What does Keyword Density mean in SEO?
This is a hotly debated concept in the world of SEO. The general idea is that a page on the internet about "cars" should contain the word "cars" fairly often if search engines are to believe it is about "cars".
The keyword density calculation is a percentage of how many times that word appears on the page. So for example if a page contains 100 words and the word "cars" is mentioned 6 times then the keyword density for "cars" on that page is 6%.
Using keyword density has been widely discounted by many but it still has a useful roll to play in the initial stages of optimizing content. For example:
- content with a keyword density for a specific term of above 4% suggests over use of that term (known as Keyword Stuffing. It is a signal to search engines that the text will be difficult to read and so provide a poor user experience. Such content is unlikely to rank well.
- content with a keyword density of less than 0.5% suggests 'keyword lacking'. A search engine may be able to work out the phrase you are trying to rank for but you are making it harder for them. Why do that?
Both of the above provide basic clues on how to optimize content but are only a first step as search engines use a number of other signals to decide if a particular phrase is important to a phrase. I'll go through these now.
One: Latent Semantic Analysis
Search engines understand words with the same meaning. So they understand SEO and Search Engine Optimization are one and the same. If the keyword density of SEO in an article was 4% and the keyword density of Search Engine Optimization was also 4% then this is the same as using 'SEO' 8% of the time.
It would be clear to a search engine that the text is difficult to read and they would be reluctant to rank it if they had alternatives.
Latent Semantic Analysis also has a second role. The more alternative words that are used, the stronger the signal that the text is well written because it shows a wider use of vocabulary.
If I am writing about New York but also use 'the big apple' and NY the article will probably be easier to read and provide a richer user experience so you get brownie points for using alternatives. In other words keep a thesaurus handy when creating or editing your work!
Two: Keyword placement
Where is the phrase on the page. It will be given more weight if search engines find it:
- In the url
- in the title tag
- in the h1 tag content
- In the alt tag of the main image
- In subheadings
They'll also give a page more weight if there are links from other pages on your website which contain that keyword.
As Google's own SEO guide says, if you have a page about baseball cards then links to it from other pages of your website should contain the text 'Baseball cards'.
Three. Entity Salience
Breaking this down:
- Entity - a subject in the content (this could be a person, place, activity, business, etc.
- Salience - how important, prominent or dominant is that entity to the page?
To work this out search engines use Natural Language Processing (NLP)
which considers what words should also
be on the page if you were talking about a specific topic and doing it comprehensively.
The more of those related words that exist the more confidence they have that the content really is about the keyword phrase you are trying to rank for and is a quality piece.
So if this article was about SEO words that you would expect to find in regular use might include 'Search Engine', 'Google', 'ranking', etc.
Testing keyword density in modern SEO
Beyond the basic tests mentioned above to ensure content is not keyword stuffing or keyword lacking there is nothing more that a keyword density check can tell you. It is too blunt a tool in the modern world and search engines now think much more widely.
So after a basic keyword density check its about keyword placement (see above). Then I work far more with Entity Salience, using the right other words to persuade Google and other search engines that I am presenting quality content. See What does Entity Salience mean in SEO?
for more on how to do this.