ATS Keywords: How to Find and Use the Right Ones (2026 Guide)
ATS keywords are search terms in queries recruiters run. Built from Greenhouse, iCIMS and Workday search documentation, with a worked example: a job advert extract and the keyword list a candidate would take from it.
The advice to "use keywords" is everywhere and explains nothing. It leaves you guessing at a hidden scoring machine, and that machine does not exist in the vendor documentation.
What does exist is a search box. Greenhouse, iCIMS and Workday all document how recruiters query stored applications, and the queries are ordinary: a phrase in quotation marks, two terms joined by AND, a job title typed into a filter. A keyword is worth having on your CV when it is a word a person searching for someone like you would type. That is the entire mechanism, and it changes what you should do about it.
This article works from those search pages, then does the thing most keyword articles skip: takes one job advert and shows the list a candidate would actually derive from it. For how the surrounding process works, see what an applicant tracking system does with your application.
What a keyword is, mechanically
Greenhouse's talent filtering lets a recruiter search "job titles, skills, locations, and other keywords in the full text of resumes and internal notes". It offers two settings, and the difference between them is the whole of keyword strategy:
- Preferred keywords: "only one term must appear in a candidate's application to be included in the search results (OR statements)".
- Required keywords: "all terms must appear in a candidate's application to be included in the search results (AND statements)".
A recruiter writing their own query gets more: Greenhouse documents AND, which "limits your search because you're requiring that any search term following AND must appear in the results"; OR, which "provides options in a search so that the results will include either or both terms"; NOT, "the command of exclusion"; quotation marks to "capture a phrase that is to be kept intact in the exact word order stated"; parentheses to "group clauses to form sub queries"; and an asterisk wildcard to "represent any number of characters". Its own examples are (MIT OR Harvard) AND "Computer Science" and Java*.
iCIMS documents the same logic through its search API, where filters carry operators for exact match and "contains" and filter groups combine with AND and OR. Workday gives recruiters search prefixes such as cand: for candidates, and — the detail worth knowing before you apply through one — its resume parsing does not auto-fill the Skills or Languages fields, so those get typed in.
Four things follow from that, and they contradict most of what gets written about keywords.
Four consequences, including the end of "keyword density"
Presence beats frequency. An AND query asks whether a term appears, not how often. No vendor page I have read describes counting occurrences or scoring density, and the advice to hit a percentage has no documented basis. Write the term once, in a sentence that earns it.
Exact phrases are exact. Quotation marks keep a phrase "intact in the exact word order stated". If the advert says "stakeholder management" and your CV says "managing stakeholders", a quoted search misses you. This is the strongest argument for lifting the employer's phrasing verbatim, and it applies to phrases far more than to single words.
Wildcards stretch stems, not meanings. Java* catches Java and JavaScript. Nothing in the documentation turns "SEO" into "search engine optimisation", which is why you write both forms once and stop worrying about it.
Your words are the index. The recruiter searches what the parser stored. A skill described only in a graphic, a rating dot or a heading the parser mangled is not text, so it is not findable — the format guide covers what breaks that step.
Worked example: one advert, one keyword list
This advert is invented. The employer, role and details are made up, and the wording is written to look like a real UK posting.
Senior Data Analyst Northgate Insurance Services Ltd — Manchester (hybrid, 2 days on site) £48,000 – £54,000 About the role Reporting to the Head of Data, you will own reporting for our claims function and work with underwriting and finance stakeholders to turn operational questions into analysis they can act on. Responsibilities • Build and maintain Power BI dashboards for the claims leadership team • Write and optimise SQL queries against our data warehouse • Own data quality checks and document data definitions • Present findings to non-technical stakeholders across underwriting and finance • Support the migration of legacy reporting into Power BI Essential • Advanced SQL • Demonstrable experience building Power BI dashboards • Experience of stakeholder management in a commercial environment • Degree in a numerate discipline, or equivalent experience • Strong attention to data quality Desirable • Python for data analysis • Experience in insurance or wider financial services • Familiarity with Agile delivery
Read it twice, then write down every term that names a thing: a tool, a technique, a qualification, a sector, a job title. Ignore the adjectives. The result is not a list of 500 industry words — it is about fifteen terms, most of them printed on the page.
| Term, as the advert writes it | Why it is a search term | Where I would put it |
|---|---|---|
| Senior Data Analyst | The job title; the most likely single thing typed into a filter | Your own job title if it is truthful, otherwise the profile line |
| Power BI | Named tool, appears three times — an AND term in any serious query | Skills, and a bullet showing what you built with it |
| SQL (and "advanced SQL") | Named language, in the essential list | Skills, plus a bullet that shows the level rather than claiming "advanced" |
| data warehouse | Names the environment; distinguishes you from spreadsheet-only analysts | An experience bullet |
| data quality | Repeated in responsibilities and essentials, so it matters to this employer | An experience bullet, with what you actually did |
| stakeholder management | A quoted phrase search matches this word order and not "managing stakeholders" | Profile and one bullet naming the stakeholders |
| claims / underwriting | Domain vocabulary; rare enough to be a strong filter | Experience bullets, if you have worked on either |
| Python | Desirable, named, cheap to include if true | Skills |
| insurance / financial services | Sector terms recruiters use to narrow a long list | Employer descriptions in your experience section |
| Agile | Desirable and commonly searched | Skills or a delivery bullet |
What did not make the list: "strong attention", "demonstrable experience", "non-technical", "commercial environment". Nobody searches for those. They are the advert's connective tissue, and copying them into your CV adds length without adding a single match.
