Welcome to WebmasterWorld Guest from 188.8.131.52
On the other hand, even if your keyword density is less than 7 percent but your information architecture is good and your keywords are surrounded by the right synonyms and your keywords are within phrases composed of three or four consecutive words supporting your theme, then you may get better ranking.
You can see pages at #1 with very low density and extremely high density. You can see pages rank at #1 for a phrase with only one of the words in the phrase on the page. As a metric, KWD comes from the last century's approach to Information Retrieval. It's an "ancient god" that should no longer be worshipped.
I know that many of us have read that as long as the text flows keyword density is not very important but I just wanted to receive some more opinions.
What Ted is saying about the keyword densities of top ranking sites for a keyword greatly differing is definitely true but I just want to sort of see if being around 7 percent if it flows well is ok?
Don't forget about the "`" operator.
Before a search term phrase, place it in front to see what words google suggested as related.
Also, if google says "do you mean", follow suit.
Google's drop down suggestions is another clue to follow.
Or do you mean the tilde "~" before the term? Notice what is in bold and they are related terms, use them throughout your content and reap the dividends. Content first, related inbounds second. Scrap the keyword density, it's worthless with Google.
I think the only reason you should worry about KD is for Yahoo and other 3rd tier search engines. Nothing else.
[edited by: MLHmptn at 5:32 am (utc) on Oct. 15, 2008]
One can think of lsi as a massive co-occurrence matrix that gives the probability of that any term would occur on the same page (or paragraph) with any other term.
So yes 2% density for a given term is not really low. More than 4% and they might assume you are spamming the keyword. The key is that your content need to contain "related terms" which is how they determine the theme.
Every time that the keyword is mentioned, however, it is not by itself. Sometimes it is in long-tail form.
E.g. Keyword density for widget building is 7%. But each time widget building appears it is not by itself. Sometimes it appears as widget building machine, widget building manual and of course sometimes just as widget building. But if you look at the keyword density of widget building it is 7%.
Is having things the way that I describe them ok?
joined:Jan 27, 2003
an unusually high KWD can possibly hurt you
Yes, but not directly. A density that was "too high" is likely to be present in a page that contains artificial (i.e. irrelevant) text. But even that doesn't mean pages with seemingly ludicrous keyword densities can't rank well, since it isn't density that's important.
Keyword density may give you an overview of certain aspects of the language used on a page, but it's a crude measure and one that does not reflect the actual aspects that are algorithmically important.
It's easy to test whether there is a correlation between keyword density and ranking. Just take a sample of well ranked pages and measure the keyword density. Do a bit of number crunching. You'll likely find no statistically reliable correlation between ranking and keyword density.
For example, I may run one tool that measures ALL the occurences of a specific word, including comment tags, file names, folder names, alt tags, meta=keywords, etc. Now my understanding is that G does not take into account meta=keywords or comments tags, so my KWD output will be by definition unreliable. It may be 4% without comments or meta kw counted, but 6% with them.
So in that sense, it's a relief to hear you and tedster and others say to "not worry". Since the measurement itself cannot be based on precise criteria, I'm happy to not think about it anymore. ;)
joined:Jan 27, 2003
Now, for a keyword density tool to be more useful it also needs to account for word stemming, the fact that words not directly adjacent to each other can still increase relevance for a phrase, and the fact that related words and phrases contribute to performance.
IMO such a tool would actually go far beyond the limited definition of density. It's one way in which Google's improvements in terms of language are of great value to site owners. I get much better results asking a copywriter to right x number of words about x topic than by asking people to think about keywords too much. And then I just need minimal editing to ensure the copy is focussed.
[edited by: Receptional_Andy at 8:10 pm (utc) on Oct. 15, 2008]
I didn't test density percentages. I just tried to review the pages to see if they looked spammy.
Matt Cutts once said Google only needs to see something a few times on a page to know what it's about.
Unique, natural writing uses natural semantics and proper stemming, variations - the very thing that Google is searching for.
I do take a quick look over copy to make sure the blatantly obvious is not missing, but the algorithm has evolved to he point where I believe measurement of specific KW density factors is a waste of time if you have good content writers (or you are a good natural writer yourself)
My experience is 99.9% of the time, writing naturally wins over keyword density.
The only thing I would add is to use a writing voice that targets your audience.