If your content brief still has a column that says "keyword density: 1.5%," it is asking your writer to chase a number Google does not use. That ratio is not a Google ranking factor and there is no percentage to hit. That does not mean words stopped mattering. Relevant wording still helps Google understand what a page is about, but where those words sit, what the page actually answers, and whether the repetition turns into stuffing matter far more than the share of the page a word takes up. The evidence here is strong where it counts: Google's own guidance is direct, and a Google representative has said as much on the record. Call the density-target claim debunked. Call word matching as a relevance signal confirmed. The exact weight Google puts on repeated words inside its own systems stays unknown, and this piece keeps those three judgments separate instead of blurring them into one verdict.
What This Ratio Actually Measures, and Why the Myth Stuck
The density figure is an arithmetic idea, not a Google concept. You count how many times a word or phrase shows up on a page, divide by the total word count, multiply by 100, and you get a percentage. Five mentions in a 500 word page works out to 1%. Different tools count differently: some include headings, some count phrase variations, some don't. None of those counting choices reveals a target Google is checking your page against, because there isn't one.
The myth survives because there is a real idea buried inside it. Words on a page genuinely can help establish what that page is about. The leap that turns this into a myth is assuming a fixed proportion of words has to match the query, or that pushing that proportion up will reliably push your ranking up. Older SEO advice sometimes did recommend a target ratio, and it is worth naming that as historical advice rather than something Google currently endorses.
A 2011 Google Webmaster video, later recounted in a January 2017 Forbes piece, has Google's Matt Cutts rejecting the idea of a magic keyword-density number and warning that repeating a term brings diminishing value fast. More recently, and more directly, Google's John Mueller was asked on Reddit in late 2021 whether that ratio was still an SEO factor. Search Engine Roundtable reported his answer on January 7, 2022: no. That is a short, specific answer from someone who works on Google Search, not a policy document, but it lines up with everything else Google has published on the subject.
The Evidence for Keyword Matching, and What It Doesn't Prove
Google is not shy about saying that words matter. Its explanation of how ranking works says that content containing the same words as a search query provides a basic relevance signal, and that words appearing on a page, in headings or in body text, might indicate that the page is more relevant to the search. That is real evidence for using the language a reader would actually search with. It is not evidence for a density target. Nowhere in that explanation does a percentage appear.
Google Search Essentials, last updated December 10, 2025, tells site owners to put the words people use to find their content in prominent places, the title and the main heading among them. The SEO Starter Guide makes the same point with an example: an expert researching a topic might search for "charcuterie," while someone newer to the subject might search for "cheese board." Anticipating how different readers phrase the same question can help a page get found by more of them, without repeating either phrase until it stops reading naturally.
Term frequency is also a legitimate idea inside information retrieval more broadly, separate from anything Google has said. It explains why some words carry more weight in a search system than others do, a distinction covered in more depth in how semantic similarity decides what AI retrieves. That is useful background. It is not a disclosure of how Google's own ranking systems actually work, and it should not be treated as one.
The Evidence Against a Density Target
Everything Google has published about keywords points toward reader language and usefulness, not a formula. The SEO Starter Guide asks for natural, easy to follow writing and calls out excessive repetition, including repeating variations of the same phrase, as tiring for readers and against its own keyword stuffing policy. Google defines keyword stuffing as filling a page with keywords or numbers in an attempt to manipulate rankings, and its examples include strings of phone numbers with no added value and blocks of city or region names dropped in purely to rank for them. Notice what is missing from that policy: a percentage at which ordinary writing crosses into stuffing. Repetition on its own is not proof of manipulation, and a page should not be treated as if it violated a rule that Google never wrote.
A vendor study adds a data point, though a weaker one than Google's own statements. One SEO tool vendor, in an analysis updated in January 2026, looked at 1,536 Google search results across 32 competitive keywords and found no consistent correlation between that ratio and ranking position. That is worth knowing, and it fits the direction of everything Google says, but it is an observational study of one sample, not a controlled experiment, and it cannot prove that density has zero effect in every situation. Treat it as supporting context, placed well after Google's own guidance, not as the headline evidence.
What Google's Own Guidance Actually Says
Strip the myth away and Google's published direction on optimizing a page comes down to four things working together, none of them a word count.
Use the wording a reader would naturally reach for when they are trying to find or understand your subject. That is one of the seo fundamentals that still carry weight, and Google names it a basic relevance signal, not a reason to avoid a clear term in favor of a vaguer one.
