 42% of hiring managers spend more time reviewing applications and 38% have added interviews per candidate. The Canadian sample size and Robert Half's fieldwork dates are not published in the releases, so give those two rows less weight.

**34% vs 22% - Hiring managers who see skill mismatches often, against job seekers who admit listing skills they lack**

Express-Harris surveyed 1,002 US hiring decision-makers and 1,003 US adults in November 2025.

The gap between those two numbers is the reason this framework exists. One likely explanation is that candidates count partial exposure as having the skill, while the reader counts it as not having it. Nobody is lying outright, but the label is wrong. A qualifier states the partial exposure on the page and closes most of that reading gap.

## The six dimensions of the Missing Skill Score

The six dimensions are posting weight, exposure depth, recency, proof, test likelihood, and synonym coverage. Three of them (depth, recency, proof) decide whether a claim is honest. Two (weight, test likelihood) decide what is at stake. One (synonym coverage) decides only the exact words.

Score each dimension 0, 1, or 2. Synonym coverage is the exception and gets yes, partial, or no.

| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Posting weight | Responsibilities or boilerplate only | Preferred | Required or form screener |
| Exposure depth | OPM Level 1 Awareness | Level 2, with guidance | Level 3+, independent |
| Recency | Before your shown history | Earlier role on the resume | Current or latest role |
| Proof | Nothing to point at | Project, repo, course, certificate | Employer bullet with a result |
| Test likelihood | Rarely checked | Likely interview probe | Testable tool, language, or task |

Synonym coverage scores **yes** if the term is on your resume under another name, **partial** if a related form is there but not one a search would match, and **no** if nothing is there.

These scales are Refolk's rubric. No hiring body publishes them. Each level is tied to something you can look up, such as a posting section, an OPM level, a year, or a named artifact, so two people scoring the same term should land on the same number.

### What each dimension proves, and when it lies

| Dimension | What it proves | When it lies |
|---|---|---|
| Posting weight | How hard the screen filters on the term | A preferred term treated as required, or a form knockout missed |
| Exposure depth | Whether you can do the work unaided | Counting a tutorial as independent use |
| Recency | Whether the skill is current enough to test well | Using the 10 to 15 year work-history rule as if it covered skills |
| Proof | Whether the claim survives a probe | A certificate with no use behind it |
| Test likelihood | How soon a false claim is exposed | Assuming software tests are rare when the rate is unpublished |
| Synonym coverage | Whether a search will find the term | Assuming the ATS links SEO and its expansion |

Some notes on scoring the harder ones.

**Posting weight.** ATS settings follow the posting's sections. In Workday, recruiters mark each skill on a requisition as required or optional. In Greenhouse, required keywords combine as AND and preferred ones as OR. On Indeed, an applicant who fails a required screening question is moved to the Rejected list automatically, while preferred questions only flag the candidate. Workday's own example is a requisition where Time Management is required but Outlook is not, because Outlook can be taught. Whether responsibilities or boilerplate text carries any filter weight is not established publicly, so I score it 0 until a form screener says otherwise.

**Test likelihood.** TestGorilla's employer sample reports 69% using soft-skills tests, 35% self-report tests, and 28% language tests. The same page gives 50% and 30% for cognitive tests, so treat that figure as unresolved. Rates for software-tool, typing, and spreadsheet tests are not published. The best signal available is that Express's examples of overstated skills name Excel and Google Analytics, which are tools that show up on the job quickly. Score any named tool you would use in your first week as 2.

## Which level word matches which exposure?

Use the OPM independence test. If you need guidance to do the task, you are at Basic and the term gets a qualifier. If you can do it on your own, you are at Intermediate or above and can claim it. This is the only line in the published scales that two people can check the same way.

OPM's 2210 FAQ gives short definitions: Level 2 Basic "can perform simple tasks with guidance", Level 3 Intermediate "can perform tasks independently". Resume vocabulary does not line up neatly with those levels:

| OPM level | OPM FAQ definition | Resume Worded term | NIST-cited scale |
|---|---|---|---|
| 1 Awareness | Basic familiarity | novice/beginner | - |
| 2 Basic | Simple tasks with guidance | familiar, working knowledge | limited use |
| 3 Intermediate | Independent | intermediate | working or functional |
| 4 Advanced | Can lead or mentor | proficient/experienced | in depth |

The rows are aligned by level label only. The sources were not written to match each other. Look at "working knowledge" in the table. Resume Worded places it at Basic, while the NIST-cited scale places "working or functional" proficiency at Intermediate. A reader who uses the second scale will expect a level higher than a writer who uses the first.

