Search engine themes

The concept of themes was first brought up by Brett Tabke, webmaster of Search Engine World and a search engine optimization expert, in an article discussing search engine themes published in his newsletter. He claimed that search engines are now widely using "themes" as a key part of their ranking system, and that adapting them is an important aspect of search engine optimization.

What exactly does this theme stuff mean? The answer is pretty simple. The common, traditional point of view is that search engines rank pages separately. The theme thinking is that search engines have now begun to rank sites instead of pages, that they combine all the pages in a single site into one big "page". This would mean that optimizing a single page for a single keyword would no longer cut it - you'd have to optimize your whole site instead of just that one page.

To optimize your site for themes, you'll need to weed out content that is not related to your site's main subject. If you have a site about pets, but there's also some pages about Formula 1 driving on the site, you'll need to move the F1 pages elsewhere so that the SE's recognize your site as being highly relevant to the keyword "pets" (and other pet-related keywords) and that the F1 pages don't confuse the ranking system. You'll also need to make sure that your most important keywords appear not just in one highly optimized page, but in all of the pages on your site, in all of your META tags, in your content and often in your headings and titles.

Does this stuff really work? I'm not sure. Brett seems to be serious about this, and he should know what he's talking about when it comes to promotion. Also, some other people have also reportedly been successful by converting their sites into "theme-based". Because of these reports, I decided to try a little experiment of my own.

I added a H2 tag at the top of each page and used CSS to reduce the size of the tags. Each H2 tag was identical, short (ab. four words) and contained the keyword I was hitting at in the beginning of the tag. I had the keyword in every title of every page (the page titles were partly similar to each other, but not completely). I made sure that the keyword appeared in most of the META tags on my pages and at least once or twice in nearly every article. I also used the keyword in the link text of just about every page. The site used in the experiment had around 30 pages, every page was submitted to the engines and verified that it did indeed get into the database.

That, I figured, was the way I should nail the theme algorithm. So I went ahead, submitted and.. got killed. I focused my search engine optimization experiment on Altavista and Google. On Altavista, I did not see any positive effect at all, but since AV has recently been pretty unfriendly towards me, I didn't lose much. On Google, not only did I fail to see any positive results, but I lost some nice rankings that were delivering pretty good traffic. To sum it all up, this experiment left a bad taste in my mouth.

Before anyone starts jumping to any conclusions, I'm not saying Brett and everyone else is wrong and I'm right. I'm saying that my tests failed to show any positive results with the approach I used. My testings weren't very extensive, and it is possible that I misunderstood what Brett was trying to say, or that I just overdid it with the H2's. It's also possible that the engines have dropped or greatly altered their 'theme'-algorithms since Brett first reported about it in the summer. I wish the results of this experiment would have been different, but I feel that I have to 'call 'em as I see 'em', even if I don't like what I see.

But even as the experiment failed, I do think that the principal thinking behind 'themes' is sound, and might well be something that will gain weight in search engine algorithms in the future. You might want to read Brent Winter's article about themes in the November 2000 Marketposition newsletter, which I feel pretty much agrees with the views I've expressed here - themes might not have a great effect yet, but they are probably coming sooner or later.

So, my recommendation about the subject would be: If you're going to convert your site into theme-based, don't just jump in head first. Do some experiments, and if you decide to use the 'theme'-approach, don't risk everything on it: In addition to using the theme-approach, optimize every page well for a different keyword.

If there's anyone out there with good or bad experience with themes, comments about my experiment, or just some general thoughts about this subject, I'd be more than happy to receive them.