How Resume Check works
Resume Check compares your resume with a job posting by fixed, published rules, not AI. Checked with the same rules on the same day, the same resume and posting always give the same result.
This page is made from those rules themselves, so what you read here is what runs on your check: when a rule changes, this page changes with it. Match pack 0.11.0, 36 rules, last changed on 9 November 2026. To try it, open Resume Check.
How sure each rule is
Every rule says what it rests on, in one of these words:
- Documented by the official source: the system’s maker, the data’s publisher, or the authority whose law it is, says so in its own documents.
- Seen in our tests: reproduced in Resume Check’s own tests.
- Reported by recruiters and testers: reported by recruiters, careers services or independent testers, but not confirmed by the maker or our tests.
- Our best judgement: reasoned but not measured; each such rule gives its reason.
Of the 36 rules: “Documented by the official source”, 4; “Our best judgement”, 32.
Reading the posting
Resume Check splits the posting into lines and reads each line with the rules below, in this order: the first rule that decides a line wins. It reads a posting of up to 200,000 characters and 2,000 lines; a line longer than 1,000 characters is shown but not read. You can move any requirement between the two lists, and the check uses your choice.
The 99 headings that open the must-haves
requirements · requirement · the requirements · job requirements · role requirements · key requirements · essential requirements · minimum requirements · basic requirements · core requirements · candidate requirements · requirements and qualifications · qualifications and requirements · required · qualifications · required qualifications · minimum qualifications · basic qualifications · essential qualifications · key qualifications · your qualifications · qualifications and experience · qualifications and skills · skills and qualifications · skills and experience · experience and skills · experience and qualifications · skills · key skills · required skills · essential skills · core skills · skills required · experience · required experience · essential experience · must have · must haves · must-have · must-haves · must have skills · must-have skills · essential · essentials · the essentials · essential criteria · what you'll bring · what you will bring · what you bring · what you'll need · what you will need · what you need · what you'll have · what you will have · what we're looking for · what we are looking for · what we look for · who we're looking for · who we are looking for · who you are · about you · you have · you'll have · you will have · your profile · your skills · your experience · your background · profile · the ideal candidate · ideal candidate · person specification · what you'll need to succeed · what you need to succeed · what you'll need to be successful · you're a great fit if · you'll be a great fit if · is this you · what it takes · you might be a fit if · you'll thrive if you have · you'll thrive if · you will thrive if · who are you · your skills and experience · essential skills and experience · education and experience · experience and education · knowledge, skills and abilities · required skills and experience · skills and requirements · requirements and skills · qualifications required · experience required · additional requirements · what we need from you · what we expect from you · what we expect · physical requirements
The 72 headings that open the nice-to-haves
nice to have · nice to haves · nice-to-have · nice-to-haves · nice to have skills · nice-to-have skills · nice-to-have qualifications · preferred · preferred qualifications · preferred skills · preferred experience · preferred requirements · preferable · preferred but not required · bonus · bonus points · bonus skills · bonus points if you have · bonus points for · bonus if you have · desirable · desirables · desirable criteria · desirable skills · desirable experience · desirable qualifications · desired · desired skills · desired qualifications · desired experience · good to have · great to have · nice extras · extra credit · pluses · plus · a plus · big plus · it would be great if you have · it would be great if you · it would be great if · it would be nice if you have · it's a plus if you have · it's a bonus if you have · even better if · even better if you have · you'll stand out if · you will stand out if · what would be nice · what would make you stand out · what will make you stand out · what'll make you stand out · ideally · ideally you have · ideally you'll have · ideally you will have · we'd love to see · what we would love to see · what we'd love to see · not required but a plus · desirable skills and experience · desired skills and experience · preferred skills and experience · preferred, but not required · bonus points if you've · bonus points if you · we'd love it if you have · we'd love it if you · we would love it if you have · we'd also love to see · ideally you'll also have · ideally you will also have
The 119 headings that close the requirements
about us · about the company · about the role · about the job · about the team · about the position · about the opportunity · about · who we are · our company · our team · our story · our mission · our values · our culture · the company · company · the role · the job · the position · the team · the opportunity · role overview · job overview · position overview · overview · summary · job summary · role summary · position summary · job description · description · role description · responsibilities · key responsibilities · main responsibilities · core responsibilities · your responsibilities · job responsibilities · role and responsibilities · roles and responsibilities · duties · key duties · main duties · duties and responsibilities · what you'll do · what you will do · what you'll be doing · what you will be doing · what you'll work on · what you will work on · your role · your mission · your impact · in this role · in this role you will · day to day · day-to-day · a day in the life · your day · your day to day · benefits · our benefits · what we offer · what we can offer · what we'll offer you · what's in it for you · what you get · what you'll get · why join us · why us · why work with us · perks · perks and benefits · benefits and perks · compensation · compensation and benefits · salary · salary and benefits · pay · pay and benefits · location · working hours · hours · schedule · contract · job details · role details · key information · additional information · other information · more information · how to apply · to apply · application process · next steps · interview process · our process · hiring process · recruitment process · equal opportunity · equal opportunities · equal opportunity employer · diversity and inclusion · diversity, equity and inclusion · what we offer you · our offer · we offer · what you can expect · your tasks · tasks · key tasks · your day-to-day · the day-to-day · why you'll love working here · why you will love working here · why you'll love it here · our perks · what you get from us
The 23 headings about the company, its offer or its pay
Under these, a line about years of experience or a screening question is read as the company’s, never as a question for you.
about us · about the company · who we are · what we do · benefits · perks · benefits and perks · what we offer · what's in it for you · why join us · why work with us · why work here · our culture · our values · our story · pay · salary · compensation · pay and benefits · salary and benefits · compensation and benefits · pay range · salary range
MT-HEAD-MUST: A heading that opens the must-haves
Postings commonly list what an applicant needs under headings such as "Requirements" or "What you'll bring"; the list is our own, not measured on real postings.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-HEAD-NICE: A heading that opens the nice-to-haves
Postings commonly list what would help but is not needed under headings such as "Nice to have" or "Preferred"; the list is our own, not measured on real postings.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-HEAD-OTHER: A heading that closes the requirements
Postings put the company, the duties and the benefits under headings of their own; reading those lines as requirements would add must-haves the posting never asked for.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 9 November 2026: The headings about the company or its offer (companyHeading: a fixed list, "Life at <X>", "About <X>" and "Why <X>" only for a capitalised name), under which a years or question line is no ask; the final pass adds the pay headings (Pay, Salary, Compensation, Pay and benefits, Salary and benefits, Compensation and benefits, Pay range, Salary range), where such a line holding pay is a pay fact
MT-HEAD-INTRO: A line that introduces the must-haves or the nice-to-haves
Postings often open a list with a sentence of their own ("The ideal candidate will have:"); its cue words say which list follows.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-AGE: A line about age is never read
Resume Check never infers or displays age, so an age line is neither matched against a resume nor asked as a question.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-NEGATION: A line that says something is not needed
A posting that waives something ("no degree required") asks for nothing, so the line is not a requirement or a question.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-YEARS: A line that asks for years of experience
