If you’re using an analyzer to optimize web pages for keyword density and you should, it’s important to measure keyword density to see if the right figure’s achieved. When using a keyword
density analyzer how do you know if the analyzer’s giving the right density & what’s the right density anyway ?
I set about trying 5 analyzer’s on 5 different pages ranked from position 1 to 30 on Google.com. I wanted to see how important keyword density was to ranking order and find the right figure.
If you don’t know what keyword density is, it’s the number of keyword’s divided by the word count on the page expressed as a percentage. So, 20 keyword’s on a page with a total word count of 500 should be 20/500 or a keyword density of 4%. As we’ll see,
it’s not quite so simple.
A keyword density of between 2% to 20% for a page is often quoted as a target figure. Personally I think at 20% you would be regarded by the search engine as spamming (to much repetition)
for which you could find your site penalised. It would also not make very interesting reading.
It might seem obvious to say, but there are only two ways to get higher keyword density. One is more frequent use of the keyword or phrase, the other is having fewer words.
It’s only when using an analyzer to look at keyword density, you start to realise how difficult & time consuming it can be to reach keyword density figures of 4 to 5% never mind anything
higher. Reducing the words on a page often seems like the easier answer to improving density.
Some optimizer’s believe that some search engines look at word count as well as density. It then adds a weighting factor against a page with a low count, compared to say another page with the same density but a higher word count.
But is this true, or just a theory and what’s an acceptable word count ?
A well known and respected analyzer that many optimizer’s use, for Google.com recommended a total word count of 505 to 795 with 436 to 629 words in the body text.
I carried out an experiment on Google.com running a query on the
exact phrase “keyword density” and then used a keyword density analyzer to determine the density & word count for chosen pages in the Google rankings.
Each page was stored to the hard drive to ensure the software was analyzing the same page each time. For the exact phrase “keyword density” there were 55,100 other competing pages on Google.
Keyword Density on the total page
The table this section refers to is at:
In the keyword density column the 1st figure is for “keyword” and in brackets for “density”. Only position 30 complies with the analyzers recommended total word count!
From the keyword density analyzer results it’s clear the No.1 position is not being significantly penalised for only having a word count of 161 and neither is No.10 with 102 words when
compared to the others.
Interestingly the No.10 & 19 positions have almost identical keyword density, still with low word count, but the one with lower count gets the higher ranking.
We know keyword density is not the only factor to determine page rank. What about keywords in the Meta Tags and link text and the main body content ?
For Google the keyword and description Meta Tags are supposed to be ignored. However a recent report by Jon Ricerca of www.searchenginegeek.com entitled “Does the Keywords Metatag
Affect SE Ranking?” would seem to suggest that might not be true.
The Meta Tag’s don’t seem to matter since the No.1 posn does not have them!
Next, a keyword in the visible link text and 2nd in the actual link URL to a page, the href part.
From the Keyword Density Analyzer, Posn 1 has keyword’s in the link text, but so do No’s 19 & 30. No’s 4 & 10 don’t have keyword’s in the link text.
Posn 1 also has a keyword in the link, but again so do No’s 19 & 30 and both 4 & 10 don’t.
Also on positions 10 & 19, No.10 beats 19 even though it has no keywords in the link text or link.
So far, No.10 beats No.19 even though they have almost the same keyword density. No 10 has a lower word count and no keywords in either the link text or link compared to No 19 which has both.
So what’s the missing factor/s that explain why No 19 has a lower ranking than No 10 when by the optimizer rules it should be the other way around.
It’s nothing to do with keywords in the title either, because they all have those.
What about keywords in the Alt tags, no its’ not that either, the only one that has those is No.4.
I do hope your following this so far !
Perhaps it’s due to keywords in the body content, but No.19 has more than No.10, so it’s not that.
Let’s not forget No.4 it has a lower keyword density than 10 or 19 but still manages to beat them both.
