Resume keywords: how to find the ones a job description asks for

A resume keyword is a term a recruiter or a search would look for to decide whether you can do the job. Most of them fall into a few kinds:

• Hard skills and tools: Python, Salesforce, Figma, GAAP, Kubernetes.
• Methods and areas of work: A/B testing, financial modelling, incident response, user research.
• Credentials: a license, a certification, a degree, a clearance.
• The job title itself, and the common variants of it.

Soft skills (“communication”, “ownership”, “team player”) appear in almost every posting. They are rarely what a search filters on, and naming them on a resume proves nothing. Show them through a line of work instead: “Ran the weekly release review with four teams” says more about communication than the word does.

A posting usually has an introduction about the company, a description of the role, a list of what you will do, a list of what you need and sometimes a list of what would be a plus. The keywords that decide a screen are in the last two.

1. Copy out the requirements. Take the lists under headings such as Requirements, Qualifications, What you bring, Must have, Nice to have or Bonus.
2. Mark each item required or preferred. The heading tells you most of the time. Words such as “ideally”, “a plus” or “familiarity with” inside a required list usually mean preferred.
3. Pull the term out of each sentence. “Strong experience designing and operating distributed systems on AWS” holds two keywords: distributed systems and AWS.
4. Note the facts you cannot change. Years of experience, a degree, a work permit or a clearance are requirements, not keywords to work in. Either you have them or you do not.
5. Check the responsibilities too. A tool named only in “what you will do” still tells you what the team uses day to day.

Doing this for three or four postings for the same title shows you the terms that come back every time. Those are the keywords your resume should carry for that kind of job, whatever the employer.

A search and a hurried reader both look for the posting's own words. Use the short, common name of each term, and match the posting's spelling where there is a choice.

• If the posting says PostgreSQL and your resume says “Postgres”, use the posting's form.
• Spell out an abbreviation once if both forms are common: “search engine optimization (SEO)”. After that the short form is enough.
• Name the tool, not the category. “Built dashboards in Looker” beats “Built dashboards in a BI tool” when the posting asks for Looker.
• Keep to a skill's name. “Kafka” is a keyword; “experience with event-driven streaming platforms such as Kafka” is a sentence copied from a posting, and reads like one.

Synonyms matter less than they used to. Search inside some tracking systems still matches words exactly, while tools that compare meaning, like the one behind our resume match score, place “Postgres” and “PostgreSQL” together. Using the posting's form covers both.

A keyword can sit in three places, and each one does a different job.

In a line of experience

This is where a keyword earns trust. “Moved the billing service to Terraform-managed infrastructure on AWS” tells the reader where you used both tools and what for. Put each required keyword you have into the line whose work used it. If no line fits a required skill you really have, add a new line under the role where you used it most.

In the skills section

A short list grouped by kind (languages, frameworks, cloud, tools) lets a reader check the posting's list against yours in seconds. Every keyword that appears in your lines should appear here once, under its group.

In the headline

The line under your name can name the job you are applying for, as the posting writes it: “Senior Data Engineer”. Your past roles keep their own titles.

Bolding the technology names inside your lines helps a skimming reader, and lets you drop a separate “Tech stack” line under each role, which saves space for work that matters.

• Hidden text. Pasting the posting in white or tiny text is an old trick. A parser pulls it out as ordinary text, it shows up in the plain-text view a recruiter sees, and it ends the application when they notice.
• Keyword lists with no work behind them. A skills section with thirty tools and lines that mention none of them reads as padding.
• Repeating a term. Naming Python eight times does not make you more of a Python developer to a reader, and tools that compare meaning do not reward it either.
• Claiming what you do not have. A preferred skill you have never used is not worth a line. Every keyword on the page is something you can be asked about.

The test for each keyword is simple: could you talk for two minutes about the work where you used it? If yes, put it in that line. If not, leave it out.

Jobapplier does the reading for you, in two places.

For one posting

When a posting comes into our catalogue, JobBERT, a language model trained on job postings, reads its requirements and marks the skills it asks for, required and preferred. A posting in another language is put into English first, because the model reads English only. When you open the job, Jobapplier lists which of those skills your resume shows and which it lacks.

Press Tailor and the missing skills are placed the way this guide describes. A skill goes into the line whose work fits it, swapping that line's similar skill for it or added into the line. A required skill with no fitting line gets a new line under the role whose work fits it best. A preferred skill with no fitting line is left out, except a programming language, which goes into its closest line. Every added skill also goes into the skills list under its group, and is bold in its line. Employers, titles, dates, degrees and work authorization never change.

For your target titles

Your resume's health score includes a keyword check that looks past one posting. It learns the keywords most asked for in postings with your target titles, from up to 200 open postings in our catalogue, and tells you which of them your master resume does not show. That is the list worth working through once, before you tailor for any single job.

See resume tailoring for the levels and what each plan includes, or the guide on tailoring a resume to a job description for the whole routine.

Say a posting for a backend engineer requires Go, PostgreSQL and Kubernetes, and prefers Kafka and Terraform. Your resume shows Go and PostgreSQL. You ran services on Kubernetes at your last job but the lines say “containers”. You have never used Terraform. You used RabbitMQ, not Kafka.

• Kubernetes is required and you have it: name it in the line about running the services.
• Kafka is preferred and you have not used it: leave it out, and keep RabbitMQ where it is true.
• Terraform is preferred and you have not used it: leave it out.
• Add Kubernetes to the skills list under its group, once.
• Put “Backend Engineer”, the posting's title, in your headline.

The resume now shows every requirement you meet, in the posting's words, and nothing you would have to explain away.