> **Rule:** One qualifier, one scale
>
> Write "working knowledge" only when you pass OPM's Level 3 test and can do the task without guidance. Below that line, describe where the exposure came from (coursework, certificate project, personal project) instead of rating your level.

Holding "working knowledge" to the higher reading is deliberate. A reader who takes it as Basic will be pleasantly surprised, and a reader who takes it as Intermediate will find exactly what the word promised. The mistake to avoid is a word that reads higher than your actual level.

Indeed's advice points the same way. It warns against labelling skills as beginner unless the role explicitly asks for a developing skill or you are listing a language at a basic level. A context word like "(coursework)" avoids a self-rating altogether and gives the interviewer a specific thing to ask about.

The two dimensions that do most of the work in the verdict are depth and proof. Plotted against each other, they give four outcomes.

#### Depth against proof

Horizontal axis runs from No proof to point at to Named artifact. Vertical axis runs from With guidance (Level 2) to Independent (Level 3+).

| Quadrant | What it means |
| --- | --- |
| Independent, no proof | Qualify, then build a project that shows it |
| Independent, with proof | Claim, after checking recency against test likelihood |
| Guided, no proof | Leave off |
| Guided, with proof | Qualify with a context word naming the artifact |

*Independent use with a named artifact is the only quadrant that earns an unqualified claim, and Level 1 Awareness is left off wherever it falls.*

## How does synonym coverage work when a recruiter searches?

Assume exact match. In Greenhouse, a keyword search only returns an application that contains exactly that keyword. A skill you hold under a different name therefore scores as missing until the posting's own string appears on your page.

What the documentation actually establishes:

- **Greenhouse:** talent filtering requires the search keyword to exactly match the keyword in the application. A secondary source says its wildcards only extend a word stem and do not translate between an abbreviation and its expansion. If you write "SEO", a search for "Search Engine Optimization" will not find you.
- **Workday:** its skills taxonomy maps synonyms, including historical names, spelling variants, and genuine synonyms, onto a canonical skill, and common skills can have 20-plus synonyms. For candidates, though, Workday does not change the skills on the candidate record. When it finds a synonym, it creates a crowdsourced skill, and it always keeps the string the candidate entered.
- **iCIMS, Lever, and Taleo:** how these systems handle synonyms in AI matching is not established from vendor documentation. Treat them as exact-match until you see otherwise.

Coverage never changes the verdict. It only changes the wording. If coverage is yes or partial and the verdict is claim or qualify, write the posting's exact string once, with the other form beside it: "Search Engine Optimization (SEO)".

### Adjacent tools are not synonyms

An adjacent tool does a similar job but has a different name, like Tableau and Power BI. It never counts as synonym coverage. Holding Tableau gives you zero coverage for a Power BI search. The tool you hold should go on its own line in its own name, and the posting's tool needs its own verdict.

Refolk's index shows how uneven the field is for one pair of adjacent tools:

| Market | Tableau | Power BI | Tableau:Power BI |
|---|---|---|---|
| United States | 7,794 | 732 | 10.6x |
| United Kingdom | 1,635 | 244 | 6.7x |
| US:UK | 4.8x | 3.0x | - |

**10.6x - US Data Analyst profiles listing Tableau for every one listing Power BI, in Refolk's index**

The UK ratio is 6.7x, so the same pair of tools sits differently in each market's peer field.

The counts come from Refolk's index of professional profiles, filtered by the title "Data Analyst" plus the exact skill tag. The ratios are derived from those counts. There is an important caution here. The low Power BI count may partly reflect naming variants, such as "Microsoft Power BI" being tagged separately, and that cause is not verified. Even if the reason is uncertain, the data shows the same problem as Greenhouse search: exact-string counting splits one skill across several names. For your resume, that means using the posting's form of the name, whatever form you think of the tool by.

