Somewhere in the mid-2010s, a piece of advice lodged itself permanently in job-hunting folklore: work out which keywords the system is looking for and repeat them. It was reasonable guidance for the technology of the time. It is now actively counterproductive, and understanding why keyword stuffing your CV no longer works tells you a great deal about what to do instead.
Why It Used to Work
Early applicant tracking systems were built on document retrieval. They indexed your CV and matched terms against a search. More matches, higher rank. Simple, and entirely gameable.
And so people gamed it. Skills sections inflated from five items to twenty-five. Candidates pasted job adverts into the footer in white text. Whole paragraphs appeared listing every tool the applicant had heard of. For a while, on those systems, some of it genuinely worked.
What Changed, in Two Steps
Step One: Screening Started Reading Context
Modern systems do not count occurrences. They assess the text surrounding a term to judge whether the claim behind it is supported. The question moved from "does this document contain Python?" to "does this document show this person doing something with Python?"
Under that model a skills list is close to worthless, because a list contains no context by definition. Twenty unsupported mentions are worth roughly what one unsupported mention is worth, which is almost nothing.
Step Two: Generative AI Destroyed the Signal Entirely
This is the part most candidates have not adjusted to. When a keyword-perfect CV took real effort to produce, term matching was a weak but real signal, it at least indicated someone had read the advert carefully.
Now anyone can paste a job description into a chatbot and receive a CV containing every requirement, phrased fluently, in fifteen seconds. Everybody has the keywords. A signal that everyone can produce for free carries no information, so employers stopped relying on it.
Which means the effort you put into stuffing is competing in a category that no longer decides anything.
Why It Now Works Against You
Three concrete ways it costs you.
It displaces evidence. Your CV has finite space and a reader has finite attention. A dense block of terms occupies room that could have carried the one specific, scaled, outcome-bearing example that would have got you interviewed.
It creates a promise the document does not keep. Evidence-based screening notices when a skills section asserts capabilities the body never demonstrates. That mismatch is itself a negative signal. It is the exact pattern inflated CVs produce.
It reads badly to humans. A recruiter reaching a paragraph of comma-separated technologies forms an impression, and it is not a good one. Hidden white text is worse: the text is extracted regardless of colour, so the system sees it, and any human who selects the text sees it too.
What Replaced It
Context. Specifically, four things attached to each claim that matters: what you did, in what setting, at what scale, with what result.
Compare two candidates for a logistics coordinator role.
Candidate A has a skills block: "Route planning, inventory management, SAP, Excel, KPI reporting, stakeholder management, continuous improvement, health and safety, team leadership, problem solving, communication, time management."
Candidate B writes: "Coordinated inbound scheduling for a regional distribution centre handling around 180 deliveries a week. Rebuilt the booking slots in SAP after a third of vehicles were arriving outside their window and queueing on the access road; late arrivals fell to under 10% within two months."
Candidate A has listed twelve things. Candidate B has evidenced inventory scheduling, SAP, measurement, problem solving and operational improvement without naming any of them as a skill, and has given an interviewer something to ask about.
Under a keyword matcher, A might score higher. Under anything modern, and with any human, B wins comfortably.
The Legitimate Version of Keyword Work
Not all keyword attention is stuffing. There is a real practice worth keeping, and it is narrow.
Vocabulary alignment. If the advert says "stakeholder management" and your CV says "client liaison", change yours to match. If they say "safeguarding" and you wrote "child protection", align it. You are describing the same thing in their words, removing ambiguity for both the software and the reader. That is translation.
Naming hard requirements explicitly. If the role requires a specific registration, licence, certification or system and you hold it, state it plainly rather than implying it. Candidates lose roles constantly by assuming a requirement is obvious from context.
Using the full form and the acronym once. "Chartered Institute of Personnel and Development (CIPD)" covers both possible searches. Do it once, not repeatedly.
The distinction is simple: aligning language you can back up is legitimate, importing claims you cannot is not.
What to Do with the Space You Free up
Cutting a bloated skills section usually recovers a third of a page. Spend it on:
- One properly evidenced example per critical requirement. Three or four total.
- Numbers. One per bullet wherever a real figure exists.
- The difficulty. Briefly, what made it hard or what you tried first that failed. Almost nobody includes this and it is the strongest authenticity signal available.
- A short, honest skills list. Six to ten items, every one of which the CV goes on to demonstrate.
Frequently Asked Questions
Does Keyword Stuffing Still Work on Any System?
It may marginally raise a score on an old keyword-matching setup or a third-party checker. It does nothing on context-aware screening, and it costs you with every human reader. Given you cannot tell which system an employer runs, the safe strategy is the one that works on both, evidence.
Should I Still Match My CV to the Job Advert?
Yes, but match the substance, not the vocabulary density. Identify the requirements that matter, evidence those specifically, and align your terminology where it accurately describes what you did.
Is a Skills Section Still Worth Having?
Yes, kept short and honest, as a quick index for a human skimming. Six to ten items, all demonstrated elsewhere in the document. What fails is a long list standing in for evidence.
What About White or Hidden Text?
Never. Text is extracted regardless of colour, so the system reads it and reads it as deliberate manipulation. Anyone who selects the text sees it immediately. The risk is total and the upside is zero.
How Many Times Should a Key Term Appear?
As often as it naturally occurs in describing what you actually did, typically once or twice. If you are counting occurrences, you are optimising the wrong variable.
The Shift in One Sentence
Screening stopped asking whether you used the right word and started asking whether you did the thing. Keyword stuffing answers a question nobody is asking any more.
CVSense evaluates the context, scale and outcome around a claim rather than its frequency, which means the candidate who describes one real problem properly outranks the one who lists twenty capabilities. That is a fairer contest, and it rewards the part of your CV that was always the most useful.
Sources
National Careers Service. CV Sections and Advice.
https://nationalcareers.service.gov.uk/careers-advice/cv-sections
Chartered Institute of Personnel and Development.
https://www.cipd.org/uk/
InsightCircle
