How to Find Keywords in a Job Description (Step-by-Step)

Every job posting tells you what the employer will screen for. The trouble is that it hides the signal inside paragraphs of mission statements, perks, and recycled requirements. If you pull out the wrong words, or the right words in the wrong priority, your resume ends up stuffed with filler while the actual must-haves stay buried.

This guide shows you how to find keywords in a job description step by step, both by hand and with tools. You'll learn which sections carry the most weight, how to rank what you find with a simple scoring method, how to handle acronyms and tool names, and how to map each keyword to real evidence from your work.

You'll also get a fully marked-up sample posting, a keyword-to-evidence template, an AI prompt you can paste, and a placement table showing where each type of keyword belongs on your resume.

What counts as a keyword in a job description#

A keyword is any word or phrase that an employer might search for, filter on, or use to judge fit. Most fall into six types. The table shows where each usually appears and how much it tends to matter.

Why these words matter to software and people#

Applicant tracking systems parse your resume into fields and let recruiters search and filter by terms like these. Some also rank candidates against the stated qualifications. Recruiters, in turn, skim for the same words to decide whether to keep reading. The keywords are where those two audiences overlap.

Exact wording matters more than many people expect. In research by Harvard Business School and Accenture, 88% of employers agreed that qualified high-skilled candidates were screened out because they didn't match the exact criteria in the job description, as HR Dive reported. For the broader picture, read our guide to resume keywords.

What keywords look like in different fields#

The six types hold across industries, but the specific words change. A few examples of what high-weight keywords tend to look like by field:

• Healthcare: licenses and certifications (RN, BLS, ACLS), unit types (ICU, med-surg), and EHR systems (Epic, Cerner)
• Finance and accounting: standards and regulations (GAAP, SOX), processes (month-end close, reconciliation), and systems (NetSuite, SAP)
• Software and data: languages (Python, SQL), platforms (AWS, Snowflake), and practices (CI/CD, data modeling)
• Education: certifications and grade levels, curriculum frameworks, and learning platforms (Google Classroom, Canvas)
• Skilled trades: licenses, equipment names, safety certifications (OSHA 10 or 30), and specific procedures
• Sales and customer roles: CRMs (Salesforce, HubSpot), sales motions (outbound, enterprise, SMB), and quota language

In every field, the most decisive keywords are concrete nouns: a license, a system, a regulation, a procedure. Those are what recruiters filter on and what qualification-based ranking checks.

What isn't a keyword#

Not every repeated word deserves a spot on your resume. Skip these:

• Company values and mission language ("we're obsessed with our customers")
• Benefits and perks
• Equal opportunity and legal statements
• Generic adjectives with no evidence attached ("dynamic," "self-starter," "rockstar")
• Location and schedule details, unless they're a requirement you need to confirm

Before you start: gather the right material#

Five minutes of setup makes the extraction faster and more accurate.

• The full posting. Open the original on the company's careers site if you can. Job board copies sometimes truncate the requirements.
• Three to five similar postings. Same title, similar company size and industry. You'll use them to separate core keywords from one employer's quirks.
• The company's careers or "about" page. It sometimes names products, customers, or tools the posting assumes you know.
• Your master resume. A long document with every role, project, and result. If you don't have one, see how to build a master resume.
• A blank document or spreadsheet with four columns: keyword, type, required or preferred, and your evidence.

How to find keywords in a job description: the 7-step method#

Here's the full process at a glance. The sections below walk through each step with examples.

1. Copy the full posting into a plain document.
2. Delete boilerplate: benefits, legal text, and company marketing.
3. First pass: extract everything from the required qualifications.
4. Second pass: extract skills and tools from the responsibilities.
5. Third pass: flag the job title, repeated terms, and anything in the opening lines.
6. Normalize each keyword to one exact form, including acronyms.
7. Rank the list and map each keyword to your evidence.

Step 1: Copy the posting into a plain document#

Pasting into a plain text editor or a blank document strips formatting and makes the posting easier to mark up. It also exposes how long the posting really is. A posting that looked short on a careers page may run 600 words once you see it all in one place.