Total years can be compared with a resume's dates; years of one named skill cannot, so they become a question; the phrasing list is our own.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: A years phrase with "experience" ends at its clause, and the rest of the line is read for another question ("Bachelor's degree and 3 years of experience")
- Version 1.2.0, 9 November 2026: The company's own years are no ask only in closed shapes ("We bring 30 years …", "Our team has …", "With 20 years in logistics, Acme …", "has been trading for over 25 years", "N years in business"); "Our …" alone no longer drops a line ("Our ideal candidate has 5+ years" is an ask)
MT-KNOCKOUT: A line that asks a screening question
Postings ask these questions in their own words; a line that asks one is answered yes or no, so it is a question to check, not a skill to match; the phrasing lists are our own.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: A skill opener asks only a language question; on-site activities, office work and "ready to join a startup" ask nothing; remote within a place and travel are location questions; CEFR levels are language questions
- Version 1.2.0, 8 November 2026: A skill opener governs only its first clause (MT-YEARS' CLAUSE_BREAK), the rest asks its own question ("Relevant experience and a full UK driving licence"); "background check" is no skill opener; "start work" and a bare time are notice periods; a level asks a language only beside it ("Advanced knowledge of English law" asks none); "on-site client work" and "degree-level understanding" ask nothing
MT-FACT: A fact about the posting, not a requirement
A posting's date, pay and details describe the job, not the applicant, so they are never matched against a resume.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-HEAD-UNKNOWN: A heading in no list closes the section
A short line ending in a colon or question mark, or set as a heading, usually starts a new part of the posting; reading what follows as under no heading is safer than keeping it under the previous one.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-PROSE: Description, duties and instructions are not requirements
A line about the company, a duty or how to apply asks the applicant to have or know nothing, so matching it against a resume would add a must-have the posting never set.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: "This role requires …", "Our ideal candidate is …" and "If you have …, even better" state requirements, not prose
MT-CUE-MUST: A cue word that makes a line a must-have
Words such as "must", "required" and "at least" say a posting needs the thing named; the word list and its exceptions are our own.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: "This role requires …" at the line's start is a must cue
MT-CUE-NICE: A cue word that makes a line a nice-to-have
Words such as "preferred", "a plus" and "ideally" say a posting would welcome the thing named without needing it; the word list is our own.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: "even better", "all the better" and "so much the better" are nice cues
MT-NICE-CLAUSE: A nice cue in a side clause keeps a must-have
A line such as "Python, ideally with Django" asks for Python and would welcome Django; reading it as a nice-to-have would drop the must-have it states first.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-SECTION: No cue word: the section decides
A line listed under "Requirements" or "Nice to have" belongs to that list unless its own words say otherwise.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
MT-NO-SECTION: No section: only a line that states something to have or know
A posting without headings mixes the title, a description and duties with what it asks for; only a line that names something to have or know, or carries a cue word, reads as a requirement.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version
- Version 1.1.0, 8 November 2026: "Our ideal candidate is …" names something to have or know
Matching your resume
Each requirement is matched against your resume’s lines. The strongest kind of match found wins, and the lines that meet it are shown (at most 50). In the verdict, a must-have counts:
| How it is met, as the screen says it | Counts | Rule |
|---|---|---|
| Met: your resume uses the same words | In full (4 quarters) | MT-EXACT, MT-LIST-ITEMS |
| Met: your resume uses the acronym or its full form | In full (4 quarters) | MT-ACRONYM |
| Met: your resume uses another word for it | In full (4 quarters) | MT-SYNONYM |
| Partly met: your resume has some of its key words | 2 quarters | MT-PARTIAL |
| Not found in your resume | Nothing (0 quarters) | None of the above |
Matching by meaning isn’t used (see below).
The 24 acronyms in Resume Check’s own table
- CAP: certified analytics professional. The credential's own short name, as its awarding body (INFORMS) writes it.
- CUA: certified usability analyst. The credential's own short name, as its awarding body (Human Factors International) writes it.
- CSPO: certified scrum product owner. The credential's own short name, as its awarding body (Scrum Alliance) writes it.
- PMP: project management professional. The credential's own short name, as its awarding body (PMI) writes it.
- CPA: certified public accountant. The licence's own short name.
- CFA: chartered financial analyst. The credential's own short name, as its awarding body (CFA Institute) writes it.
- ACCA: association of chartered certified accountants. The body's own short name.
- SEO: search engine optimization, search engine optimisation. The usual short name, in US and UK spelling.
- PPC: pay per click. The usual short name for pay-per-click advertising.
- CRM: customer relationship management. The usual short name.
- ERP: enterprise resource planning. The usual short name.
- KPI: key performance indicator. The usual short name.
- KPIS: key performance indicators. The usual short name, plural.
- AWS: amazon web services. The service's own short name.
- GCP: google cloud platform. The service's own short name.
- GDPR: general data protection regulation. The regulation's own short name.
- UX: user experience. The usual short name.
- UI: user interface. The usual short name.
- QA: quality assurance. The usual short name.
- HR: human resources. The usual short name.
- IT: information technology. The usual short name.
- AI: artificial intelligence. The usual short name.
- ML: machine learning. The usual short name.
- NLP: natural language processing. The usual short name.
The 14 acronyms taken from O*NET® data
O*NET® 31.0 Database, built 2026-11-09.
- ABAP: advanced business application programming. Listed in the Software Skills of O*NET® 31.0 data as "Advanced business application programming ABAP", for 23 occupations.
- BGP: border gateway protocol. Listed in the Software Skills of O*NET® 31.0 data as "Border Gateway Protocol BGP", for 9 occupations.
- DDL: data definition language. Listed in the Software Skills of O*NET® 31.0 data as "Data definition language DDL", for 5 occupations.
- DML: data manipulation language. Listed in the Software Skills of O*NET® 31.0 data as "Data manipulation language DML", for 5 occupations.
- GIMP: gnu image manipulation program. Listed in the Software Skills of O*NET® 31.0 data as "GNU Image Manipulation Program GIMP", for 6 occupations.
- HCPCS: healthcare common procedure coding system. Listed in the Software Skills of O*NET® 31.0 data as "Healthcare common procedure coding system HCPCS", for 40 occupations.
- IAFIS: integrated automated fingerprint identification system. Listed in the Software Skills of O*NET® 31.0 data as "Integrated Automated Fingerprint Identification System IAFIS", for 6 occupations.
- JCL: job control language. Listed in the Software Skills of O*NET® 31.0 data as "Job control language JCL", for 23 occupations.
- NIBIN: national integrated ballistics information network. Listed in the Software Skills of O*NET® 31.0 data as "National Integrated Ballistics Information Network NIBIN", for 5 occupations.
- SAML: security assertion markup language. Listed in the Software Skills of O*NET® 31.0 data as "Security assertion markup language SAML", for 5 occupations.
- SQR: structured query report. Listed in the Software Skills of O*NET® 31.0 data as "Structured Query Report SQR", for 3 occupations.
- UML: unified modeling language. Listed in the Software Skills of O*NET® 31.0 data as "Unified modeling language UML", for 25 occupations.
- VBA: visual basic for applications. Listed in the Software Skills of O*NET® 31.0 data as "Microsoft Visual Basic for Applications VBA", for 45 occupations.
- VHDL: vhsic hardware description language. Listed in the Software Skills of O*NET® 31.0 data as "Very high-speed integrated circuit VHSIC hardware description language VHDL", for 6 occupations.
The 14 groups of names for the same thing
- postgres, postgresql. PostgreSQL's own documentation calls it Postgres too.
- customer interviews, user interviews. Both name interviewing the people who use a product.
- supplier management, vendor management. Supplier and vendor name the same party in purchasing.
- kubernetes, k8s. The project's own short name.
- golang, go. The Go project's own alternate name (golang.org).
- javascript, js. The usual short name.
- react, react.js, reactjs. Spellings of the one library's name.
- vue, vue.js, vuejs. Spellings of the one framework's name.
- node.js, nodejs. Spellings of the one runtime's name.
- e-commerce, ecommerce. One word, spelt with or without a hyphen.
- front-end, frontend. One word, spelt with or without a hyphen.
- back-end, backend. One word, spelt with or without a hyphen.
- full-stack, fullstack. One word, spelt with or without a hyphen.
- microsoft office, ms office. The suite's name, with its maker's name written out or short.
Table version 1.4.0. Every row is “Our best judgement”, with its reason.
