I ran a relevancy test comparing all 3 SEs

Discussion in 'Google' started by FastWeb, Sep 11, 2007.

  1. #1
    Google did not fare well in my first run of tests on keyword relevancy.

    The first keyword I decided on was "tool chest". Pretending to be a user in search of a tool chest for sale, I entered the keyword into all three search engines. I am only interested in sites that have tool chests for sale. Here are the results (out of the top 20 results):

    1. Google - 7 relevant results
    2. Yahoo - 15 relevant results
    3. MSN - 6 relevant results

    Read more at my blog

    I'll be doing tests once per week to test the relevancy of the three big boys. I decided to do so because the search results from Google have been getting worse while Yahoo has improved A LOT.

    Stay tuned... :)
     
    FastWeb, Sep 11, 2007 IP
  2. amanamission

    amanamission Notable Member

    Messages:
    1,936
    Likes Received:
    138
    Best Answers:
    0
    Trophy Points:
    210
    #2
    Excuse me, but one aspect of your methodology I don't understand...are you excluding Google results that relate to tool chests, but don't sell them? Because those are still relevant to the search, even more so (how to make your own toolbox, articles about antiques, I dunno)? How irrelevant are the poor results? Are they not exactly what you wanted, or were they completely different subjects?
    To really be scientific your test should distinguish between relevant but unhelpful, and totally irrelevant, especially because you didn't include a "buy" term in your keyword.
    Anyway, I'll be following this. Interesting test...it could be some interesting link bait if your results are solid.
     
    amanamission, Sep 11, 2007 IP
  3. FastWeb

    FastWeb Peon

    Messages:
    812
    Likes Received:
    45
    Best Answers:
    0
    Trophy Points:
    0
    #3
    That's something to think about. Google came up with some pretty irrelevant results also, and I think that's almost worse than not having strongly relevant results.

    I think for my test next week (or in the next couple of days if I get time), I'll use 4 or 5 keyphrases based on a single keyword. For example, instead of just doing "tool chest", I might do:

    Tool chest
    Buy tool chest
    Tool chests for sale
    Tool chest website
    Tool chest sale

    and see if the long tail keywords score better in certain SEs. I think I'll also show Relevant, Semi-relevant, Semi-irrelevant, and Irrelevant results for each one and maybe a simple scoring system.

    Thanks for the input!
     
    FastWeb, Sep 11, 2007 IP