Step 2: Delete the boilerplate#

Remove the benefits list, the equal opportunity statement, the generic "about us" paragraph, and the application instructions. What's left is usually 40% to 60% of the original. That's the part the hiring manager actually wrote or approved.

Be careful with "about us." Delete the marketing language, but keep any concrete detail such as product names, customer types, or regulated industries. Those can become domain keywords.

Step 3: First pass through the required qualifications#

This section has many names: "Requirements," "Qualifications," "What you'll bring," "You might be a fit if," or "Must have." Whatever it's called, it's where the employer lists what it will screen on.

Write down every:

• Years-of-experience phrase, with the function named ("3+ years in marketing operations")
• Hard skill and tool
• Degree, license, or certification
• Domain or industry term

Mark everything from this section as required, even if the wording is soft ("familiarity with").

Step 4: Second pass through the responsibilities#

The responsibilities section ("What you'll do," "In this role") describes the work. It's rich in skills that the requirements section takes for granted. If the requirements say "marketing automation experience" and the responsibilities say "build nurture workflows and manage lead scoring," those specific skills are keywords too.

Pull out the noun phrases, not the verbs. "Build and launch email campaigns" gives you "email campaigns." The verb is useful later when you write bullets, but recruiters search for nouns.

Step 5: Third pass for title, repetition, and position#

Now look at the posting as a whole:

• The job title and any title variants in the text ("specialist," "analyst," "manager")
• Repeated terms. Anything mentioned two or more times is central.
• The first paragraph and the first two bullets of each section. Hiring managers tend to put the most important items first.
• Bold or capitalized terms. Formatting is a deliberate signal.

Step 6: Normalize each keyword#

Employers name the same skill in different ways, and so do resumes. Before you add anything, decide on one exact form that matches the posting, then add the alternate form once.

Skills taxonomies can help you decide which name is standard. The Lightcast Skills Taxonomy lists more than 34,000 skills and is free to browse, which makes it handy for checking the common name of an unfamiliar skill.

Step 7: Rank and map#

The last step turns a raw list into a plan. You'll score each keyword, sort by score, and match the top items to proof from your experience. Both parts get their own sections below.

A marked-up example: one job description, start to finish#

Here's a sample posting written for this guide. It's generic on purpose, but its structure matches what you'll see on most careers sites.

What the three passes pull out#

After the first, second, and third passes, the extracted list looks like this:

What the example teaches#

HubSpot appears in every section, including the opening line. That's the clearest possible signal. Salesforce appears only twice, but both times in high-weight spots, so it's also a must-have. SQL shows up in the requirements section, but the phrase "is a plus" demotes it to preferred.

The soft skills at the end, "clear written communication and attention to detail," didn't make the table. They're real expectations, but you'll prove them with how your resume reads and with a bullet about documentation or reporting accuracy, not by listing them.

How to rank keywords by importance#

A list of 25 keywords isn't a plan. You need to know which 6 to 8 must appear and which you can let go. This scoring method makes the call consistent.

Score 6 or more and it's a must-have. 3 to 5 is important. 1 or 2 is optional.

The scoring method applied to the sample posting#

Here's how three keywords from the marketing operations example score, assuming the five-posting tally shown later in this guide:

• HubSpot: required (+3), opening paragraph (+2), responsibilities (+1), three extra mentions capped at +2, and in all five comparison postings (+2). Total: 10. Clear must-have.
• A/B testing: responsibilities only (+1) and in three of five comparison postings (+2). Total: 3. Important, worth a bullet if you have the experience.
• SQL: listed as "a plus" (+1 total) and in two of five comparison postings (no points). Total: 1. Optional; list it in skills if you've used it.

The scores aren't precise science. They force you to look at where a term appears instead of reacting to whichever word caught your eye first, which is the main reason people end up with a skills section full of filler.

