Friday, 25 July 2008
Jennifer Laycock's recent post on How Search Engines See Keywords over at Search Engine Guide really surprised me. It harks back to the 1990s, with talk of keyword density, and doesn't match my understanding of modern search engines at all. It especially surprised me given the author: I felt that Jennifer was pretty savvy about these things. Maybe I'm just missing something really crucial.
Anyway, my understanding is that the search engines index each and every word on your page, and store a count of each word and phrase. If you say "rubber balls" three times, it doesn't matter if you also say "red marbles" three times: the engines don't assign "keywords" to a page, they find pages that match what the user types. This is why if I include a random phrase on a web page exactly once, and then search for that phrase then my page will likely show up in the results (assuming my phrase was sufficiently uncommon), even though other phrases might appear more often on the same page.
Once the engine has found the pages that contain the phrase that users have searched for (whether in content, or in links to that page), the search engine then ranks those pages to decide what to show. The ranking will use things like the number of times the phrase appears on the page, whether it appears in the title, in headings, links, <strong> tags or just in plain text, how many other pages link to that page with that phrase, and all the usual stuff.
Here, let's put it to the test. At the time of writing, a search on Google for "wibble flibble splodge bucket" with quotes returns no results, and a search without quotes returns just three entries. Given Google's crawl rate for my website, I expect this blog entry will turn up in the search results for that phrase within a few days, even though it only appears the once and other phrases such as "search engines" appear far more often. Of course, I may be wrong, but only time will tell.