MT-KEY-WORDS: The words of a requirement that count
A requirement names its skill in its key words; the frame around them ("experience with", "a strong knowledge of", "is a must") and the small function words name nothing a resume has to hold.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
- Version 1.1.0, 8 November 2026: An "or" list is read as alternatives (joiner); an address is never a key word
- Version 1.2.0, 8 November 2026: The frame of "This role requires …" and "Our ideal candidate is …" is never a key word
MT-NAME-GUARD: Names that are also everyday words, people, employers or letters
A false exact match tells someone they wrote a skill they did not, so a tool named with an everyday word, a given name or a letter counts only where it is written as the name, and never in a person's name, an address or an employer.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
- Version 1.1.0, 8 November 2026: Addresses, employers (after "at", before a company form), all-capitals lines, reference lines, "Surname, Given" lines and role lines never count a name; notBefore and years ("Spring 2021", "Go live"); letters never as grades, categories or levels, and also after "in", "using", "with", "and", "or"; capitals rows not after a number, and own forms ("SAFe"); a name inside a phrase must be written as the name
- Version 1.2.0, 8 November 2026: "Surname, Given" only where a word is no tool (TECH_NAMES) or both are names ("React, Redux" counts); a tech name between slashes is no address ("ASP.NET/C#")
MT-EXACT: Exact: the requirement's own words on one line
A line that holds every key term of a requirement, word for word, says what the requirement asks in the posting's own words; anything looser is partial.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
MT-LIST-ITEMS: A list met item by item
A resume's skills section lists one name per line or item, so a posting's list of names is met when every name is listed, though no one line holds them all; a list joined by "or" asks for any one of them.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
- Version 1.1.0, 8 November 2026: A list joined only by "or" or "and/or" is met by any one item (`any`)
MT-ACRONYM: Acronym: the table, or defined in either text
An acronym and the words it stands for name the same thing when the table says so or the text defines it; written in capitals, an acronym cannot be the everyday word it spells.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
- O*NET® 31.0 Database: Software Skills (CC BY 4.0), the names written with their own acronym (checked 9 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
- Version 1.1.0, 9 November 2026: The acronym rows taken from the Software Skills names in O*NET® 31.0 data, conservatively (onet-acronyms.ts)
MT-SYNONYM: Synonym: another name from the table, or another form of the words
Two names of one thing (by its maker's naming, a spelling variant, or plainly the same task), or the same words as a plural or in UK or US spelling, meet the same requirement; only table rows a person chose count, so nothing is guessed, and it is never called exact.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
- Version 1.1.0, 8 November 2026: Plurals and UK/US spellings folded, at synonym strength
MT-PARTIAL: Partial: some of the requirement's key words
A line that shares some of a requirement's distinctive words may be about it, so it is shown, labelled partial and counted half; sharing only a word found in most job ads says nothing.
- Gold postings (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (the match map, rules only)
Screening questions and years
From the posting’s own words, Resume Check lists the questions an application form may ask. You write your own answer to each, and Resume Check compares it with what the posting asks. The kinds, as the screen names them: Work authorization, Years with a skill, Degree, License, Location, Relocation, Language, Notice period and Salary expectation.
Each question shows “Met”, “Not met by your answer”, “To check” or “Years compared”. A question is “Not met by your answer” only when your own answer contradicts what the posting asks; one your resume doesn’t answer is “To check”, never a failure.
MT-ANSWER: What the user's own answer says
An answer is read only for yes or no, a sponsorship need, a time, an amount or a number of years, by fixed word lists; anything else is left unclear rather than guessed.
- Gold pairs' answer sheets (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (knockouts, years and the verdict)
- Version 1.1.0, 8 November 2026: "I am / I have / I live / I hold" are a yes only before an ability or a permission; explicit statements of a right to work; "not applicable" is unsure and "not currently employed" no notice; in a posting a bare number is no pay figure; the years tied to a skill
MT-RESUME-CONFIRMS: Where the resume confirms a screening question
A degree, a place, a language or a licence the resume states can confirm the question that asks for it; a resume that does not mention one says nothing either way, so it never makes a question fail.
- Gold pairs' knockout statuses (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (knockouts, years and the verdict)
- Version 1.1.0, 8 November 2026: A place stands as itself (no capitalised word before it, no other qualifier after it); an associate or foundation degree confirms no bachelor's degree
MT-KNOCKOUT-STATUS: A screening question: met, failed, short or check
A question can only fail on the user's own answer contradicting what the posting states (rule-pack-spec.md §6, "clearly failed"); a resume that says nothing, or a place that differs, leaves it for the user to check.
- Gold pairs' knockout statuses (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (knockouts, years and the verdict)
- Version 1.1.0, 8 November 2026: A bare yes or no decides only a line of plain polarity (a sponsorship line or one with its own negation gives check: answer-ambiguous), explicit right-to-work statements decide any; a pay period on one side only is check unless the figures are of one order; a met salary is "not above the top"; the skill years tied to the skill; the resume never confirms travel; "start in N weeks" is a notice limit
- Version 1.2.0, 8 November 2026: A skill's years below the posting's are short (half or more, or under half), never failed, with the plain facts as a note; a bare no to them is check
MT-EXPERIENCE: Years of experience from the role dates
Total experience is the months the role dates cover, both end months counted and overlaps merged; a year written alone is read as widely as written, so no shortfall is put on the person that the resume does not state.
- Gold pairs' months and years statuses (tests/tools/gold-match.ts), scored by npm run gold:match (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (knockouts, years and the verdict)
- Version 1.1.0, 8 November 2026: The resume reading's month grammar ("Dec2019", "Sept.2016"), "Present." and "Till now", a month alone as a start takes its end's year; the narrowest reading of year-only dates; yearsUncertainty (dates not read, or year-only readings that disagree)
The verdict
The fit for a job is a verdict, never a number, and every verdict says “Our best judgement”: its weights and thresholds are this published formula, not measured against hiring results. It sits beside the readings of the resume itself and is never combined with them into one score.
Coverage is the must-haves’ quarters (above) out of 4 for each must-have, compared in whole numbers. The worst verdict that applies wins:
- Strong fit: coverage of at least 90%, no screening question asked as a requirement that your answer fails, and the years asked for met, or none asked for.
- Plausible fit: coverage of at least 70% and under 90%, and no screening question asked as a requirement that your answer fails.
- A stretch: coverage of at least 40% and under 70%, or years, in total or with a skill, of at least half but not all of those asked for.
- Not competitive for this one: coverage under 40%, a screening question asked as a requirement that your own answer clearly fails, or years, in total or with a skill, under half of those asked for.
Nice-to-haves never lower the verdict; they are reported apart. A posting with no requirements at all gets no verdict (“This posting lists no clear requirements”), and when there is no must-have and no years to compare, the verdict waits for your answers to the screening questions (“Answer the screening questions to see a verdict.”). Formula version 1.3.0.
MT-FIT: The fit verdict and its reasons
Our best judgement until outcome data exists: the weights and thresholds are the published formula, not measured against hiring results.
- rule-pack-spec.md §6 (the formula) and the gold pairs' verdicts (tests/tools/gold-match.ts) (checked 8 November 2026)
Changes
- Version 1.0.0, 8 November 2026: First version (knockouts, years and the verdict)
- Version 1.1.0, 8 November 2026: Years the dates cannot decide (dates not read, or year-only readings that disagree) never lower the verdict and are said to be for the user to check; FIT_SWITCHES holds two rules then still open
- Version 1.2.0, 8 November 2026: No verdict only with truly no requirements (no must-have, no years requirement, no must-class question), so a posting of questions and years gets a verdict without a coverage reason; a skill's short years weigh by the total-years rule, in plain facts; FIT_SWITCHES removed (both rules decided)
- Version 1.3.0, 8 November 2026: A verdict needs a deciding fact, so with no must-have and no years that can decide, questions all still to check withhold it ("Answer the screening questions to see a verdict.")
Occupation, nearby roles and pay figures
Each country has its own list of occupations. Resume Check looks up the posting’s job title in the US list and the UK list separately, by published rules. Check both, and change either one if it doesn’t fit.
The job title is read from the posting’s first 20 lines: a line labelled “Job title:”, “Title:”, “Position title:” or “Role title:” first; otherwise the first line when it reads as a title (at most 120 characters and 12 words); otherwise a line labelled “Position:”, “Role:”, “Job:”, “Job role:”, “Vacancy:” or “Post:” whose words read as a title (at most 10 words). It is also tried without the level words at its start (senior, sr, snr, junior, jr, jnr, lead, principal, staff, graduate, entry level, mid level) and a level number at its end (i, ii, iii, iv, v, 1, 2, 3, 4, 5). One occupation is shown by itself only when the country’s own index says which; otherwise you choose from a list, of which the first 30 are shown:
- In the US: Listed with an occupation whose own O*NET® title it is first, then the one named for it by the most O*NET® sources, ties in code order.