Signals that something is required, even when it isn't labeled#

• Phrases like "must," "required," "minimum," "at least," or "you will need"
• Years of experience attached to the skill
• Licenses or legal requirements tied to the job, such as a driver's license for a field role
• Anything that also appears as an application question

Signals that something is softer than it looks#

• "Familiarity with," "exposure to," or "comfortable learning"
• A long laundry list of 15 tools, where no candidate could know all of them
• Requirements that look copied from an older version of the role

That last point is common. In the same Harvard Business School research, 72% of employers said they recycled existing job descriptions for middle-skill openings, and 38% used existing or slightly altered descriptions for high-skill roles, according to HBS Working Knowledge. Dated requirements happen. Comparing several postings helps you spot them.

Compare several postings to find the core keywords#

One posting tells you what one employer wrote. Three to five postings for the same title tell you what the role actually requires across the market. That's especially useful if you're applying to many similar jobs and want a strong base resume.

How to run a quick cross-posting tally#

1. Collect three to five postings with the same or similar title.
2. Run the three-pass extraction on each, or use a keyword finder tool.
3. Put all keywords in one list and count how many postings include each.
4. Mark anything that appears in at least three postings as core.
5. Make sure every core keyword you have appears in your master resume.

The chart below models what a tally might look like for five marketing operations postings. The counts are invented to show the method.

In this model, Marketo shows up in two postings. If you've used it, add it to your master resume. If you haven't, you don't need to chase it, because the role's core is elsewhere.

Build a keyword bank for your target role#

If you're applying to the same kind of role repeatedly, keep the tally going. Each time you extract keywords from a new posting, add them to a running list with a count. After 10 or 15 postings, you'll have a keyword bank that shows the role's true core, the common secondary skills, and the rare extras.

That bank pays off in three ways. Your master resume can cover every core term in advance, so tailoring becomes a matter of reordering instead of rewriting. You'll spot skill gaps worth closing with a course or a project. And you'll walk into interviews already knowing the vocabulary the field uses.

Look beyond the posting for hidden keywords#

Some of the most useful keywords never appear in the posting itself:

• Product and customer names on the company's website, which tell you the domain language the team uses daily
• Profiles of people who hold the role now, which often list the tools they actually use
• The team's engineering, design, or company blog, if there is one, for named platforms and methods
• Recent press releases for initiatives the new hire might support, such as a system migration or a new market

Treat these as context, not requirements. They help you choose which of your true skills to emphasize and give you specifics for your cover letter.

Use O*NET and skills taxonomies when the posting is thin#

Some postings are three sentences long. Others say "proficient in relevant software" without naming any. When that happens, borrow structure from public occupation data.

O*NET OnLine#

O*NET OnLine, run by the U.S. Department of Labor, publishes a profile for hundreds of occupations. Each profile lists typical tasks, software skills, knowledge areas, and core skills. Software flagged as a Hot Technology is one that appears frequently in employer job postings.

O*NET's Hot Technologies list ranks software by how many US job postings mentioned it in 2025, using job posting data from Lightcast. The top of the list shows how often everyday tools appear as requirements.

Two lessons from that data. First, basic tools are keywords too. If a posting names Excel, don't assume it's too obvious to list. Second, notice how O*NET lists SQL as "Structured query language SQL." Employers write tool names in different forms, which is why Step 6 matters.

How to use occupation data without overreaching#

• Use it to find likely keywords when the posting is vague, then confirm them against similar postings.
• Add only skills you've actually used.
• Never list a tool just because O*NET associates it with the occupation.
• Use the task lists for bullet ideas, rewritten with your own results.

Map each keyword to your experience#

Ranking tells you what matters. Mapping tells you whether you can claim it and where the proof lives. This is the step most guides skip, and it's the one that keeps your resume honest.

The keyword-to-evidence template#

Fill this in for every must-have and important keyword. It takes about ten minutes for a typical posting.

What to do when you don't have a keyword#

You'll rarely match everything. Decide case by case:

If you're changing careers, the evidence column will often come from projects, volunteer work, or adjacent tasks. Our guide to transferable skills shows how to frame them.

