Monday, May 30, 2005

Yahoo! Mindset

A really cool BETA tool from Yahoo labs. It allows you to run a search, and then resort your results based on whether you expect a commercial (you're buying a product or service), or informational (you're doing research), result.

What's really interesting is that it gives you a chance to gather some perspective on how your site is viewed. Do Yahoo! (and likely other search engines), 'see' your web site correctly? Do they think your site is informational in nature when you're really trying to sell a product or a service?

For example, or main site; was viewed as informational, whereas our site for the hospitality industry - was seen as being heavily oriented towards selling. I would say this is a correct assumption; objectively.

Our main site is there to sell SEO services. But it's also used to inform people about the search engines, and optimization in general. We've collected and posted so much information over the years, that I can understand why an algorithm would place it has heavily favouring the "informational site" side.

Then I tried something different. I did the equivalent of 'google myself' using Mindset, and got interesting results. This blog was #3 overall. I guess it's relevant. :-) Number 1 was ISEDB That too was relevant. The remainder were articles I had written over the years. I moved the slider around, and the results changed, although it mostly seemed to be based on the site the article was posted to - so the same article might be more valuable on an informational site than on a service based site. I find that to be interesting in and of itself. Pages are not weighted within the site, the over site score is counted. This is a BETA, but this feature does not seem as detailed as PageRank, which varies from area to area within a site.

I then happened to spot a particular article in the results which I had written and published in October 2003. In it I had written:

"Search engines need to keep an eye to new technologies and innovative techniques to improve the quality of their search results. Their business model is based on providing highly relevant results to a query quickly and efficiently. If they deliver inaccurate results too often, searchers will go elsewhere to find a more reliable information resource. The proper and carefully balanced application of usage data, such as that collected by Alexa, combined with a comprehensive ranking algorithm could be employed to improve the quality of search results for web searchers.

Such a ranking formula would certainly cause some waves within the search engine community and with good reason. It would turn existing search engine results on their head by demonstrating that search results need not be passive. Public feedback to previous search results could be factored into improving future search results."

I'm no genius, or psychic. I can't remember Oct 2003 that well, but I'm sure that a lot of people wrote similar comments before and after my article was published. The argument was one that I'm sure many people at the time agreed with, and wanted to see happen. I wrote that article in particular about ways you could use Alexa data to model behaviour, and automatically start filtering results based on prior tendencies.

This route followed by Yahoo! is much on the same lines, except that it doesn't assume they your search tendencies are always relevant - it asks you to decide if you want information or to make a purchase. I like it.

If nothing else, this new tool gives some insight into the logic that Yahoo uses today, and opens new paths for optimization strategies.