- In the UK: Listed by the coding index’s own marks: its entry for the title alone first, then the entries with no condition, its default first, then the entries with a condition, defaults first, each in the index’s order.
Nearby roles are shown only for “A stretch” or “Not competitive for this one”: up to 20 related occupations from O*NET® data, in its own order and tiers (the five most closely related; the next five; ten more), never compared with your resume.
What employees in this occupation were paid, as each publisher measured it. These figures are not an estimate for you or for this job. Each table shows the 10th percentile, 25th percentile, median, 75th percentile and 90th percentile, with its source, year and the publisher’s own notes.
MT-OCCUPATION: The posting's title, and its occupation in each country's own index
The indexes are the publishers' own; which line is the title, the level words and the order of a list are ours. One occupation is shown by itself only when the index says which (its own title, the only one, the coding index's default), never on a title's last words; otherwise the person chooses, and can always choose another or give another title.
- O*NET® 31.0 Database: Job Titles, Occupation Data and Sample of Reported Titles (CC BY 4.0) (checked 9 November 2026)
- O*NET® 31.0 Data Dictionary: Job Titles (the Source(s) codes) (checked 9 November 2026)
- ONS, SOC 2020 Volume 2: the coding index and coding rules and conventions (version 14, 27 August 2026) (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version
MT-NEARBY: Nearby roles: related occupations from O*NET® data, for a stretch or not competitive
The list, its order and its tiers come from O*NET® data; Resume Check only decides when to show it (a stretch or not competitive).
- O*NET® 31.0 Data Dictionary: Related Occupations (10 primary and 10 supplemental per occupation, three tiers, ordered by expert review) (checked 9 November 2026)
- O*NET® 31.0 Database (CC BY 4.0) (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version
MT-PAY-PANEL: Public pay figures for the occupation, with their source, year and caveats
The figures, their quality markers and their gaps are the publishers' own; which state the posting names is read from its place only, and a code that is also a country's needs a US sign (a province code, a Canada sign).
- BLS, Occupational Employment and Wage Statistics, May 2025 (public domain; BLS asks to be cited) (checked 8 November 2026)
- BLS, OEWS May 2025 national and state files (oesm25nat.zip, oesm25st.zip), their "Field Descriptions" sheet: the notes defining "*" and "#", and the "annual" and "hourly" fields (checked 9 November 2026)
- ONS, ASHE Table 14, 2026 provisional (22 October 2026): Table 14.7a and its CVs, 14.7b (Open Government Licence v3.0) (checked 9 November 2026)
- USPS state abbreviations; ISO 3166-1 alpha-2 country codes (the codes both use) (checked 9 November 2026)
- Canada Post province and territory symbols (the codes the place reading takes) (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version
- Version 1.1.0, 9 November 2026: One place reading for the pay panel and the pay-law fact (postingPlaceRead): labelled lines and header-like lines only; the District of Columbia also as "Washington, D.C.", "Washington DC" or "District of Columbia", never "… metro"; Canadian provinces and territories; Cleveland and Columbus, Ohio, when the only Ohio place; up to three place words, quoted without a label or stop word; never a Team or Department line, a head-office segment, a "Washington, D.C. metro"; "Based in" as a label; a ZIP or US sign as a word; any other possible place (a code without its sign, a foreign country or region, a labelled location value or list part with no state, province or law city, such as "Remote" or "Multiple Locations", unless a money amount) blocks the state
- Version 1.2.0, 9 November 2026: The US figures from BLS's May 2025 files: BLS's "*" ("a wage estimate is not available", in its notes) is shown as not available, never a number, and "#" as "or more" than the top figure its notes give; a row with only "*" is said as such instead of a table; for the posting's state, a state where BLS publishes no row for the occupation, or only "*", is said beside the national table
About this posting
Facts the posting gives about itself, and lines to read side by side. None of this changes the verdict.
A posting fetched from a job board shows the board’s own date, named as the board’s documents describe the field:
- Greenhouse (
first_published): the date the posting was first published. - Lever (
createdAt): the date the posting was created in Lever, which can be before it was first shown. - Ashby (
publishedAt): the date the posting was last published, which may not be the first time.
These put the posting’s own lines side by side. They say nothing about the employer. What the screen says, with the figures it uses:
- The posting calls this role entry-level and asks for 3 years of experience.
- The posting lists 13 must-haves. Resume Check points out lists of more than 12: a number we chose as a reading aid, not one from research.
- The posting asks for 3 years in one line and 5 years in another.
- One line of the posting says no degree is needed, and another asks for one.
- The job title is a senior one, and another line calls the role entry-level.
- One line of the posting says the role is fully remote, and another asks for set days in an office.
Entry-level wording is put beside a must-have of 3 or more years, and a list is pointed out above 12 must-haves: figures we chose as reading aids, not from research.
MT-POSTING-DATE: The posting's date, from the job board's data or its own words
Each board's date is labelled as its own documentation describes the field (Greenhouse's by its name, as the documentation shows it without a description); a date in the posting's words is read only when written in full, and days are counted from the analysis date.
- Greenhouse Job Board API: first_published in the "Retrieve a job" example answer (not described) (checked 9 November 2026)
- Lever API, Postings: createdAt, "Datetime when posting was created in Lever" (checked 9 November 2026)
- Lever Postings API (public): no date field documented (checked 9 November 2026)
- Ashby public job posting API: publishedAt, "ISO DateTime when the job was last published" (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version (the posting's facts and warnings)
MT-PAY-LAW: No pay stated, where the place has a law asking for pay in job postings
Each row is the official authority's own statute, ordinance or guidance; the posting never shows the employer's size or the other conditions a law sets, so the fact says what the law asks and whom it covers, never that it applies; any money amount in the posting blocks it, since a false "no pay" claim is worse than a missed one.
- California: California Labor Code section 432.3 (checked 9 November 2026)
- Colorado: the Equal Pay for Equal Work Act, C.R.S. 8-5-201(2) (checked 9 November 2026)
- Connecticut: Connecticut General Statutes section 31-40z, as amended by Public Act 26-12 (checked 9 November 2026)
- Delaware: Title 19 of the Delaware Code, section 709C (checked 9 November 2026)
- Hawaii: Hawaii Revised Statutes section 378-2.8 (checked 9 November 2026)
- Illinois: the Illinois Equal Pay Act of 2003, 820 ILCS 112/10(b-25) (checked 9 November 2026)
- Maine: Title 26 of the Maine Revised Statutes, section 622-A (checked 9 November 2026)
- Maryland: the Labor and Employment Article of the Maryland Code, section 3-304.2 (checked 9 November 2026)
- Massachusetts: Massachusetts General Laws chapter 149, section 105F (checked 9 November 2026)
- Minnesota: Minnesota Statutes section 181.173 (checked 9 November 2026)
- New Jersey: New Jersey’s pay transparency law, N.J.S.A. 34:6B-23 (checked 9 November 2026)
- New York State: New York Labor Law section 194-b (checked 9 November 2026)
- Vermont: Title 21 of the Vermont Statutes, section 495p (checked 9 November 2026)
- Virginia: Code of Virginia section 40.1-28.7:12 (checked 9 November 2026)
- Washington State: RCW 49.58.110 (checked 9 November 2026)
- the District of Columbia: the Wage Transparency Omnibus Amendment Act of 2023, D.C. Code 32-1453.01 (checked 9 November 2026)
- Cleveland, Ohio: Cleveland’s Codified Ordinances, chapter 669 (checked 9 November 2026)
- Columbus, Ohio: Columbus City Codes chapter 2335, as amended by Ordinance 2898-2025 (checked 9 November 2026)
- British Columbia: the Pay Transparency Act, SBC 2023, c 18, s 2 (checked 9 November 2026)
- Prince Edward Island: the Employment Standards Act, RSPEI Cap. E-6.3, s 19 (checked 9 November 2026)
- Ontario: the Employment Standards Act, 2000, s 8.2 (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version (the posting's facts and warnings)
MT-ENTRY-YEARS: Entry-level wording beside a must-have of 3 or more years
An entry-level role is one a person can start with little experience, so a must-have of 3 or more years beside that wording is worth reading side by side; 3 is our own reading aid, not a statistic.
