Resume keywords for ATS: how matching really works

There is no universal keyword list that makes a CV qualified. The useful keywords are in the posting in front of you, and the only ones worth adding are the ones your work already supports.

What an ATS actually reads

An applicant tracking system, or ATS, helps an employer manage applications. When discussing keywords, separate three questions: can the system extract your information, can a search find it, and does the information show you can do the job?

Parsing extracts information. A parser may turn the document into profile fields such as your name, employers and employment dates. Greenhouse describes this process in its resume parsing documentation. It also explains that some formatting can cause incomplete extraction. A successful parse means information was extracted; it does not establish your suitability.

Keyword search retrieves information. A recruiter can search for terms in an application. Depending on the feature, the search may inspect the attached resume, profile fields or application answers. The words you use therefore matter, but a search result only establishes a match under that search's rules.

Review evaluates the evidence. A bullet explaining how you used PostgreSQL gives a reader something to assess. A skills list containing PostgreSQL gives much less context. Prepare for a person to question the work behind each term, even when your first concern is automated screening.

This distinction helps diagnose a missing match. If your exported PDF loses a skills section when copied into plain text, fix the document structure. If the text is readable but says only “database work,” name the database you actually used. If you have never used the required database, wording cannot supply that experience.

For the document checks, use the separate ATS-friendly resume guide. Here, the task is choosing truthful words for readable content.

Required vs preferred vs nice-to-have

Treat these labels as an order for preparing your application. Address the requirements first, then the preferences, then the extras. This follows the employer's stated priorities. It does not tell you which filters or searches the employer will use.

Read the sentences around a term. “Experience maintaining Kubernetes deployments” asks for different evidence from “interest in learning Kubernetes.” “SQL required; PostgreSQL preferred” also differs from a requirement for PostgreSQL specifically. Preserve those distinctions in your notes.

Consider this fictional example. Amir is applying for a backend engineer role at the fictional company Elm Parcel. The posting requires Python, PostgreSQL and Kubernetes; it prefers incident response experience and Redis, with AWS listed as a bonus. His records support the following map:

The table exposes a useful difference: PostgreSQL is a wording gap, while Redis is an experience gap. The first can be fixed today. The second needs actual work or a project, if learning it is worth the effort for Amir's target roles.

Do not let a long list of matched extras distract you from a missing requirement. A mandatory license, a language requirement or substantial production experience needs its own honest assessment. Matching more words does not cancel it out.

Variants, acronyms and synonyms

Use the posting's wording where it accurately describes your work. When a common acronym is useful, include both forms naturally: “Maintained Kubernetes (k8s) deployment manifests.” This makes the meaning explicit without repeating a list of variants throughout the page.

There is no safe promise that every ATS will recognize equivalent terms. Greenhouse's Talent Filtering documentation describes exact keyword matching. Workable's candidate search documentation describes fuzzy matching by default and options for exact phrases. Those are examples of different search behavior, not evidence that either system recognizes every acronym or synonym.

The practical response is clear naming:

Avoid trying to anticipate every possible search query. Name the specific tool or practice, explain its use, and move on. For engineering examples, see resume keywords for software engineers.

Keep past job titles accurate, too. You can explain an unfamiliar internal title with a truthful description of the work. You should not replace it with a more senior target title just to create a match.

The honest gap: what to do when you lack a term

Before adding a missing keyword, find the source fact. Ask: where did I use it, what did I do, and what could I explain if questioned?

For Amir's PostgreSQL gap, the answer is already in his work notes. This rewrite uses only the fictional facts shown in the component:

Worked on database performance.

cv['backend engineer, elm parcel']

Investigated slow order-history queries in PostgreSQL with EXPLAIN ANALYZE and added a missing index.

The new bullet answers the posting more clearly. It does not claim that latency fell by an invented percentage. The action, tool and problem make it specific without a fabricated result.

Use a different response for each kind of gap:

• The experience exists but the term is missing. Add the accurate name to the relevant bullet and, if useful, the skills section.
• You used it only in an independent project. Describe the project and its setting. Do not imply commercial production experience.
• You studied it but have not used it. Describe completed training accurately under education or training. Keep it out of a list that implies practical proficiency.
• You have no evidence. Leave the skill out. Decide whether the remaining fit justifies an application or whether you need to build experience first.

For a larger move between fields, the career change resume example shows how to distinguish transferable work from new project experience.

Why stuffing and hidden text are poor fixes

Repeating “Python” across unrelated bullets adds no evidence of another Python task. A pasted job description adds the employer's requirements, not your experience. White text hides content from a reader while leaving your visible CV unable to support it. None of these tactics makes you better qualified. The same applies to a chatbot asked to "add the missing keywords"; see where ChatGPT resume prompts stop working.

Keep keywords in sentences that remain useful without a search engine: what you did, with which tool, for what purpose, and with what supported outcome. If you lack an outcome measurement, state the concrete work. Do not manufacture a metric to make the bullet look finished.

Checking coverage against the posting

You can do a useful check manually before using a scanner:

1. Save the exact posting. Work from its requirements and responsibilities, rather than a generic keyword list for the profession.
2. Group its terms by importance. Keep explicit requirements separate from preferences. Flag unclear wording instead of guessing.
3. Locate the evidence. For each term, point to an experience or project bullet. A relevant term may also belong in the skills section, but the bullet supplies context.
4. Edit only supported gaps. Name a tool already used, expand an acronym, or explain a responsibility. Keep experience gaps visible in your private notes.
5. Check the final file and application fields. Copy the PDF text into a plain document to inspect the reading order. If an application shows extracted fields, review those too. Read the resulting CV as a hiring manager would.

In cv[0], the keyword check separates required, preferred and nice-to-have terms from a posting and shows which your CV contains. It recognizes variants and acronyms such as k8s for Kubernetes. A free account includes 15 keyword checks a month; sign-up is required. Pro includes unlimited checks. Current plan details are on the pricing section.

That coverage is information for your review. It cannot tell you an employer's search settings or hiring decision. Treat a scanner's percentage, if one is offered elsewhere, as a tool's own calculation rather than an employer verdict. Read the terms and supporting bullets instead of pursuing a number.

Once the keyword map is sound, use the full tailoring workflow to choose which facts deserve space. Clear emphasis matters as much as including an accurate name.

To compare a real posting with your own history, start with your CV in cv[0].