The Search Engine Relevancy Challenge: Pepsi vs. Coke

Apr 13, 2005 • 4:49 pm | comments (10) by twitter Google+ | Filed Under Search Engine & SEO Theory

As a follow up to my entry named Google Relevancy Poor Based on Generalization, I have started a new thread at Search Engine Watch Forums named Coke vs. Pepsi Challenge for Search Engines.

I understand that relevancy is extremely subjective, however, search engines all want to achieve the most optimal level of relevancy possible. The engines all want their results to be the most relevant for each individual searcher, no matter who it is. Based on that, I figured it would be cool to set up a Coke versus Pepsi challenge, but for the search engines. Basically, the concept would be to create a "white labeled" search engine page, allow users to query the search engine and return results from one of the big four (Google, Yahoo!, MSN and Ask Jeeves). Then the user would mark which results are relevant on some sort of scale. Collect this data over time and then show the results.

Currently, many search engine users accept that Google or Yahoo! or the engine of their choice is relevant. So if I am a Google user, I will compare those results to Yahoo! and say, wow Yahoo!'s results are far less relevant to me then Google. Relevancy, as mentioned above, is extremely subjective. And with subjective concepts, comes bias. Search engines spend a lot on branding, to influence user bias. If we take branding out of the equation, I strongly wonder what the user would consider relevant and what the user will consider irrelevant. Even more so, I wonder which of the four engines would win the contest.


I am strongly considering building a white labeled search engine to do this. Since Google and Yahoo! both have APIs and MSN has an RSS feed, I can make it work. In addition, I think Ask Jeeves would be happy to provide some sort of feed. If you are interested in building out this white labeled search engine to collect subjective relevancy data, feel free to let me know. This can be fun!

Previous story: Domain Names & Hyphens, Lots of Hyphens
blog comments powered by Disqus