- Gold posting data-analyst-wexcombe (tests/tools/gold-match.ts), scored by npm run gold:match (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version (the posting's facts and warnings)
MT-LONG-LIST: More than 12 clear must-haves
A very long must-have list is worth a second look at which lines the role truly needs; 12 is our own reading aid, not a statistic, and no source is claimed for it.
- Gold posting full-stack-engineer-quillmark (tests/tools/gold-match.ts), scored by npm run gold:match (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version (the posting's facts and warnings)
MT-CONTRADICTION: Two lines of the posting that clearly say opposite things
Only pairs whose own words disagree are shown (two numbers of years for one subject, no degree needed and a degree required, entry-level and a senior title, fully remote and set office days), because a false contradiction misleads more than a missed one.
- Gold posting frontend-developer-halyard (tests/tools/gold-match.ts), scored by npm run gold:match (checked 9 November 2026)
Changes
- Version 1.0.0, 9 November 2026: First version (the posting's facts and warnings)
The pay-law table
Where a posting states no pay and its place has a law in this table, “About this posting” says so: what the law asks and whom it covers. It never says the law applies to the employer, which depends on things the posting doesn’t show, such as the employer’s size or where the work is done. A row is used only from its date, and a row checked more than 180 days before a check says it may be out of date. This is not legal advice.
Places are read as US states, the District of Columbia, Canada’s 13 provinces and territories, and Cleveland and Columbus in Ohio. Table version 1.0.0.
| Place | Law | From | Who it asks | Official source |
|---|---|---|---|---|
| California | California Labor Code section 432.3 | In force from 1 January 2023 | Employers with 15 or more employees | Source, checked 9 November 2026 |
| Colorado | The Equal Pay for Equal Work Act, C.R.S. 8-5-201(2) | In force from 1 January 2021 | Employers with one or more employees in Colorado | Source, checked 9 November 2026 |
| Connecticut | Connecticut General Statutes section 31-40z, as amended by Public Act 26-12 | In force from 1 October 2026 | Every employer | Source, checked 9 November 2026 |
| Delaware | Title 19 of the Delaware Code, section 709C | In force from 26 September 2027 | Employers with 26 or more employees | Source, checked 9 November 2026 |
| Hawaii | Hawaii Revised Statutes section 378-2.8 | In force from 1 January 2024 | Employers with 50 or more employees | Source, checked 9 November 2026 |
| Illinois | The Illinois Equal Pay Act of 2003, 820 ILCS 112/10(b-25) | In force from 1 January 2025 | Employers with 15 or more employees | Source, checked 9 November 2026 |
| Maine | Title 26 of the Maine Revised Statutes, section 622-A | In force from 29 July 2026 | Employers with 10 or more employees | Source, checked 9 November 2026 |
| Maryland | The Labor and Employment Article of the Maryland Code, section 3-304.2 | In force from 1 October 2024 | Every employer | Source, checked 9 November 2026 |
| Massachusetts | Massachusetts General Laws chapter 149, section 105F | In force from 29 October 2025 | Employers with 25 or more employees | Source, checked 9 November 2026 |
| Minnesota | Minnesota Statutes section 181.173 | In force from 1 January 2025 | Employers with 30 or more employees at sites in Minnesota | Source, checked 9 November 2026 |
| New Jersey | New Jersey’s pay transparency law, N.J.S.A. 34:6B-23 | In force from 1 June 2025 | Employers with 10 or more employees over 20 calendar weeks | Source, checked 9 November 2026 |
| New York State | New York Labor Law section 194-b | In force from 17 September 2023 | Employers with 4 or more employees | Source, checked 9 November 2026 |
| Vermont | Title 21 of the Vermont Statutes, section 495p | In force from 1 July 2025 | Employers with 5 or more employees | Source, checked 9 November 2026 |
| Virginia | Code of Virginia section 40.1-28.7:12 | In force from 1 July 2026 | Every employer | Source, checked 9 November 2026 |
| Washington State | RCW 49.58.110 | In force from 1 January 2023 | Employers with 15 or more employees | Source, checked 9 November 2026 |
| The District of Columbia | The Wage Transparency Omnibus Amendment Act of 2023, D.C. Code 32-1453.01 | In force from 30 June 2024 | Employers with one or more employees in the District | Source, checked 9 November 2026 |
| Cleveland, Ohio | Cleveland’s Codified Ordinances, chapter 669 | In force from 27 October 2025 | Employers with 15 or more employees in the city | Source, checked 9 November 2026 |
| Columbus, Ohio | Columbus City Codes chapter 2335, as amended by Ordinance 2898-2025 | Enforced from 1 January 2027 | Employers with 15 or more employees in the city | Source, checked 9 November 2026 |
| British Columbia | The Pay Transparency Act, SBC 2023, c 18, s 2 | In force from 1 November 2023 | Every employer | Source, checked 9 November 2026 |
| Prince Edward Island | The Employment Standards Act, RSPEI Cap. E-6.3, s 19 | In force from 30 June 2026 | Every employer | Source, checked 9 November 2026 |
| Ontario | The Employment Standards Act, 2000, s 8.2 | In force from 1 January 2026 | Employers with 25 or more employees on the day the job is posted | Source, checked 9 November 2026 |
Wording
The Wording tab reads the words of your summary and your experience bullets and points out what a person reading them may notice. It quotes your own words and never writes a line for you. These are its 20 rules, version 1.8.0.
The human readability reading
One reading from 0 to 100, worked out from five parts. It is one of Resume Check’s readings and is never combined with the others into a single score. With no experience bullets there is not enough to measure, and no number is shown.
| Part | Points out of 100 |
|---|---|
| Specific lines | 40 |
| Measurable results | 20 |
| Stock phrases | 15 |
| Verbs | 15 |
| Grammar | 10 |
Specific lines earn full points when 7 in 10 lines are not generic; measurable results when 2 in 5 bullets hold one. Stock phrases earn nothing at 3 per 100 words, verbs at 1 issue for every 2 bullets, and grammar at 3 slips per 100 words. These figures are our own best judgement; no study gives them.
Which lines are read
WD-LINES: The lines the words checks read
The summary's sentences and the roles' bullets. The summary headings, where a summary ends and the sentence split are our own reading, checked on the words gold; no source says which lines a human reader judges first.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version
Specific or generic
WD-SPECIFICITY: Generic, plain and distinctive lines
A line is generic with no concrete anchor, distinctive with two or more. What counts as an anchor, that generic also needs two signs of vagueness (or a summary sentence about oneself), and both thresholds are our own reading, set on the words gold and frozen.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (specificity)
- Version 1.1.0, 13 November 2026: A summary sentence opening with a pronoun is one sign of vagueness, no longer generic by itself
Stock phrases
WD-STOCK: Stock phrases
Our own dictionary: phrases common enough on resumes that they tell a reader little. No public list was cited, as none is a source we could check.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
Numbers and results
WD-MEASURABLE: Bullets with a measurable result
A digit, a percentage, an amount, a multiple or a written number. That a year, a version, a standard, a grade, a time or a label with a number is no measure, and a fraction after "by" is one, is our own reading.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
WD-SECOND-LOOK: Figures worth a second look
Round figures and "every" or "all" claims, asked about as a question, never called false. Which figures count, and when "every" or "all" is a frequency, a count or a set phrase, are our own reading.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
- Version 1.1.0, 10 November 2026: "every" and "all" only when about a result (CLAIM_NOUNS: "every deadline"), never plain scope
Verbs
WD-WEAK-VERB: Weak opening verbs
Our published list ("helped", "worked on", "assisted with", "was involved in"): openings that say you were near the work, not what you did. Our own judgement, no study.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
WD-TENSE: Tense slips
A past role's bullet in the present tense, or a current role mixing tenses. The verb tables and the noun-like words are our own; a word not listed has no tense, never a guess.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
WD-PASSIVE: Passive voice
"was", "were" or "been" with a past participle. The adjectives left out are our own.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
WD-RESPONSIBLE-FOR: "Responsible for"
"responsible for" names a duty, not what was done. Our own judgement, no study.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
WD-REPEATED-VERB: The same opening verb on three or more bullets
The same opening verb on three or more bullets reads as a list of the same thing. The count of three is our own.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version (stock phrases, numbers and verbs)
Grammar
WD-GRAMMAR: Five clear grammar slips
Our own short rules in place of a grammar engine. Which cases count as clear, and every word list, are our own reading, kept narrow so a careful sentence is never corrected wrongly.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (our own grammar checks)
The human readability reading
WD-READING: The human readability reading
The weights and the zero and full points are our own published formula, labelled an assumption until outcome data calibrates it; no study gives them.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 10 November 2026: First version
Conventions where you are applying
WD-TARGET: The country a resume is written for
From the resume’s own location line only; otherwise the person picks. Which names and codes a location line may end with is our own list; an ambiguous code gives none.