A mapping example for a career changer#

Say you're a high school teacher applying for instructional designer roles. The posting's must-haves are "instructional design," "Articulate Storyline," "needs analysis," "LMS administration," and "stakeholder management." Your mapping might look like this:

• Instructional design: you designed a semester-long unit sequence used by four other teachers. That's direct evidence; describe it in design terms.
• Needs analysis: you surveyed students and reviewed test data to rebuild a struggling unit. Name it as a needs analysis and include the result.
• LMS administration: you managed courses in Canvas or Google Classroom for 150 students. List the platform by name.
• Stakeholder management: you coordinated with parents, administrators, and a department team. Frame it with a concrete project.
• Articulate Storyline: you haven't used it. Complete a short project with a free trial before applying, then list it under projects with a date.

Four of five must-haves turn out to be real experience written in a different vocabulary. That's the most common situation for career changers, and keyword mapping is how you translate it.

Where to put the keywords on your resume#

Each section of your resume does a different job. Spread keywords so that software finds them and a recruiter sees proof.

A bullet before and after keyword mapping#

The after version places two must-have keywords (HubSpot and lead scoring), names the partner team from the posting, and gives a result. For more on writing bullets like this, see how to write resume bullet points. For grouping your skills list, see the skills section guide.

How many keywords is enough?#

There's no magic count. A practical target is to include every must-have you genuinely have, plus about half of the important and preferred items. For most postings, that's 10 to 15 keywords in total, with the top 6 to 8 appearing both in your skills section and in a bullet.

Tool-assisted methods: keyword finders, scanners, and AI#

Manual extraction teaches you how postings work. Tools make it fast once you know what you're looking for. Use them to speed up Steps 3 through 6, not to replace your judgment in Step 7.

Keyword finders and scanners#

A job description keyword finder pulls skills, tools, and qualifications from a pasted posting and shows which ones it emphasizes. A resume keyword scanner goes one step further and compares the posting with your resume, listing matched and missing terms.

These tools are good at:

• Catching tools and acronyms buried in long paragraphs
• Counting repetition across a long posting
• Showing your gaps side by side with the posting's wording

They're weaker at:

• Telling required from preferred when headings are unusual
• Recognizing that two different phrases mean the same skill
• Knowing which "missing" keywords you should leave out

An AI prompt you can paste#

General AI assistants are good at structured extraction if you constrain them. Harvard's career services office also suggests using generative AI to work in keywords from a job description, while treating it as an editing aid rather than the author, in its resume guide.

Check the AI's work#

Read the output against the posting. Look for three common errors: skills the posting never mentioned, preferred items labeled as required, and requirements dropped because they sat inside a long sentence. More ready-to-use prompts are in our ChatGPT resume prompts guide.

Mistakes to avoid when finding job description keywords#

Most keyword problems come from extracting too much or the wrong kind of thing. Watch for these:

• Treating every word as equal. A tool named once under "nice to have" doesn't deserve the same space as a requirement named four times.
• Copying sentences. Recruiters recognize their own posting. Lift the terms, not the prose.
• Listing soft skills as keywords. Show them through bullets instead.
• Ignoring the job title. Title alignment is one of the strongest signals for both search and ranking.
• Skipping acronyms. Write the full form and the abbreviation once each.
• Chasing every gap. Some missing keywords should stay missing because you don't have the skill.
• Using one keyword list for every job. Core terms carry over, but each posting has its own priorities.

A final check before you edit your resume#

• Boilerplate removed and the posting saved
• Required, responsibilities, and title passes complete
• Each keyword normalized to the posting's exact wording
• Every keyword scored and sorted into must-have, important, and optional
• Must-haves mapped to specific evidence and a metric
• Missing skills sorted into "add," "mention lightly," and "leave out"

Next steps: turn your keyword list into a tailored resume#

Finding keywords is the research half of tailoring. The writing half is putting them in the right places with proof. Here's how to finish the job:

• Update your headline and summary with the top must-haves.
• Reorder your skills section so the posting's priorities come first.
• Rewrite the two or three bullets that best match the must-haves.
• Check your result with a resume-to-job match tool and aim for a natural match rather than a perfect one.

If you want a target to aim for, read how closely your resume should match the job description and what an ATS score means. For the full tailoring routine, see how to tailor your resume in 30 minutes. The free job description keyword finder can run the extraction for you, but the ranking and mapping steps above are what make the result yours.