Forum Moderators: open
The SEW tool gives my index page a KWD of 16.58% for my main keyword, including the title and meta tags. At first I worked on the assumption that this was accurate, but then a manual check showed a more reasonable 4% on visible text only. (The manual check was not truly manual -- I saved the page as a Word doc, ran a word count, and then did a search and replace on the keyword to determine how many instances there were). The manual search of course missed the meta tags, but even factoring those in the KWD would only be around 4%.
A different online tool gave a breakdown as follows:
HTML...........Keywords.....Total.........%
Title.............1...........4.......25.00
Meta_Description..2..........11.......18.18
Meta_Keywords.....3...........6.......50.00
Visible_Text.....22.........537........4.09
Alt_Tags..........2..........28........7.14
Comment_Tags......0..........26........0.00
Domain_Name.......0...........2........0.00
Image_tags........0.........115........0.00
Linked_Text.......0...........0........0.00
Option Tags.......0...........0........0.00
Reference_Tags....0..........14........0.00
Total............30.........743........4.03
The 4.03% total is close enough to my "manual" count that I'm reasonably sure this is accurate without the tedium of a recount by hand.
(Not sure what it means by "linked text" -- my anchor text certainly does include my keyword).
So, a question: how does the SEW tool determine KWD? What is it actually counting? Is it skewing the results, or is there some logic behind what it's doing? And if it is skewing the results, does it do it equally for your competitor's sites? Eg, one of my competitors has a lower KWD with the SEW tool but a higher KWD on the tool that gives a breakdown.
Thoughts, anyone?
1. Use our default stop word list.
These words will make a difference if your other program does not have stop words.
2. Ignore words of "3" characters in length or less.
The other tool has a stop word list, but it's not customizable. By default it ignores words of 2 letters or less rather than 3 letters or less as the SEW tool does. But going back to the SEW tool and setting that to ignore words of two letters or fewer, it still gives me a KWD of over 16%.
16% is a heck of a lot higher than the 4% that the other tool or a manual count is giving me.
A third tool gives me a "Weighted Keyword Density" of 7.13%.
A fourth tool gives 4.4%. When I use SEW's stop word list with this tool the KWD actually *drops* -- to 4.25%.
Now all the other tools, although different, are in the same ball park, while the SEW tool is more than twice as high as any other. Even assuming that the SEW is basing its KWD calculation on unique words rather than a total count, the results it's giving are highly anomalous.
Something seems screwy. Since there's a connection between Searchengineworld and Webmasterworld, it would be good to hear from someone who knows more about how the SEW tool works.
For example, a sentence like "you have to carefully layout your page for search engine optimization or it may not work" might be listed as "for search engine", "engine optmization or" etc. (I know it's a bad example, containing stop words, but I couldn't think of anything else--the meaning will be clear, I hope).
I haven't found a good analyzer yet...
Yes, for phrases it's easy to imagine that there would be wide variance -- but for single words? You'd imagine that would be much more straightforward.
I've stopped using the searchengineworld tool in favor of another tool, and the results have been very interesting. Every month I collate information on competing sites -- PR, backlinks, KWD, outgoing links, etc -- and put them in a spreadsheet.
When I was using the SEW tool I couldn't see any reason for why there were sites above me with lower PR and backlinks. KWD seemed to vary widely and gave no indication either.
Using another KWD tool I now see a pattern and have a hypothesis -- based on KWD -- as to why each site is in the position it is in. Naturally, I've made changes to my site as a result of this hypothesis, and I guess the test will come around Christmas, or maybe January if I've missed the crawl for this month.
Given the close links between SEW and WWM, I'm surprised no one has stepped in to explain how the KWD tool works.
Anne
When I look at the results it gives the keyword, the count, and the density. But it never seems to show words having a count of one -- only two or more.
This also explains why it's inaccurate in different ways for different pages. The more single-occurence words there are on the page, the more inaccurate it is.
Anne, I'm still trying to figure what's too much and what's too little. For my own keyword, I'm seeing sites in the top 12 with KWD from 1.4 to 9.4% The site with 9.4% is doing worse than you might expect given its PR and number of links. Most of the sites that are doing better than you might expect are hovering around 4.5%. I'm working on the assumption for now that there is a point somewhere above 5% and below 9% where your KWD starts to count against you.
This is for a single keyword. I haven't started to look at double-keywords yet.
I always shied away from the keyword density discussions, because it doesn't feel natural. Going along w/ Brett's article, put out the best content and they will come. Seems unnatural to analyze how many times your keyword appears and modify it to suit an engine. What if the algo changes and you've built 1000 pages, you have to redo them all?
But since there's so much discussion about it, I'm wondering, does it really make all that much difference? Do you use different keyword densities for the different engines?
BTW, would be interesting to know the stop words the SEs use and the methodology too ;)
I try to keep the keyword density for that page's main keywords below 7% as it's a nice average that is acceptable to all of the engines. I do not optimise per engine as it's useless unless you are using multiple doorways and/or cloaking.
I suppose those optimising for extremely competitive terms, such as adult or gambling sites, might have a different approach where it may be necessary to optimise on a per engine basis.
When I was using the SEW tool I couldn't see any reason for why there were sites above me with lower PR and backlinks. KWD seemed to vary widely and gave no indication either. Using another KWD tool I now see a pattern and have a hypothesis -- based on KWD -- as to why each site is in the position it is in.
What tool have you switch too?