Two entries deserve a note. "Degree in a numerate discipline, or equivalent experience" is a requirement rather than a keyword: write your actual degree, in full, and let the recruiter judge. And "advanced SQL" is a self-assessment. A line reading "rewrote the claims reporting queries, cutting the monthly refresh from four hours to twenty minutes" carries the term and the evidence at once.
Here is one bullet built from the list, showing how three terms land in a single true sentence: "Built and maintained 12 Power BI dashboards for the claims leadership team, sourced from SQL queries against the group data warehouse." That sentence is findable under Power BI, SQL, claims and data warehouse, and it still reads like something a person wrote about their own job.
Job titles are the keyword people underestimate
Greenhouse lists job titles first among the things recruiters search for. Jobscan's State of the Job Search puts numbers to it from its own data: 55.3% of the 384 recruiters it surveyed in February and March 2025 said they use job titles as a keyword to filter applicants, and across more than 2.5 million applications made through Jobscan's tools, candidates whose CV titles matched the target title "had an interview rate 10.6 times higher than those who did not". That second figure describes users of a CV-optimisation product rather than a controlled experiment, so treat the size of it with caution and the direction of it as consistent with what the vendors document.
The practical move is modest and honest. If your internal title was "Reporting Lead" and the work was a senior data analyst's work, write Reporting Lead (Senior Data Analyst). You have not claimed a promotion you did not get, and both terms are now in the stored text. What you should not do is relabel a junior role as a senior one; the interview will settle that quickly.
Abbreviations, synonyms and the honest limit
Write the full term and the abbreviation together once — "Search Engine Optimisation (SEO)", "Chartered Institute of Management Accountants (CIMA)" — and never again. Nothing in the search documentation bridges the two forms for you, and the wildcard only extends a stem.
For genuine synonyms, pick the advert's term for the CV you send to that advert. That is the point of tailoring per application: you are not maintaining one document that pleases every search, you are sending the version that matches this one.
The limit worth stating plainly: none of this makes an unqualified application succeed. Keywords decide whether a qualified person is found in a list. They do not decide whether that person is hired, and a CV stuffed with terms you cannot discuss will fail at the interview that the keywords won you. If you are moving between fields and the vocabulary gap is the real problem, the career change CV guide handles the reframing.
Sources
- Greenhouse Support: Talent Filtering (preferred and required keywords)
- Greenhouse Support: Search candidates using Boolean queries
- iCIMS Developer Resources: Search API (filters and operators)
- Workday Administrator Guide: Concept: Resume Parsing (Skills and Languages not auto-filled)
- Workday Administrator Guide: Reference: Search Prefixes for Recruiting
- GOV.UK JobHelp: Tailoring your CV
- Jobscan: State of the Job Search (job title figures and survey method)
Frequently Asked Questions
What are ATS keywords?
They are the terms a recruiter types when searching the stored text of applications. Greenhouse's talent filtering searches 'job titles, skills, locations, and other keywords in the full text of resumes and internal notes', and its boolean search supports AND, OR, NOT, exact phrases in quotation marks, parentheses and a wildcard. A keyword is useful on your CV to the extent that it is a word someone would search for.
How many times should a keyword appear on my CV?
Once is enough for the search to find it. Greenhouse's required keywords are AND statements — every term must appear in the application — which is a test of presence, not of frequency, and no vendor documentation I have found describes counting repetitions or scoring density. Repeat a term only where a second mention genuinely describes different work.
Should I use the exact wording from the job advert?
Yes, for the terms that name a skill, tool, qualification or job title. An exact phrase search in quotation marks matches the phrase as written, and iCIMS filters match a field exactly or by 'contains' depending on the field. Paraphrasing a named tool or credential is the one place where creative wording costs you.
Should I write the abbreviation or the full term?
Both, together, the first time: 'Search Engine Optimisation (SEO)'. You cannot see which form the recruiter will search, and Greenhouse's wildcard only extends a word stem rather than translating between an abbreviation and its expansion. Writing both costs three words.
Do soft skills work as keywords?
Only when the advert names them as a phrase, such as 'stakeholder management'. A generic claim to be a good communicator is not a search term and takes space a specific one could use. My order of priority: job title, named tools and qualifications, the functional phrases the advert repeats, then everything else.
Does keyword stuffing get detected?
No vendor documentation I have found claims to detect it, and white text pasted at the end of a CV will be extracted like any other text. The reason not to do it is the person who reads the document afterwards: a CV that reads as a word list fails the stage that actually decides whether you are interviewed, and a claim you cannot discuss in an interview is worse than a missing keyword.
Where do keywords matter most on a UK CV?
In your job titles, your skills section and the bullets under your two most recent roles. Workday's parser does not auto-fill the Skills field at all, so on those applications you often type your skills into the form yourself — which makes the words you have chosen the words that get stored.
Keep reading
What Is ATS? How Applicant Tracking Systems Work (2026 Guide)
What an applicant tracking system does with your CV, step by step, from Workday, Greenhouse, iCIMS and Oracle Taleo documentation — including the difference between a CV the software could not read and an application an employer's own rules excluded.
Read articleHow to Tailor Your Resume to a Job Description (Step-by-Step)
Learn how to tailor your resume to any job description. A step-by-step guide to getting past ATS filters and landing more interviews.
Read articleATS-Friendly Resume Format: Complete Checklist for 2026
An ATS-friendly resume format checked against what Greenhouse, Workday, iCIMS and Oracle Taleo document: accepted file types, size limits, what breaks parsing, which popular rules are myths, and a before-and-after example.
Read article