Most resume advice about keywords is vague. "Include relevant keywords." "Match the job description." These instructions are not wrong — they're just incomplete. The part that's missing is the mechanics: which words, exactly, where in the resume, and why the placement matters.
Where ATS keywords come from
The keywords an ATS system looks for are not chosen by the software company. They come from the job description, which is why the job description is the only source you should use.
When a recruiter enters a resume search in Greenhouse, Workday, or Lever, they search using terms from the role they're filling. When the ATS ranks incoming resumes automatically, it does the same thing — it compares resumes against the language in the JD. This means the JD is both the map and the answer key.
How to read a job description for keywords
Not every word in a JD is a keyword. Here's how to separate signal from noise:
Hard requirements carry the most weight. Any skill, tool, language, platform, or certification in the "Requirements" or "Must have" section is a keyword the ATS will actively look for. If the JD says "5+ years of Python required," Python is a mandatory keyword. If it says "experience with Kubernetes," Kubernetes is a keyword.
Repeated terms are high-signal. If "SQL" appears four times across the JD — in the requirements, in the responsibilities, and in the nice-to-haves — SQL is a core keyword. The ATS is likely weighted toward it. Your resume should include it multiple times as well (in context, not as padding).
Job title and level matter. If the posting is for "Senior Data Analyst," those exact words — "Senior" and "Data Analyst" — should appear in your resume. Job title searches are common in ATS systems and in recruiter Boolean searches.
Soft skills are lower priority. "Strong communication skills" and "collaborative team player" appear in almost every JD and carry little ATS weight. Focus your keyword effort on technical skills, tools, domains, and certifications.
The exact-match problem
The single biggest mistake candidates make with ATS keywords is assuming that synonyms and paraphrases work. Many ATS systems perform string matching, not semantic matching. This means:
- "Machine learning" ≠ "ML" (unless the JD uses both)
- "JavaScript" ≠ "JS"
- "Project management" ≠ "program management"
- "Data analysis" ≠ "data analytics"
The fix: use the exact phrasing from the JD. If the JD uses "ML," use "ML." If it uses "machine learning," use "machine learning." If it uses both, include both in your resume.
Spell out acronyms on first use: "Natural Language Processing (NLP)" covers both variants. Then you can use either form throughout.
Where keywords carry the most weight
ATS systems typically score keyword placement. The hierarchy from highest to lowest ATS weight:
- Job title / role title — if you've held a title matching the posted role, that's the highest-weight match possible.
- Experience bullets — keywords in a bullet point with context ("Built recommendation engine using Python and collaborative filtering") score higher than bare keyword mentions.
- Summary section — the opening paragraph is read early and weights well. Include your 3–4 most important keywords here with one line of context.
- Skills section — this is where most candidates list keywords, but it's actually lower-weight than experience bullets. Still essential — just not sufficient on its own.
- Education / certifications — relevant degrees and certs (AWS Certified Solutions Architect, PMP, CFA) are keywords. List them in full.
How many keywords do you need?
There is no universal threshold, but a practical target is 70–80% keyword coverage of the core requirements section. For a JD with 15 listed requirements, you want at least 11–12 represented in your resume. Below 60%, most ATS systems will rank your resume low regardless of your actual qualifications.
The easiest way to measure this is to use a tool that compares your resume against a specific JD. Applyr's free ATS checker gives you a match percentage and lists the specific keywords you're missing — so you're not guessing which terms to add.
Common keyword mistakes
Keyword stuffing. Do not list 40 skills in your skills section, many of which you have only passing familiarity with. ATS systems may flag keyword density anomalies, and human reviewers will notice. Include keywords you can actually speak to in an interview.
Burying keywords in image or table cells. ATS parsers often skip tables and cannot read text in images. A resume that looks keyword-rich in a Word document can parse as nearly empty if those keywords are inside table cells or embedded in a logo.
Using only one section. Candidates who list technical skills only in the skills section miss keyword credit from experience bullets. A keyword that appears only once, in one section, scores lower than a keyword that appears in multiple sections with supporting context.
The workflow that works
- Open the job description. Highlight every specific skill, tool, technology, certification, and domain mentioned in the requirements and responsibilities.
- Check your resume against that list. For every highlighted term you're missing, decide: can you honestly add it?
- For terms you can add, insert them into an experience bullet where they're accurate — not just into the skills list.
- Check your coverage with Applyr's ATS checker — paste your updated resume and the JD, confirm your match score is above 70%.
- Submit.
This process takes 15–20 minutes per application. The alternative — submitting the same resume to every role — is faster and produces significantly worse results.