The only thing in the data so far that sets No.4 apart from the others is a higher keyword count in the body content. But that’s not likely to help since it also has a much higher body word count.
Keyword density in the body content
The table this section refers to is at:
*The SEO Analyzer did not return results, so word count was
calculated from the keyword density SEO did return & knowing the body content keyword count.
In the Body word count, none of these pages comply with the analyzer recommended figure of 436 to 629.
Now although the keyword density in the body content of No.10 is slightly higher than 19 it’s also higher than No.1, but it doesn’t get ranked No.1 because of it.
No.4 has a lower body keyword density than all except No.30, yet it still gets ranked No.4 !
Another factor to consider is keyword prominence or how close keywords are to the beginning of either the title, the link text or in the main page content.
The analyzer reported a minor keyword prominence problem with No 19 in the link text and body content and more of a problem with prominence of No.30 in the link text.
But to keep these prominence reports in perspective for No 19 & 30 we must not forget that No.4 and 10 have zero keyword prominence in the link text because they don’t have any link text !
Online Analyzer Keyword Density
One of the most detailed online keyword density analyzer’s is that from www.ranks.nl
The table this section refers to is at:
The above results show the keyword density & prominence for the words “keyword” & “density” as separate words and also as a two word phrase “keyword density”.
It seems a higher keyword density coupled with a higher prominence goes some way to explain the ranking order. The exception to this No.4 whose ranking is not due to keyword
density or prominence since on both counts it’s only slightly different to No.30.
Two other things are striking about the results from all analyzer’s, firstly the very wide variation in word count and secondly the variation in keyword count.
The word count by analyzer’s varies by as much as 81.3% (No.10), 44% for No.19 & 41.9% for No.30.
In the worst extreme (No.10) the keyword density calculated by the analyzer’s varies by 154.4%
The keyword counts for No.10 & 30 show the most variation. The keyword density software counts 3 occurrences of the word “keyword” and 3 of “density”. The online analyzer counts 8 and 5 respectively.
For position 30 the software counts 13 for “keyword” and 9 for “density” whilst the online tool counts 8 and 6
respectively. In the worst case this is a keyword count variation of 62.5%.
The software was set to look for keywords in the Title, Link text, Links, Header tags (h1-h6), Meta Keywords & Description tags, Body text & Alt tags. In fact the only place
it was not set to look are the comment tags, but then neither was the online analyzer.
Both the software and the Ranks online analyzer can exclude words shorter than a defined setting. In all keyword
density analyzer’s where the option was available it was set to ignore words of 2 letters or less.
To another online analyzer position No.4 had only 10 keywords and a body word count of 969 both of which differ from the figures already seen.
Can we use any tool to compare one page with another. Given that none produce consistently high word & keyword counts or consistent low ones, a relative analysis of pages is also questionable.
Keyword Density Analyzer Conclusions
Three software tools and two online tools have been shown to all produce widely differing results for word count (variation 81.3%), keyword count (variation 62.5%) and as a result wide variations in keyword density (variation 154.4%). This whilst analyzing 5 different pages with Google rankings from No.1 position to No.30.
It seems likely with 5 different tools all producing different results that Google itself also arrives at yet other figures for word count, keyword count, prominence & keyword density.
Based on the tests, these tools don’t produce absolute results or consistent relative comparisons between pages. The
differences in word count and keyword count between these analyzer’s could not be explained by a word exclusion filter.
The rankings of the test pages can not be explained just on the basis of keyword density, prominence, word count, keywords in the meta tags, headers, links or Alt tags.
This should come as no surprise, since we know other factors like on & off site links and link popularity all play a role.
What may come as a surprise is that the analyzer tools we rely upon give such widely differing results. Does this mean we should stop using them, no I don’t think so, at least not until
something better comes along.
Andy Theekson http://www.the-search-engine-optimizer.com provides free information on SEO for small web businesses and conducts independent research on software tools and services. Andy has been involved in small business start-ups both offline and online since 1988.