- Country conventions table, each row with its official sources (checked 13 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (conventions)
WD-LENGTH: Longer than the target country’s guidance
Each country’s figure is the largest page count an official or public-service career source for it gives, named in the row. Taking the largest where sources differ is our own choice, so a resume is called long only when every official figure says so.
- Country conventions table, each row with its official sources (checked 13 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (conventions)
WD-NAMING: The document’s own name
Only where the row’s source says a CV is another document than a resume (the US: CareerOneStop). That the name is read from the first three lines, as a line of its own, is our own reading.
- Country conventions table, each row with its official sources (checked 13 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (conventions)
Personal details
WD-PERSONAL: Personal details the resume states
Lines recognised as personal details, quoted, never worked out further. Which labels count, and that a line must open with one and a colon, are our own reading, chosen so nothing is inferred.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (personal details)
WD-PHOTO: A photo where guidance advises against one
Only that the file holds a picture near the top, and only where the target country’s row says a photo is advised against. The picture is never looked at.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (personal details)
WD-GRADUATION: An education year long ago
20 years or more before the analysis date. The year is the education entry’s own; no age is ever worked out from it.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (personal details)
Certifications a reader can verify
WD-CREDENTIAL: Where a certification can be checked
A curated table of certifications with each issuer’s own verification page, opened on the day in the table. Which names and acronyms a line is matched by, and that a name several bodies issue gets no link, are our own reading.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (verify at source)
Where a missing must-have would belong
WD-KEYWORDS: Where a missing requirement belongs
A skills line for a tool or skill, the carrier bullet when Job match names one, otherwise the summary; always "only if it’s true for you". The ceiling of three mentions is our own published figure, an assumption with no study behind it; the length cost reads the target country’s official page count.
- Our own fictional test resumes, scored by a test run (checked 10 November 2026)
Changes
- Version 1.0.0, 13 November 2026: First version (keyword guidance)
The conventions table
Each row is taken from official or public-service career guidance for that country, checked on 13 November 2026. Where the guidance gives no figure, the table gives none and nothing is compared.
The certifications table
33 issuing bodies, each with its own verification page, checked on 13 November 2026. Resume Check contacts no issuer and looks nothing up: it only links the page.
- Project Management Institute: anyone can look a holder up
- PeopleCert: a reader needs the credential number
- Engineering Council: a reader needs the credential number
- ACCA (Association of Chartered Certified Accountants): a public register
- SHRM: anyone can look a holder up
- Scrum Alliance: anyone can look a holder up
- Scrum.org: anyone can look a holder up
- The Linux Foundation: a reader needs the credential number
- Amazon Web Services: the holder shares a link
- Google Cloud: anyone can look a holder up
- Microsoft: the holder shares a link
- Cisco: a reader needs the credential number
- CompTIA: the holder shares a link
- ISC2: a reader needs the credential number
- ISACA: a reader needs the credential number
- CFA Institute: a public register
- ICAEW: a public register
- CIPD: a reader needs the credential number
- CIMA (AICPA & CIMA): a public register
- INFORMS: anyone can look a holder up
- Human Factors International: anyone can look a holder up
- IAAP (International Association of Accessibility Professionals): anyone can look a holder up
- Salesforce: anyone can look a holder up
- Snowflake: anyone can look a holder up
- Databricks: anyone can look a holder up
- The Open Group: anyone can look a holder up
- HashiCorp: anyone can look a holder up
- Oracle: the holder shares a link
- Red Hat: a reader needs the credential number
- NEBOSH: a reader needs the credential number
- IOSH: a reader needs the credential number
- HRCI (HR Certification Institute): anyone can look a holder up
- Google (Skillshop): the holder shares a link
How the wording rules are tested
The rules are checked against 25 resumes written for these tests, for made-up people. Every expected result was written by hand, never copied from Resume Check’s output. When these rules were published: 1,464 checks, 1,464 correct, 0 failed. These counts are from Resume Check’s own test resumes. They don’t measure how it does on real resumes, and they aren’t a promise about yours.
Wording rules fingerprint: b9887d15df61.
Coach mode
Its questions, the way your answers are laid out, the career read’s eight rules and the interview prep cards’ three patterns are fixed and published with the other rules. The same resume gives the same result. A line it lays out holds your own answers and four joining words (“for”, “from”, “to” and “with”), and nothing else.
The questions about a line
For a line the Wording tab calls generic, Coach mode reads which kind of work the line is about from its own words, says so, and asks that kind’s four questions. You can say it read the line wrongly and get the general questions. An answer you leave empty shows as [...]; nothing is put in its place.
| A line about | The four questions |
|---|---|
| Leading people | What was the group responsible for? What kind of people were in it? Say their jobs, not their names. How many people? What was the result? Give a number if you have one, or leave it empty. |
| Making or building something | What did you make? Who was it for? Say what kind of people or team, not a name. What did you make it with? Name the tools. What was the result? Give a number if you have one, or leave it empty. |
| Helping customers | What kind of people did you help? Say what kind, not names. How did you help them? How many did you help, and in what period? What was the result? Give a number if you have one, or leave it empty. |
| Regular, recurring work | What was the work, exactly? How often did you do it? How much work was it? Give a count if you have one. What was the result? Give a number if you have one, or leave it empty. |
| Selling | What did you sell? What kind of buyers did you sell to? Say what kind, not names. How much did you sell, and in what period? What was the result? Give a number if you have one, or leave it empty. |
| Analysis or reporting | What did you study or check? What did you use? Name the tools. Who used your findings? Say what kind of people or team, not a name. What was the result? Give a number if you have one, or leave it empty. |
| Teaching or training others | What did you teach? What kind of people did you teach? Say what kind, not their names. How many people? What was the result? Give a number if you have one, or leave it empty. |
| Changing something for the better | What did you change? How did you change it? What was it before? Give the figure if you have one, or leave it empty. What was it after? Give the figure if you have one, or leave it empty. |
| Organizing work between groups | What did you organize? Which groups did you bring together? Say what kind, not names. How large was it? Give a count if you have one. What was the result? Give a number if you have one, or leave it empty. |
| Writing for readers | What did you write? Who reads it? Say what kind of readers, not names. How much did you write, or how often? What was the result? Give a number if you have one, or leave it empty. |
| Any other kind of work | What did you do? Say it in one short phrase. How did you do it? How much was it, or for how long? What was the result? Give a number if you have one, or leave it empty. |
The career read
Eight rules about the document, never about you. Each quotes up to 3 of your own lines. None gives a number that isn’t counted from your resume, and none says what any employer will do. These rules are our own reading of resumes, not a measured fact about hiring.
