Forum Moderators: open
Many other factors such as incoming links, titles, alt tags etc. play a roll in ranking placement at the various search engines.
Do a site search (link at top of page) for keyword density to find oodles of info on this subject. :)
If you take the keyword and divide it by the word count, it gives you a percentage.
I understand I can run a page report against my top competitors and just beat the highest keyword density percentage, but what I'm really wondering is if the very best keyword density goes up exactly as the word count goes up, as in a line. If that were the case, I could just shoot for using a certain amount of keywords depending on how many words there are in the page. This way, I wouldn't have to beat the highest percentage of my competitors keyword density because I would already have the optimal keyword density that my page could possibly have.
But do SE's actually work like that? Or do you think the optimal percentage is the percentage of the #1 ranked page for that keyword, which explains why it is #1 in the first place?
I know there are many other factors that are considered when an SE looks at a page, but I'm just curious about this one topic for now.
Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Aenean placerat volutpat metus. Curabitur faucibus diam eget justo. Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae; Etiam tempus blandit lectus. Proin ullamcorper. Proin iaculis est ut lacus. Morbi nec libero. Donec molestie ipsum non dui. Vestibulum porttitor pede sit amet purus. Mauris ligula. Cras arcu. Pellentesque lorem mi, consequat eu, tempus nec, lobortis ullamcorper, magna. Donec rhoncus laoreet leo. Vivamus ipsum lacus, venenatis vel, vestibulum a, mollis ac, lorem. Sed ultricies, metus ac fermentum pulvinar, sapien tortor aliquet turpis, vitae venenatis risus wisi vitae risus. Integer vel lacus quis turpis pellentesque accumsan. Integer condimentum ligula. Suspendisse sapien dolor, luctus sed, tempor vitae, tincidunt quis, est.
Yep, greeking. I left keywords in anchors, titles and headers on those pages and keywords in anchors pointing to those pages. The pages maintained their position in the serps for 4 months and I took them down.
So, while other folks are busy analyzing KW density down to exact percentages, I write pages of content with no regard to KW density and they seem to do just fine.
Look at it this way: You take like 1000 queries, and make up a statistic of keyword densities. The top rankers establish the optimal keyword density, wit some further refinements.
Now what do you have? A pile of BS. The data you fed into it are BS, and so is the outcome.
The feed is BS, because the ranking you based it upon was in reality not all determined by the density.
So it is a futile attempt to derive further conclusions from recommendations/rules which are based on totally flawed statistics.
The placement of keywords and phrases is important in any engine, even Google. But the density?
There are pages with 50% density ranking anywhere from #1 to #637.928, likewise pages with a density of 5%.
Okay, and now I'll shut up! :)
As you were saying heini, WebPosition doesn't know the SE's algorithm. And even if it did, I think the best keyword density might depend on the page anyways, and so I doubt the SE's have set values that they look for. Which makes it impossible to know what that percentage might be for each individual page.
Thanks everyone.
In very competitive areas kw density may play a less significant role.
DG, I'm sorry you took those pages down after your experiment, I was wondering what would have changed if you'd have the text replaced with high density relevant text (unless you were #1 already).
The thing is just that density, as in statistical frequency, is a minor factor. Only when all other factors are set it's worth it to play around with density.
And, to repeat, the possible win of writing for density IMO rarely justifies the limitations it puts on usability and persuasiveness of your copy.
DG, am I wrong in assuming you were talking about Google rankings? In Google that would make sense, as off page factors play such a large role, that on page factors, especially body text can be completely neglected. We all know the effects of that, as when pages are returned which contain nothing but "under reconstruction", or when google bombing takes place.
If you have a particluar page that performs well, using 4 keyword repeats and 100 words, and then you want to create a 200 word page on the same keyword and keep all other things equal (external factors too) as much as possible, I would go for a proportional keyword count first, 8 keywords per 200 in this example.
Doug's question is not wheter keyword density is that important in this case , just wheter it would make more sense to try only 4 repeats for the 200 word page or 8. It's presumably just as much work to get an exact keyword count of 4 as it is to reach 8, so in that case I'd choose 8 for what it's worth.
I really would not be surprised if the 200 word page ranks much better then the 100 word page or the other way around, so I agree with Heini..don't worry about it too much before hand...first make your copy read well and sell to your visitors that you do have.
After it's indexed, see if you can tweak it by playing with keyword density/repeats without hurting your copy. I think even for the most seasoned seo's it would only be an educated guess in what direction to tweak..more or less keyword density/repeats..as you already mentioned it largely depends on the particular page.
IMO keeping track of the density numbers does help to compare situations relatively over time about one page, so it does have it's uses in giving you a model of a page. If you raise density from 1 to 5% and the page jumps to the top of the serps it's a fair bet to say 5% is better for that page. If you then raise it to 10% (possibly just because you want to change the copy) and you drop again you might like to go back to the 5% setting...be it with different copy then the copy you had when you first set the page to 5% density.
So let's say I'm optimizing for AltaVista. My first coordinate would be (5,445) and the second coordinate would be (26,819) - based on WebPosition data. Based on the line this creates, I can enter in my word count, and it tells me exactly how many keywords to use. So for 625 words, it says to use 5.8 keywords.
Is this reliable? assuming the WebPosition data is correct and that the content still reads well to customers with exactly 5.8 keywords and exactly 625 words - however impossible that may be :)
No, you're correct. :) However, none of the pages slipped in position for Inktomi, Fast or AV. They slipped in MSN but never lower than #11 in that 4 month period. I used 5 different pages, all ranked #3 on Google for a specific KW phrase. I used different greeking paragraphs on all of them and all of them had 5 paragraphs of greeking.
What I did find interesting was that 2 of the pages ranked quite well on Google for a phrase that appeared in the greeking. ;)
I'm not suggesting that KW density doesn't have a role to play, just that it is a very small role.
LOL DG! It prooves Google IS reading page content as well, which I've always been convinced of. (Do I remember a similar thread between us about this?).
Heini: I agree that KW density can be a critical factor, in ATW even more then in Google. But as DG states, other factors are more important, like *where* you put them (on the page).
That's bad news for Dough: SEO and keyword density are no sciences with ever lasting rules. Even if you could find such a formula for now, it will change with every algo tweek and you have to test again.
Goodnight!
Patterns in Unstructured Data
[javelina.cet.middlebury.edu...]
I think it has been posted here before.
local weight, global weight and the normalization factor:
Google and AltaVista must not because they don't support partial matching (or Porter Stemming) which was described as part of the LSI technique in the article you suggested. Which also makes sense since DG's "greeked" pages still held positions in Google without actually containing any keywords in the text (seems LSI revolves around textual content, not title, meta, alt tags and anchor text).
Also, after reading that article I can tell that there's a lot more that goes into calculating keyword relevency in the body text than just the keyword to word count ratio and now I see where heini was coming from in saying that statistics don't matter. Although the term-document matrix values are arrived at in a linear fashion, the page results for a specific keyword depend on the topic. So I guess I was right in saying that 'the best keyword density might depend on the page'.
Even if SE's aren't using the LSI model, they are probably using something like it, so thanks for your post. It really cleared some things up for me.
