Somebody once told me Google has no predefined keyword density, so keywords no longer matter. Half of that is right. There is no magic keyword density percentage. But keywords, and the entities around them, still decide whether Google understands your page at all. The mistake is chasing a fixed count. The target should come from the pages that rank today, and that is what entity coverage measures.
KEY TAKEAWAYS
- Keywords still matter. Without them, Google cannot tell what the page is about.
- A fixed count (“use the keyword 8 times”) ignores what the ranking pages actually do.
- Entities are the people, places, things and concepts that belong with a topic.
- A fair target is the range the ranking pages use, per 1,000 words.
- Count an entity only when it shows up on two or more different websites, and leave alt text out.
Keyword density: keywords still matter, counts do not
Early search engines were simple. You typed a word, and the engine looked for the page that repeated it most.
“it would find the page that had the most times of that keyword on the page”
Beth Guide, Keywords and Why They Are Still Relevant (Internet Marketing Clinic, June 2021)
Search is nothing like that now. But that does not make keywords dead. If the keyword is missing from the places Google looks, it cannot get its hooks into the page, and it cannot understand what the page is about. If a site about Italy tours suddenly starts talking about vacations to Zimbabwe, Google does not know what to do with it.
So the keyword goes where it always went: the title tag, the H1, the URL, the headings and the copy. Keyword research in our SEO tools shows the volume, difficulty, intent and ranking pages, so you pick the right phrase before you write. What changed is what surrounds it.
WATCH THE CLASS
Keywords and Why They Are Still Relevant (Internet Marketing Clinic, June 2021). Beth Guide teaches this live, most months, with HCC and SCORE.
Keyword density vs entity coverage
An entity is something Google can identify: a person, a place, a product, an organization, a concept. Pages that rank for a topic tend to mention the same set of related entities, because those things belong in a complete answer. A page about lead tracking that never mentions GA4, forms or conversions is missing part of the topic, no matter how many times it repeats “lead tracking.”
Entity coverage is how much of that expected set your page covers, and how naturally. It is a better target than keyword density because it measures whether you covered the subject, not whether you repeated your target keyword. Synonyms and related phrases count toward the topic; a percentage does not.
Google also files every page into a category. The elev8tor content scanner runs your page and each ranking page through Google’s Natural Language categories, then groups entities by category, so you can see which part of the topic you are thin on.
Why a fixed keyword density is the wrong target
Many on-page checkers hand you a rule: use the keyword this many times, use this term at least twice. A fixed number ignores context. A 600 word page and a 2,400 word page should not use a term the same number of times. And the right amount for a medical topic is not the right amount for a roofing page.
Worse, chasing a keyword density percentage pushes people into keyword stuffing. They repeat a phrase until the checker turns green, and the page reads worse for the human it was written for. The scanner has an over-optimization check for exactly that.
How the elev8tor scanner sets entity targets
These are the rules the scanner follows, and they are a big part of why elev8tor scores differently from a generic checker.
Targets come from the ranking pages' range
For each entity, the scanner looks at how often the ranking pages use it per 1,000 words and takes the middle of that range, the 25th to the 75th percentile. For example, if the pages on page one use a term between 2 and 5 times per 1,000 words, that band is your target. Below it, you are thin. Far above it, you are probably stuffing.
An entity counts only when it shows up on 2+ different websites
If one ranking site mentions its own product name forty times, that is not a topic entity. It is that site’s brand. The scanner only counts an entity when it appears on two or more different websites in the set, which filters out one site’s quirks.
Image alt text is not counted
Alt text describes images for people who cannot see them. Counting it would reward stuffing keywords into image attributes, so the scanner leaves it out.
Questions count as coverage too
The scanner pulls questions from People Also Ask and the ranking pages and marks each one answered, partly answered or missing. Answering the real questions often closes entity gaps on its own, because good answers use the right terms naturally.
Entity coverage is the floor, originality is the ceiling
Covering the same entities as page one gets you into the conversation. It does not make you the best answer. If every page covers the same set, the one that adds something new stands out. That is why the scanner also shows terms beyond page one, the information gain you bring that the top 10 does not, plus an originality score. I cover that side in AI content for SEO.
Coverage on one page is not enough if the page sits alone. Related pages that link to it with keyword anchors carry the topic across your site. That is the idea behind website structure for AI search and the link suggestions in internal linking for AI search.
A quick keyword density and entity check for your page
- Put the keyword in the title tag, H1, URL and at least one H2.
- Read the top five results and list the things they all mention. That is your entity list.
- Check which of those your page skips, then add them where they genuinely belong.
- Answer the People Also Ask questions in plain sentences.
- Check keyword frequency paragraph by paragraph, not just the page total. If one paragraph repeats the phrase three times, rewrite it.
- Add one thing none of the top five say: a number from your own work, a step in your process, a mistake you see customers make.
Doing that by hand takes an afternoon per page. A keyword brief in elev8tor does the reading before you write, and the scanner checks the draft after. Both sit alongside the SEO audit in the same workspace, and the free demo includes 5 real scans so you can try it on a page of your own.
FAQ
Frequently asked questions
There is no fixed keyword density or percentage Google rewards. Keywords still belong in the title, H1, URL, headings and copy so Google can tell what the page is about, but the right amount depends on what the ranking pages do, not a universal percentage.
A person, place, organization, product or concept Google can identify. Pages that rank for a topic tend to share a set of related entities, and covering that set shows you answered the whole topic.
Use the range the ranking pages use. The elev8tor scanner takes the 25th to 75th percentile per 1,000 words across the ranking pages, so the target scales with your page length and your topic.
Alt text describes images for people who cannot see them. Counting it would reward stuffing keywords into image attributes, so it is excluded from entity counts.
There is no ideal keyword density for every page. Put the keyword in the title tag, meta description, H1 and opening paragraph, then follow the range the ranking pages use for your topic. The elev8tor analysis does that per 1,000 words.
Yes, if it crams them in. The scanner has an over-optimization check, and going far above the ranking pages’ range usually means the copy reads worse.
Check your own page against today's top 10
The free demo gives you 5 real Content Scanner scans on your own pages. No card, no sales call.
Related posts and pages
- Content Scanner: entities by category, questions and originality
- SEO action plan: ranked fixes instead of a data dump
- AI content for SEO: using ChatGPT, Claude and Gemini without hurting rankings
- SEO audit: titles, H tags and indexability with How to fix
elev8tor.ai is developed by Vertical Web and Beth Guide, drawing on more than 20 years of SEO work.
- Internal Linking for AI Search: The Exact Sentence and Anchor Text - October 23, 2026
- How to Prove SEO ROI with GA4 Lead Tracking - October 22, 2026
- Keyword Density Is the Wrong Target: Use Entity Coverage Instead - October 21, 2026