Forum Moderators: open
For example one could build a plain text page with a good keyword density. However, one could instead build a page with lots of images with alt text, links, etc etc as well as text. The image filenames, alt text, links etc could be keyword rich.
My hunch is that a good search engine will rank pages higher with a variety of elements within them containing keywords than just a page of optimised plain text.
This makes the issue of keyword density analysis, however, rather complex. What weighting do search engines give to image filenames, image alt text, links, etc etc.? What are the correct density levels here?
Perhaps I am wrong and search engines still like plain ol' text, but I suspect that they will want to "see" relevant images, links etc etc as well.
Unfortunately, there's not a magic formula for the keyword density in every part of the page, particularly across multiple SEs. In general, though, it's safe to say that visible stuff carries more weight than invisible stuff. Comment tags haven't had any effect for ages, image alt tags seem to be declining toward no effect, etc. Heading text and link text, on the other hand, are probably a bit more important than plain text.
If you want to do some analysis of successful pages, use some of the free keyword density checkers. Web Position Gold also has that feature (and even has broad statistical values), but be mindful that some SEs don't like automated queries. Since off-page criteria count more and more these days, collecting meaningful data is that much more challenging.
Take a look at the original paper on The Anatomy of a Large-Scale Hypertextual Web Search Engine [www7.scu.edu.au]. Guess those guys have nothing against large amounts of text :)