| It says | When | In these words |
|---|---|---|
| Lines that say what came of the work What the document does well | An experience bullet holds a result word and a number, and names something specific. | {count} of your experience bullets name a result and give a figure. A figure gives a reader something definite to ask about. |
| Lines that name what the work was done with What the document does well | Two or more lines name a tool, a method or a product and are specific. | {count} of your lines name a tool, a method or a product and say something specific about the work. A reader can tell from them what the work was done with, without guessing. |
| More than one title at the same employer What the document does well | One employer appears with two or more different titles. | The document shows {count} different titles at one employer. A reader can see that the title changed; the bullets under each title are where to say how the work did. |
| No bullet gives a figure Where the document says less | The resume has 4 or more experience bullets and none holds a number. | None of your {bullets} experience bullets gives a figure. The document says what the work was, and leaves a reader to guess how much of it there was and what came of it. |
| Most lines could be on anyone’s resume Where the document says less | 4 or more lines were read and more than half name nothing specific. | {generic} of the {lines} lines read name nothing specific: no figure, tool, name or result. A reader can’t tell from these lines what your part of the work was. |
| The current role says less than a past one Where the document says less | The one current role is at least 6 months old, and a role that has ended has 4 or more bullets and at least twice as many. | Your current role has {current} bullets; a role that has ended has {earlier}. The document says more about that past work than about the work you do now. |
| A long time under one title What a reader may notice without saying so | The one current role, the only role at its employer, runs 10 years or more by the narrowest reading of its dates. | Your current role runs ten years or more under one title. The title alone doesn’t show how the work changed over that time. If it did change, the bullets are where to show it: what you took on, and when. |
| Nothing was read under the current role What a reader may notice without saying so | The one current role has no bullets read under it. | No bullets were read under your current role. It is the role the document gives as your present work, and nothing was read about what you do in it. |
Interview prep cards
One card for each must-have of a Job match, up to 40. The question on a card is one of three patterns, chosen by how the Job match found the requirement on your resume. It is not what any employer is known to ask. The boxes are empty until you write in them.
| The card | The question | The boxes |
|---|---|---|
| Your resume shows this | Can you tell me about a time you did this? | Situation · Task · Action · Result |
| Your resume shows part of this | How far does your experience with this go? | Situation · Task · Action · Result |
| Prepare an honest answer | Do you have any experience with this? | The closest thing I have done · What I have not done · How I would pick it up |
Resume and profile, side by side
You paste the experience section of a profile you keep elsewhere, up to 20,000 characters. Nothing is fetched from any site, and the pasted text is not kept. A role is lined up with an entry only when the employer’s name matches as a whole and there is one clear candidate; anything less certain is left alone. It can show three things: the title differs; the dates differ; a role of your resume whose employer is nowhere in the pasted text. It never says which side is right.
Freelance work and gaps
Questions whose answers are laid out for you to edit, the same way as a line. For a gap, which kind you pick only chooses the note shown; it stays on your device and is never part of your line.
| You pick | The note | The questions |
|---|---|---|
| Freelance, contract or platform work | One heading for the whole stretch of work, then a line for each piece of work you want a reader to see. If you don’t want to name a client, you can describe them instead. | The entry’s heading: What do you call yourself in this work? Use the words a client would use. When did it start? When did it end? If it hasn’t, say so in your own words. A piece of work: What was the work, in a few words? Who was it for? You can describe the client instead of naming them: what kind of organization, and how big. What tools or methods did you use? (optional) What was the result, if there was one? (optional) |
| Something else | It can be just a name for this time and its dates. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| Caring for someone | This line doesn’t have to say who you cared for, or why. It can be just a name for this time and its dates. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| Health | This line doesn’t have to say anything about health. It can be just a name for this time and its dates. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| Study | You can name the course or the subject if you want a reader to know it. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| A role that ended | This line doesn’t have to say why the role ended. It can be just a name for this time and its dates. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| Relocating | This line doesn’t have to say why you moved. It can be just a name for this time and its dates. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
| Working for myself | If you worked for clients in this time, choose “Freelance, contract or platform work” above to write that work as entries. | The line: What do you want to call this time, in your own words? When did it start? When did it end? If it hasn’t, say so in your own words. Is there anything you did in this time that you want a reader to know? (optional) |
What you write here is your choice of words, and so is how much you say. Nothing here is legal, medical or employment advice.
Coach tables: questions 0.1.0, career read 0.1.0, cards 0.1.0, profile comparison 0.1.0, builders 0.1.0.
Known limits
Some postings are read wrongly. Some of these are on purpose, because reading them the other way could show a false fact; others are left for now. Each example below is run through the rules whenever this page is made, so the page can’t keep a limit that is no longer true.
- “Viterbo, VT”: read as Vermont. Two capital letters after a place name are read as a US state or a Canadian province when the posting names no other country, so a place elsewhere that uses the same letters is read as one too (here, an Italian province’s code). The screen always shows the words it read, so you can check them.
- “Le Mans, Maine”: read as Maine. A US state’s name after a place name is read as that state, wherever the place is.
- “Charlottetown, PE”: no fact about a pay law. A province or territory code that is also a country’s code (NL, NU, PE, SK, YT) is read only when the posting also says Canada.
- “Downtown Cleveland, OH”: no fact about a pay law. A city with its own law is read only when no other word is joined to its name before it (other than “in”, “and” or “or”).
- “Salary: competitive”: no fact about a pay law. A pay line with no amount counts as pay stated, so no fact about a pay law is shown.
- “We raised $50 million last year.”: no fact about a pay law. Any amount of money in the posting stops the fact about a pay law, even one that isn’t pay: saying “no pay stated” wrongly would be worse than not saying it.
- “Location: On-site - Austin, TX”: read as a screening question (Location). A location line, including the one Tideline adds from the job board’s own fields, can be read as a screening question about where you are, as the same line pasted would be.
Matching by meaning: not used
Resume Check doesn’t match a requirement by what it means, only by the rules above. Before a small model for this was added, its scores were studied on the test postings: no level of closeness told the lines meant to match apart from lines that only share some words, so it would either almost never count or count as met a line that only shares a word with the requirement. So no model is used, and nothing is downloaded. The formula keeps a weight for it (3 quarters), so a model can be added later without changing the rest, but only once a study like this finds a level that works.
Data and licences
Each check’s receipt names the data it used, with these versions and the first 12 characters of each file’s SHA-256:
| Data | Version | SHA-256 |
|---|---|---|
| O*NET® 31.0 occupations, job titles and related occupations | O*NET® 31.0 Database | 6977767a193c |
| ONS SOC 2020 coding index (version 14) and unit groups | SOC 2020 coding index v14 (27 Aug 2026) | 4efc771e6683 |
| ONS ASHE Table 14.7a and 14.7b: gross annual pay of full-time employee jobs, by SOC 2020 unit group | ASHE 2026 provisional (22 Oct 2026) | 3fedfda32f96 |
| BLS OEWS May 2025: national and state wage percentiles | OEWS May 2025 | c088f68ea025 |
| Pay-in-postings laws | 1.0.0, checked 2026-11-09 | 72a148a1948f |
O*NET® 31.0 occupations, job titles and related occupations
Licence: Creative Commons Attribution 4.0 International (CC BY 4.0)
This page includes information from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Tideline Resume Check has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications.
What Tideline Resume Check changed: Tideline Resume Check keeps only each occupation's code and title, its job titles, its sample of reported titles and its related occupations, keys the titles as Resume Check's published title rules read them (folded plurals and spellings), and, from Software Skills, only the names written with their own acronym; descriptions, ratings and the other files are left out.
ONS SOC 2020 coding index (version 14) and unit groups
Licence: Open Government Licence v3.0
Source: Office for National Statistics. Contains public sector information licensed under the Open Government Licence v3.0.
What Tideline Resume Check changed: Tideline Resume Check keeps each title in natural word order as written, with each of its entries' SOC 2020 unit group, default marker and qualifiers, keyed by the title as Resume Check's published title rules read it, and each "see another indexing word" entry (}}}}) with its SEE text; the older classifications' codes are left out.
ONS ASHE Table 14.7a and 14.7b: gross annual pay of full-time employee jobs, by SOC 2020 unit group
Licence: Open Government Licence v3.0
Source: Office for National Statistics. Contains public sector information licensed under the Open Government Licence v3.0.
What Tideline Resume Check changed: Tideline Resume Check keeps, for each four-digit unit group, the full-time sheet's 10th, 25th, 50th (median), 75th and 90th percentiles in whole pounds and their coefficients of variation; the other sheets, including the male and female ones, and the other columns are left out.