The market difference matters too. In both countries, Tableau is far more common than Power BI among analyst profiles, but the ratio is smaller in the UK. A recruiter searching for Power BI is looking at a smaller field in each market. Whether that makes a well-supported qualified line more visible is plausible, but no data here proves it.

## How does a claimed term get checked after you send it?

A term on your resume can be checked as many as five times: by a form screener, by keyword search or AI matching, in a skills test, by a recruiter screen, and by an interview question about the bullet that contains it. Test likelihood is the dimension that predicts how early in that sequence a weak claim gets exposed.

#### Where a gap term gets checked

1. **Form screener** - A required question can auto-reject regardless of the resume
2. **Keyword search or AI match** - Exact strings decide whether the term is found at all
3. **Skills test** - 36% of testing employers test before resume screening
4. **Recruiter screen** - A short question on the term's depth and source
5. **Interview probe** - Questions on the bullet where the term appears

*An inflated claim can be checked at five points, and for some employers the test comes before any human reads the resume.*

TestGorilla's figures change the order in which this decision should be made. Of the employers surveyed, 76% use skills tests. Among those, 36% test before screening resumes and 62% after. For more than a third of testing employers, then, a claim gets checked before anyone reads your page. Keyword guides usually start with synonym matching and wording. Testing data suggests you should first estimate whether the term will be tested, because that determines how much the claim needs to withstand.

> Choose the wording after you know how the term will be tested.

This is why the verdict rules below treat recency more strictly when test likelihood is 2. If you used a tool independently two jobs ago, you can talk about it convincingly in an interview and still be too rusty for a timed test.

## Scoring one gap term, step by step

Score each gap term in a fixed order, from the posting's text through to the line on your page, and write each score down so the call can be repeated. My working estimate is about five minutes to extract the terms and a couple of minutes per term after that. These timings are not from any study.

The extraction step is the tedious part when you are applying to many companies. [Refolk](/candidates) tailors your resume to each posting from your own work history, so the posting's terms and your existing wording are already side by side when you start scoring. The verdicts are still yours to make.

#### The Missing Skill Score procedure

1. **Extract the gap terms** - Compare the posting with your base resume and list every term your resume lacks in the same words. Tag each one as required, preferred, responsibilities, boilerplate, or form screener.
2. **Score posting weight** - Required and form-screener terms score 2, preferred 1, everything else 0. Open the application form first so a knockout question does not surprise you.
3. **Score synonym coverage** - Record yes, partial, or no, and copy the posting's exact string. An acronym does not cover its expansion in exact-match search.
4. **Score exposure depth and recency** - Map to OPM levels (Awareness 0, Basic 1, Intermediate or above 2) and note the year and role of last use.
5. **Score proof** - Name the artifact: an employer bullet with a result scores 2, a project, repo, course, or certificate scores 1, nothing scores 0.
6. **Score test likelihood** - Score 2 for testable tools, languages, and tasks, 1 for likely interview probes, 0 for rarely checked terms, with one line of reasoning.
7. **Apply the verdict rules** - Leave off, claim, or qualify by the rules in the table below. Do not let weight or coverage change the honesty verdict.
8. **Write the term on its set line** - Claims go on the core skills line and into a bullet, qualified terms go on the second skills line with a context word, and left-off terms appear nowhere.

### The verdict rules

| Verdict | Score rule | Where it goes |
|---|---|---|
| Leave off | Depth 0, or depth 1 with proof 0 | Nowhere; answer screeners truthfully |
| Claim | Depth 2, proof 1+, recency 1+, and recency 2 if test is 2 | Core skills line plus a bullet |
| Qualify | Every other combination | Second skills line or project bullet |

Two adjustments come after the verdict, and neither turns a dishonest claim into an honest one:

1. **Weight 0 and qualify:** a qualified term that appears only in boilerplate is not worth the space. Drop it if the skills block is crowded.
2. **Weight 2 and leave off:** the posting requires something you cannot honestly list. Answer any screener on it truthfully, and decide whether the application is worth sending as a stretch. If your fit score in Refolk is low mainly because of required gap terms, that is the signal to move your time to a better-matched posting.

## Writing the verdict on its set line

Every verdict has a fixed location: claimed terms on the core skills line and in a bullet, qualified terms on a second skills line with a context word in parentheses, and left-off terms nowhere. Fixed locations are what keep dozens of tailored versions consistent, so a recruiter who sees two of your applications sees the same claims both times.

