Keyword Density Is Not a Thing

Keyword density targets, TF-IDF scores, and "optimal" word counts are engineering folklore dressed up as SEO science. Here is what the evidence actually says, and how to build content that ranks without chasing metrics that Google stopped caring about years ago.

Where Did Keyword Density Come From?

In the late 1990s and early 2000s, search engines were crude. They counted how many times a word appeared on a page relative to total word count, and used that ratio to decide relevance. A 2% keyword density score for "blue running shoes" meant the page probably talked about blue running shoes. Simple, gameable, and eventually gamed to death.

By 2003, Google's Florida update had already begun penalising keyword stuffing. By 2011, Panda was penalising thin, repetitive content at scale. By 2015, RankBrain introduced machine learning into the ranking pipeline. By 2019, BERT was processing natural language at a transformer level, understanding context and intent rather than word frequency.

The keyword density advice you still see in blog posts and SEO checklists is citing a model that has not reflected Google's actual algorithm for roughly two decades.

What Does Google Actually Measure?

This is the more useful question, and the honest answer is: Google does not publish its ranking signals. What we have is a combination of confirmed statements from Google engineers, patent filings, and observable correlation data from large-scale studies.

Entity recognition, not keyword matching

Google uses the Knowledge Graph and entity disambiguation to understand what a page is about, not just what words it contains. A page about Django does not need to repeat "Django" 47 times. It needs to mention related concepts: WSGI, ORM, migrations, middleware, settings.py, Gunicorn, the difference between request.GET and request.POST. The presence of semantically related terms signals genuine topical depth.

This is closer to how a subject matter expert would read a document than how a keyword counter would.

Search intent alignment

Google classifies queries into intent categories: informational, navigational, commercial, transactional. If you rank for a commercial query with an informational page, you will underperform regardless of keyword frequency. The SERP itself tells you the expected format. Look at the top 10 results for your target query. If they are all comparison tables and product pages, a 2,000-word tutorial will not rank there, no matter how many times it repeats the keyword.

User engagement signals

Dwell time, pogo-sticking back to the SERP, scroll depth, and click-through rate all feed back into Google's quality assessments. A page that buries its answer in keyword-padded paragraphs will see users bounce faster than a page that answers the question in the first two sentences.

E-E-A-T and author signals

Since the Helpful Content System updates of 2022 and 2023, Google has leaned harder on Experience, Expertise, Authoritativeness, and Trustworthiness. This is not about keywords at all. It is about who wrote the content, whether they have demonstrated expertise, and whether external sources corroborate their authority.

So What Should You Actually Optimise For?

Topical coverage, not keyword repetition. The practical implication is this: write one page that covers a topic so completely that a reader does not need to go elsewhere. Then build supporting pages that cover adjacent subtopics, linking between them.

This is called a topic cluster model, and it reflects how Google's crawlers build an understanding of your site's authority on a subject.

Old thinking	Current reality
Repeat the keyword at 1–2% density	Use the keyword naturally; prioritise related entities
Target one keyword per page	Cover one topic per page; keywords follow from that
More words = better ranking	Match word count to query intent; do not pad
Meta keywords matter	Meta keywords have been ignored since 2009
Exact match anchor text	Descriptive, contextual anchor text

Specific things that do matter, with evidence behind them:

• Core Web Vitals: LCP under 2.5 seconds, FID under 100ms (now INP under 200ms as of March 2024), CLS under 0.1. These are confirmed ranking signals.
• Structured data: Schema.org markup, particularly Article, FAQPage, and HowTo, increases eligibility for rich results, which improves CTR.
• Internal linking: Pages that are well-linked internally are crawled and indexed more reliably. Orphaned pages underperform structurally.
• Page title and H1: The keyword should appear in the title tag and H1 once, naturally. Not four times with slight variations.
• First 100 words: Google parses the opening of a page heavily for relevance signals. State what the page covers early.

Does Word Count Matter at All?

Weakly, and only as a proxy for coverage.

A 300-word page on a complex technical topic probably does not cover it adequately. A 3,000-word page stuffed with padding does not outrank a 900-word page that answers the question precisely.

The Semrush 2023 ranking factors study found a correlation between longer content and higher rankings in informational queries, but correlation is not causation. Longer pages on informational topics tend to rank higher because they tend to cover the topic more completely, not because of the word count itself.

Write as many words as the topic requires. Cut everything else.

How Should You Actually Research Keywords Then?

Keyword research is still necessary. What changes is how you use the output.

Use tools like Ahrefs, Semrush, or Google Search Console to identify:

1. What terms your target audience uses to find topics in your domain
2. What search volume and difficulty look like for those terms
3. What intent category dominates the SERP for each term

Then write content that genuinely serves that intent. The keyword informs the topic. It does not dictate how many times you type it.

One useful discipline: write the page first, then check whether the keyword appears naturally. If it does not appear at all, you may be off-topic. If it appears 20 times in 800 words, you have a problem. Somewhere between those two is where real content lives.

Conclusion

Stop counting. Start covering.

Pick a topic, understand the intent behind how people search for it, write something that genuinely addresses it with depth and accuracy, and make sure the technical fundamentals (Core Web Vitals, structured data, internal linking, a clear title) are in place.

If you are auditing existing content, open Google Search Console, find pages with impressions but low CTR, check whether the title and meta description match search intent, and fix those first. That will move the needle faster than any keyword density adjustment ever will.

---

FAQ

Is there any keyword density percentage I should aim for? No. Google has not used keyword density as a ranking signal in any meaningful way since the mid-2000s. The 1–2% figure you see in older SEO guides has no basis in current algorithm behaviour. Write naturally, ensure the keyword appears in the title and opening, and focus on topical coverage.

Does repeating a keyword in image alt text help? Alt text should describe the image accurately. If the keyword is genuinely relevant to what the image shows, it belongs there. If you are adding it purely for keyword density, it does not help ranking and may be flagged as keyword stuffing. Descriptive, accurate alt text is also an accessibility requirement under WCAG 2.1.

If keyword density does not matter, why does my SEO tool flag it? Most SEO tools flag keyword density because it is easy to measure programmatically, not because it correlates with ranking. Tools like Yoast or Surfer SEO are scoring proxies that can be useful guardrails, but treat their keyword density scores as rough signals, not targets to optimise against.

How do I know if I have covered a topic well enough? Read the top 5 ranking pages for your target query. Identify what subtopics, questions, and related terms they cover. Make sure your page addresses all of those, plus anything they miss. Tools like AlsoAsked or the "People also ask" section in Google SERPs help surface related questions.

Can keyword stuffing actually hurt my rankings? Yes. Google's spam policies explicitly cover keyword stuffing, and pages that engage in it risk manual actions or algorithmic demotion. Beyond penalties, stuffed content degrades readability, which increases bounce rates, which itself feeds back negatively into quality signals.

Have a project in mind? Contact Sodio Technologies to discuss your requirements and explore the right technology solution for your business.