Hard skills, soft skills, and job titles as keywords

Three kinds of match terms, not one keyword soup

Hard skills are named capabilities: tools, languages, methods, certifications. Soft skills are workplace traits: communication, leadership, collaboration. Job titles are the role labels — in the posting and on your timeline.

The distinction is lexical, not moral. Hard skills and titles are often rare enough that a search can treat them as a match. Soft skills are common English, so they are weak filters even when they are true of the work. Mix the three into one list and you will chase the ad’s marketing paragraph instead of the named tools and titles that distinguish the role.

• Hard skills: named tools, methods, licenses — match the posting’s wording when you have done the work.
• Soft skills: traits — show them in bullets, not as a second copy of the ad’s adjectives.
• Job titles: the role name as written — a phrase match, not a paraphrase of your internal grade.

Hard skills: the terms most likely to be searched

Hard skills are the terms a recruiter can type without drowning in false positives. “Kubernetes,” “GAAP,” “HubSpot” are specific. “Experienced professional” is not. When a posting repeats a tool in the requirements, that string is the name the hiring team already uses.

If the posting says “PostgreSQL” and your resume only says “relational databases,” a person might infer the overlap. A string match will not. The fix, when the work is real, is to use the posting’s name in a sentence you can defend — not a dump of every adjacent technology. You do not need every name in the stack; you need the ones that describe work you actually did, written the way the posting writes them.

Soft skills: real work, weak match terms

Soft skills are not fake; interviews fail on them constantly. As keywords they are poor discriminators. “Communication,” “team player,” and “fast-paced environment” appear in so many ads that matching them says almost nothing about this job.

Function words are stripped before a term list is built. What survives is often a handful of common nouns every candidate can paste, which raises overlap with boilerplate without adding evidence. Put the behavior in a bullet: who you coordinated, what changed. “Led a weekly stakeholder review that unblocked a launch” is evidence. “Excellent communication skills” is a claim.

Job titles: a phrase, not a translation

Titles are match terms because people search them. A posting titled “Senior Product Manager” will usually repeat “product manager” in the body. If your resume only says “Owner, Growth Squad” or “IC5,” the work may be identical and the string is not.

Keep the official title you held. Where it is true, you can add the posting’s title in the same line. Inflating a title is a credibility problem; omitting a true equivalent is a matching problem. Treat two-word titles as a unit: matching only “engineer” or only “analyst” is a noisier claim than matching “software engineer” or “data analyst.”

Why exact wording matters more than meaning

Keyword comparison on extracted text is not reading. It does not know that “JS” means JavaScript, that “PM” means product manager, or that “worked with internal teams” is stakeholder management. It looks for the term as a string after light normalization such as lowercasing and collapsing extra spaces.

A small wording change can move a term from missing to found without inventing experience. You are using the posting’s name for work you can discuss — not gaming a secret ranking. A resume that matches every string and cannot survive “tell me about a time” is still a worse outcome. Exact wording is a matching constraint, not a license to copy the ad.

Synonyms versus compact match

Synonym matching and punctuation-stripped matching get mixed up online. They are not the same.

A synonym matcher would treat “JavaScript” and “JS,” or “Amazon Web Services” and “AWS,” as equivalent. ResumeTune’s free check does not. It does not call a language model on the free path, and it does not keep a thesaurus. If you never wrote the posting’s string, that term is missing — even when a person would nod.

A compact match is mechanical. After a case-insensitive, whitespace-normalized check, ResumeTune also strips dots, spaces, slashes, and hyphens and compares the compacted strings, if the compacted term is at least three characters. That is why “Node.js” matches “nodejs” and “CI/CD” matches “cicd.” It is not why “Node.js” would match “JavaScript runtime,” or why “product manager” would match “PM.” Compact match is spelling variation of the same token, not conceptual equivalence.

• Exact-ish match: the posting’s term appears in your text after lowercasing and space-collapsing.
• Compact match (ResumeTune): the same letters with “.”, spaces, “/”, and “-” removed — “Node.js” ↔ “nodejs.”
• Not a match: a true synonym, acronym, or paraphrase you never wrote.

How ResumeTune builds the keyword list

ResumeTune extracts keywords from the job text you paste, not from your CV. The CV is only the haystack. Skip the job description and keyword matching is skipped: its weight is zero, and the remaining score is sections, format, and action verbs.

Stopwords are dropped. Remaining unigrams and bigrams are counted. A short curated list of known technical names — languages, frameworks, cloud products — gets extra weight when those strings appear in the posting, so they are more likely to make the comparison set. At most twenty terms are kept, ranked by that weighted frequency.

Each term is then looked up in your CV with the normalize-and-compact check above. The keyword component is found divided by total, times 100. With a job description that component is 50% of the overall score (sections 20%, format 15%, action verbs 15%). Without a job, the weights are 40 / 30 / 30 for those last three, and keywords do not contribute.

Reading “missing” through this taxonomy

A missing-term list is not an order. Sort it the same way you sorted the posting. Hard skills you have used: rewrite with the posting’s name. Hard skills you have not used: leave them off. Soft skills from the culture paragraph: usually ignore them, or put the behavior in a bullet. Titles: keep your real title; add a true equivalent only if it is honest.

Because the list is capped at twenty and frequency-ranked, boilerplate can sneak in and a skill you care about can miss the cut. That is a property of the extract, not a verdict on your career. Treat the list as a prompt about wording, not a ranking of fitness for the role.

What this does not tell you about an employer ATS

Employers do not share one keyword engine. Some search extracted text for exact phrases; some allow looser queries; some lean on recruiter habit. You cannot infer their formula from a public scanner, including this one.

ResumeTune reports overlap with the posting you pasted, using the rules in this article. It does not predict whether you will pass a vendor’s filter, get an interview, or “beat the ATS.” Those claims would require a system nobody outside that company can see. Use the taxonomy to know which kind of term you are looking at: match hard skills and titles in the posting’s language when the work is real, keep soft skills as evidence, and do not expect a synonym you never wrote to count as a hit.