**Skills block with set lines**

```
Skills: Tableau, Excel, Search Engine Optimization (SEO), Time Management
Also used: Power BI (certificate project), Google Analytics (coursework)

Bullet for a claimed term:
Built [what] in [exact posting term] for [who used it], which [result with a number].

Bullet for a qualified term with project proof:
[Project name]: built [what] in [exact posting term] to [purpose], [where it can be seen].
```

*Replace the example terms with your own; keep the posting's exact string first and the other form in parentheses.*

Placement rules:

- **The core skills line holds claims only.** Each one should also appear in at least one employer bullet. A skill that sits only in a list is the easiest kind to probe and the hardest to defend.
- **The second line holds qualified terms only.** The context word goes in parentheses directly after the term, so a keyword search still matches the exact string.
- **Write each form once.** Repeating a term doesn't make your claim stronger. It only makes the page harder to read.
- **Never put a level word on the core line.** "Proficient in" adds nothing to a claim that is backed by a bullet, and it's the word most likely to be tested.

Keep a written record for each term so that the next posting naming it reuses the same call and you don't score it from scratch.

**Gap term score record**

```
Posting: [company], [role], [requisition ID]
Term (exact posting string):
Section: required / preferred / responsibilities / boilerplate / form screener
Posting weight (0-2):
Synonym coverage (yes / partial / no), form already on page:
Exposure depth (OPM level 1-5, scored 0-2):
Recency (year and role of last use, scored 0-2):
Proof (named artifact, scored 0-2):
Test likelihood (0-2), reason:
Verdict: claim / qualify / leave off
Line: core skills / also used / bullet in [role] / none
Exact words written:
```

*One block per term per posting; keep them with your base resume so repeat terms reuse the same call.*

## How the score goes wrong

Most bad calls come from scoring a dimension against the wrong yardstick. The common ones are exact-match search, synonym merging, qualifier words, posting sections, tested tools, and recency rules. Each has a false positive that looks like a good score and a check that catches it.

> **Watch out:** The most common false positive
>
> Holding SEO and assuming you are covered for Search Engine Optimization. In Greenhouse that term counts as missing, because search needs the exact string. Put both forms on the page once.

1. **Synonym coverage counted as yes when search is exact-match.** You hold "SEO" and the recruiter searches "Search Engine Optimization". Greenhouse requires an exact match, and wildcards only extend a word stem. *Check:* look for the posting's exact string on your page, not something that means the same thing.
2. **Assuming Workday's synonym merging helps you.** Workday resolves synonyms for its own canonical skills, but for candidates it creates a crowdsourced skill and keeps your string. It does not rewrite your entry as the canonical skill. *Check:* write the posting's string even when the system is Workday.
3. **Qualifier words read two levels apart.** "Working knowledge" is Basic to Resume Worded and intermediate on the NIST-cited scale. *Check:* tie each qualifier to one named scale. In this framework, that means OPM's independence test.
4. **Wrong posting weight.** Treating a preferred term as required wastes effort, but missing a form-level knockout costs you the application. On Indeed, a required screener rejects automatically, whatever the resume says. *Check:* open the form before scoring weight.
5. **Underscoring test likelihood for everyday tools.** Excel and Google Analytics appear in Express's examples of overstated skills. These are the claims most likely to be caught on the job, not just in a test. *Check:* any tool you would use in your first week scores 2.
6. **Applying a work-history rule to a skill.** The 10 to 15 year guidance from Debra Wheatman and the eight to 10 year guidance from Dana Leavy-Detrick are about career history. No source sets a recency threshold for skills. *Check:* score recency against your own resume's roles, and drop obsolete software, as Indeed advises.
7. **Reading headcount gaps as skill scarcity.** In Refolk's index, gaps between exact skill tags can reflect how a skill is named as much as how many people hold it. The Power BI count is the open example. *Check:* treat the ratios as a measure of naming and market differences, not a precise count of skill holders.

A further limit applies to the framework as a whole. The surveys measure what hiring managers believe and report, not how often an inflated line actually costs someone a job. The rubric's cut-points are a judgement built on the OPM levels and the ATS documentation, not the result of a validation study. If you have hiring-side access, the test is whether two colleagues scoring the same five gap terms on the same posting reach the same verdicts.

