• http://www.mattcutts.com/blog/ Matt Cutts

    Danny, the InfoWeek article is a fun read: “The forum posts contain links to bot code that can be used for automating Live Search Club games.”

    That’s an interesting article about people using bots to send traffic to Microsoft and then getting prizes as a result.

    It reminded me about this previous situation:
    http://publications.mediapost.com/index.cfm?fuseaction=Articles.showArticle&art_aid=39885

    That’s an interesting 2006 article about people using bots to send traffic to Microsoft and then getting prizes as a result. :)

  • joe

    @Matt:
    If you really trust Compete so as to enjoy the good press it gives google, then you also have to believe that they really do have mechanisms to avoid counting the traffic generated by bots. I am not saying there are no bots involved, am darn sure there would be.

  • http://searchengineland.com Danny Sullivan

    Matt, both Compete and Microsoft says that automated queries run by the bots are getting filtered out. I mean, they’ve both been asked directly and point blank about it. Are you saying they aren’t?

    I don’t know enough technically on how the bots work to determine if they’d pollute Compete’s results. Compete takes in ISP data (which I assume would make it hard to filter out the bots), as well as toolbar data (here I think bots wouldn’t have an impact) plus a traffic panel (again, bots wouldn’t be an impact, it seems).

    For me, I’m waiting to see what the other three rating services say before I make a firm conclusion, as I said in my original article.

    Also, in my originally story, you’d asked if Compete was measuring overall share or search share. It’s search share, so things like AJAX shouldn’t have an impact, as I noted in the piece.

    You’d also wanted the raw numbers of searches rather than the estimates. I agree, that’s useful if you want to point out that even though Google’s share of searches might be reported as down, the overall number there might still be higher than the previous month.

    NetRatings: Google Spike, Live.com Rise In February 2007 explains this issue to some depth, back when people looked at Yahoo’s drop. I said that when you think about it in terms of raw searches, they don’t look as bad.

  • http://www.mattcutts.com/blog/ Matt Cutts

    @joe, I’ve written skeptically on Compete’s metrics before: http://www.mattcutts.com/blog/review-compete/

    @danny, speaking as someone who has seen a lot of creative ways for bots to send traffic to a search engine, I know that it’s a hard problem to tackle well. Do I believe that Compete or Live make earnest efforts to find bots? Sure. But the Comscore numbers estimate that MSFT went from 757M queries to 1.1B queries, an increase of 343M queries. That’s nearly a 50% increase in queries in 30 days, which seems pretty artificial. Do I think that Compete/Live successfully detected all bot traffic with 100% accuracy? Color me skeptical. :)