Put that wording where it is descriptive and useful: an accurate title, a main heading that says what the page is about, and body copy that actually explains the thing rather than repeating its name. Search Essentials calls for prominent, descriptive placement. That is an instruction about location, not frequency.
Cover the reader's actual question, including the follow-up distinctions that belong on that page. This is where Google's people-first content guidance lands: answer what someone came to learn, completely enough that they do not have to go looking elsewhere for the missing piece.
Be useful beyond the word match itself. Google's own guide to how those ranking systems work says that after it identifies content as relevant, it looks at other signals too, including how helpful the content is and how trustworthy the source appears. No single editorial move, including fixing your keyword usage, decides a ranking by itself.
A comparison worth building into your content brief makes the difference concrete. "Our cloud backup pricing guide explains cloud backup pricing, cloud backup pricing plans and the best cloud backup pricing" repeats a phrase without saying anything. "Compare storage limits, retention periods and restoration costs before choosing a cloud backup plan" gives the reader something to act on. That second sentence is not a documented Google experiment, it is just what clearer writing looks like, and clearer writing is what the guidance above is actually asking for.
TF-IDF SEO: Useful Vocabulary, Poor Prescription
If a density target feels outdated, the next thing marketers often reach for is a term-weighting model, the idea that a content-optimization tool's frequency scores are the modern replacement. TF-IDF stands for term frequency, inverse document frequency: a term's presence within one document, weighted against how rare that term is across a wider collection of documents. Terms that show up in fewer documents get more distinguishing weight. A related model, BM25, builds on the same idea and also accounts for how long a document is, and it gives repeated mentions of a term diminishing returns rather than a straight line up.
These are real concepts from information retrieval, and a tool built on them can be a useful prompt: it can point out vocabulary or subtopics your competitors cover that your draft does not. What it cannot do is tell you Google's exact scoring formula, because the evidence available does not establish that Google runs a particular TF-IDF or BM25 implementation with a published target score for your page. Treat a tf-idf seo tool's word list as a checklist of questions to consider, not a set of words to insert. A term can be statistically unusual across a set of competing pages and still be irrelevant to what your reader actually needs, and inserting it anyway trades one arbitrary quota for another.
What to Check in a Brief or Draft Instead
Replace the density column in your content brief or review checklist with a short set of questions that get at what actually matters. Can a reader, and a search system, tell what this page is about from its title, heading and opening paragraph? Does it answer the question the title promises? Are the real distinctions explained, the ones a reader would otherwise have to go elsewhere to sort out, the same depth question that decides how much authority a topic earns? Does any sentence sound repeated purely to hit a target, rather than because the idea needed repeating?
Keep a natural, recognizable version of your primary phrase where it genuinely clarifies the subject, and let variations and necessary repetition happen without editing them out. Readers themselves have moved past typing rigid keyword strings into AI tools and search boxes alike, so a page written for a natural question tends to serve both. If a content-optimization tool flags a page for having "too few" or "too many" occurrences of a term, treat that as a prompt to ask what real problem it is pointing at, not an instruction to edit until the number changes. There is no Google-endorsed range to satisfy, and chasing one wastes review time that would be better spent checking whether the page actually answers the question.
Editorial review time is the resource this really costs teams. A marketing lead reviewing drafts against a density target is spending that time on a number that was never Google's, instead of on the coverage and clarity questions above, the same shift that separates a working keyword list from a real tracking system. Content tooling that already checks keyword coverage and structure against a brief during drafting, rather than as a separate pass after the fact, gives that reviewer time back for the calls that actually need editorial judgment: whether the piece answers the question, and whether it says anything a competitor's page does not already say.
The Verdict, and What Would Change It
That word-count ratio is not a ranking factor Google uses, and there is no percentage worth writing into a brief. Word matching, placement and topic coverage are the real, documented directions Google gives, and none of them come with a formula to hit. A brief built around a niche's real questions, the kind covered in how to own a niche instead of chasing keywords, holds up better under review than one built around a percentage. What would change this verdict is a new, explicit disclosure from Google describing a density-based scoring mechanism, or credible controlled evidence, not another observational study, showing that adjusting density on its own changes ranking. Neither exists in the evidence reviewed here.
DeepSmith's writing pipeline already builds keyword coverage, heading structure and internal linking into a draft as it is written, so a reviewer is checking the questions above rather than a percentage a tool made up. If that would save your team the hour a week it currently loses to manual keyword review, the 7-day free trial covers real drafts against your own briefs before you pay for anything.