## Before you send, and keeping the scores current

A tailored resume is ready when every gap term has a written verdict, every claim has a bullet behind it, and every exact string from the posting that you have earned appears on the page. Run the checklist on each version before you send it.

#### Gap term checks before sending

- [ ] Every gap term on the posting has a score record with a verdict
- [ ] Every claimed term appears on the core skills line in the posting's exact string
- [ ] Every claimed term also appears in at least one employer bullet with a result
- [ ] Every qualified term carries a context word in parentheses, not a level word
- [ ] "Working knowledge" appears only for skills you can use without guidance
- [ ] Acronym and expanded forms each appear once for terms that have both
- [ ] No left-off term appears anywhere on the page or in a screener answer
- [ ] Every form screener answer matches the verdict on the resume
- [ ] Terms scored test likelihood 2 have a recency score of 2 or are qualified

Scores go stale in predictable ways, so re-score at these points instead of on a schedule:

- **When you finish a project or certificate.** Proof can rise from 0 to 1, which can turn a leave-off into a qualify. Update the score record so future postings get the new verdict.
- **When you change roles.** Recency for everything you used in the old role drops from 2 to 1. Check any claimed term with test likelihood 2.
- **When a vendor changes its search.** The exact-match rule rests on Greenhouse's talent filtering documentation and the synonym behaviour on Workday's skills documentation. Re-read both when you see results that don't fit, and look for published behaviour for iCIMS, Lever, and Taleo, which is not established here.
- **When new employer surveys come out.** Recheck the mismatch and testing rates against the sample sizes and fieldwork dates in the first table before quoting them, because a release date is not a fieldwork date.

The score record is what makes this work at volume. The tenth posting that asks for Power BI should take seconds, because you already made the call, wrote the line, and kept the reason with it.

## Frequently asked questions

### Should I put a skill on my resume I barely know?

No, not if barely means awareness only. At OPM Level 1 Awareness you apply the skill only in the simplest situations under close guidance, and this framework leaves those terms off. If you can do simple tasks with guidance and can name a project, course, or certificate that shows it, add it on a second skills line with that context in parentheses, such as Power BI (certificate project). That way the reader knows what you have before they test it.

### What is the difference between working knowledge and proficient on a resume?

There is no single standard, and that is the problem. Resume Worded lists working knowledge as a Basic-level term, while a scale cited by NIST treats working or functional proficiency as intermediate. Proficient sits higher, at the advanced or experienced level. To avoid being read two levels apart, use working knowledge only when you can do the task independently, which is OPM's Intermediate line, and use a context word like coursework or project for anything below that.

### Can I add job description keywords I don't have if I qualify them?

Only if you have some real exposure and something you can point to. A qualifier describes partial exposure accurately. It does not cover a skill you have never used. Under this framework, a term with no exposure beyond awareness, or with only guided exposure and no proof, is left off whatever the qualifier says. If the term is required and you leave it off, answer any form screener truthfully and treat the posting as a stretch.

### How do I list beginner skills on a resume without saying beginner?

List where the exposure came from instead of rating yourself. Indeed advises against labelling skills as beginner unless the role asks for a developing skill or you are listing a language at a basic level. A line like Google Analytics (coursework) or Power BI (certificate project) tells the reader the depth without a self-rating that undersells you, and it gives an interviewer a specific thing to ask about.

### Does the ATS understand that SEO and Search Engine Optimization are the same thing?

It depends on the system, so do not count on it. Greenhouse talent filtering needs an exact keyword match, and a secondary source says its wildcards only extend a word stem rather than linking an abbreviation to its full form. Workday resolves synonyms for its own canonical skills but keeps the string a candidate types. Behaviour in iCIMS, Lever, and Taleo is not established publicly. Write both forms once, for example Search Engine Optimization (SEO).

### How recent does a skill need to be to list it on a resume?

No published source sets a recency cut-off for skills. The 10 to 15 year guidance from resume writers is about work history, not skills. This framework ties recency to your own resume instead: full credit if you used the skill in your current or latest role, partial credit if you used it in an earlier role still on the page, and none if you used it before the history you show. Drop obsolete software, as Indeed advises.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/missing-skill-score-claim-qualify*