BLS OEWS May 2025: national and state wage percentiles
Licence: Public domain (U.S. government work); BLS asks to be cited
Source: U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025.
What Tideline Resume Check changed: Tideline Resume Check keeps, nationally for each detailed and broad occupation and for each state and DC for each occupation BLS publishes there, the 10th, 25th, 50th (median), 75th and 90th percentile yearly wages, or the hourly ones where BLS publishes no yearly figure, with BLS’s “#” and “*” as BLS writes them, and each occupation’s title; the employment, mean and other columns, the territories, metropolitan areas and industries are left out.
How the rules are tested
The rules are checked against 14 job postings written for these tests, for made-up companies, each paired with one or more made-up people’s resumes (20 pairs). Every expected result was written by hand from what each posting was written to ask, never copied from Resume Check’s output, and each posting is read twice to show both readings are the same.
When these rules were published: 327 checks, 327 correct, 0 failed.
| What was checked | Checks | Correct | Failed |
|---|---|---|---|
| Headings that open a list: must | 11 | 11 | 0 |
| Headings that open a list: nice | 10 | 10 | 0 |
| Requirements found, by list: must | 65 | 65 | 0 |
| Requirements found, by list: nice | 18 | 18 | 0 |
| Screening questions found, by kind: degree | 2 | 2 | 0 |
| Screening questions found, by kind: language | 2 | 2 | 0 |
| Screening questions found, by kind: licence | 2 | 2 | 0 |
| Screening questions found, by kind: location | 4 | 4 | 0 |
| Screening questions found, by kind: notice period | 2 | 2 | 0 |
| Screening questions found, by kind: relocation | 2 | 2 | 0 |
| Screening questions found, by kind: salary | 2 | 2 | 0 |
| Screening questions found, by kind: work authorization | 2 | 2 | 0 |
| Screening questions found, by kind: years | 1 | 1 | 0 |
| Years of experience asked for: none | 6 | 6 | 0 |
| Years of experience asked for: years | 8 | 8 | 0 |
| Facts about the posting: pay range missing | 2 | 2 | 0 |
| Facts about the posting: posted date | 2 | 2 | 0 |
| Facts about the posting: repost wording | 2 | 2 | 0 |
| Lines to read side by side: contradiction | 1 | 1 | 0 |
| Lines to read side by side: entry level high years | 1 | 1 | 0 |
| Lines to read side by side: long must have list | 1 | 1 | 0 |
| The same result when read twice: twice | 14 | 14 | 0 |
| How each requirement is met: exact | 46 | 46 | 0 |
| How each requirement is met: acronym | 3 | 3 | 0 |
| How each requirement is met: synonym | 4 | 4 | 0 |
| How each requirement is met: partial | 7 | 7 | 0 |
| How each requirement is met: missing | 43 | 43 | 0 |
| Screening questions compared: met | 13 | 13 | 0 |
| Screening questions compared: failed | 2 | 2 | 0 |
| Screening questions compared: check | 9 | 9 | 0 |
| Years of experience compared: met | 8 | 8 | 0 |
| Years of experience compared: half | 1 | 1 | 0 |
| Years of experience compared: under | 1 | 1 | 0 |
| Years of experience compared: not required | 10 | 10 | 0 |
| Verdicts: strong | 3 | 3 | 0 |
| Verdicts: plausible | 6 | 6 | 0 |
| Verdicts: stretch | 3 | 3 | 0 |
| Verdicts: not competitive | 5 | 5 | 0 |
| Verdicts: no verdict | 3 | 3 | 0 |
A further 123 expected results are for matching by meaning, which isn’t used, so they aren’t checked or counted.
These counts are from Resume Check’s own test postings. They don’t measure how it does on real postings, and they aren’t a promise about yours.
Versions and the fingerprint
Every check’s receipt (“How this check was made, and how to reproduce it”) names the date it was made, the match pack, the data and a match fingerprint: a short code worked out from every rule, word list, table, formula and limit that decides a result, and from the version of Unicode (the standard for letters and symbols) the server’s software follows. The same resume, posting, answers and date, with the same fingerprint and data, give the same result.
- Match pack: 0.11.0
- Resume Check’s own table: 1.4.0
- Acronyms from O*NET® data: O*NET® 31.0 Database, built 2026-11-09
- Verdict formula: 1.3.0
- Pay-law table: 1.0.0
- Receipt format: 2; title keys:
occupation-keys-3 - Code: posting 1.1.0, map 1.0.0, fit 1.2.0, occupation 1.1.0, about 1.0.0
With these rules, a server that follows unicode-band:>=16.0 gives the match fingerprint 3e49cbb0f36f.
Changes to the rules
- Match pack 0.11.0, 9 November 2026: The US pay figures from BLS's May 2025 files: BLS's "*" and "#" in BLS's meaning, and the posting's state said when BLS publishes no figures for it (MT-PAY-PANEL 1.2.0)
- Match pack 0.10.0, 9 November 2026: The posting's facts and first warnings (5 rules: MT-POSTING-DATE, MT-PAY-LAW, MT-ENTRY-YEARS, MT-LONG-LIST, MT-CONTRADICTION; the pay-law table 1.0.0); one place reading for the pay panel and the pay-law fact (MT-PAY-PANEL 1.1.0); after the review, the company's own years and questions are no ask (MT-YEARS 1.2.0, MT-HEAD-OTHER 1.1.0)
- Match pack 0.9.0, 9 November 2026: Occupation, nearby roles and the pay panel (3 rules: MT-OCCUPATION, MT-NEARBY, MT-PAY-PANEL); the acronym rows from the Software Skills in O*NET® 31.0 data (MT-ACRONYM 1.1.0, the curated table 1.4.0)
- Match pack 0.8.0, 8 November 2026: A verdict needs a deciding fact; questions alone, all still to check, withhold it (MT-FIT 1.3.0, FIT_FORMULA 1.3.0)
- Match pack 0.7.0, 8 November 2026: No verdict only with truly no requirements; a skill's years below the posting's are short (never failed), weighed by the total-years rule in plain facts (MT-FIT 1.2.0, MT-KNOCKOUT-STATUS 1.2.0, FIT_FORMULA 1.2.0; FIT_SWITCHES removed)
- Match pack 0.6.0, 8 November 2026: Yes and no have a polarity (a bare one decides only a plain line), explicit right-to-work statements, statements are a yes only before an ability; unread or year-only dates never lower the verdict; a pay period on one side only, a posting's bare number is no pay; a skill opener governs its first clause; the skill years tied to the skill; places stand as themselves; FIT_SWITCHES for two decisions then still open (MT-ANSWER, MT-KNOCKOUT-STATUS, MT-RESUME-CONFIRMS, MT-EXPERIENCE, MT-FIT 1.1.0; MT-KNOCKOUT 1.2.0; the curated table unchanged)
- Match pack 0.5.0, 8 November 2026: Knockouts, years and the verdict (5 rules: MT-ANSWER, MT-RESUME-CONFIRMS, MT-KNOCKOUT-STATUS, MT-EXPERIENCE, MT-FIT); the question false alarms fixed (MT-KNOCKOUT, MT-YEARS, MT-PROSE, MT-CUE-MUST, MT-CUE-NICE, MT-NO-SECTION, MT-KEY-WORDS 1.1.0 or 1.2.0; the curated table 1.3.0)
- Match pack 0.4.0, 8 November 2026: Two-tool lines are skills lines, tech names are no addresses (MT-NAME-GUARD 1.2.0, the curated table 1.2.0)
- Match pack 0.3.0, 8 November 2026: Names, addresses, employers and personal lines never match; "or" lists; folded plurals and spellings; the curated table 1.1.0 (MT-KEY-WORDS, MT-NAME-GUARD, MT-LIST-ITEMS, MT-SYNONYM 1.1.0)
- Match pack 0.2.0, 8 November 2026: The match map, rules only (7 rules and the curated table 1.0.0)
- Match pack 0.1.0, 8 November 2026: Reading a posting (16 rules)