By Dan Thies

MSN biased to favor IIS servers?

By Dan Thies

Ivor Hewitt has done some research which appears to show that a higher proportion of top ranked pages in the MSN search engine are hosted on Microsoft’s own IIS server platform. This work is interesting enough to get slashdotted, but it’s too early to draw conclusions.

I discussed Ivor’s results with him a couple months ago, and there are a lot of variables that need to be ruled out before I’ll buy into conspiracy theories. One very big question is whether the skew is due to a bias against ASP.NET and IIS servers by other search engines.

  • As you may know, ASP.NET applications like to insert a form field (VIEWSTATE) into web pages. This is a huge string (usually around 15 kilobytes) of absolute gibberish text. The impact of this on SEO is unknown, but if it does make a difference, you can see why it would tend to push ASP.NET applications running on IIS servers down in the search results on search engines that haven’t made it a priority (as MSN would) to work around it.
  • It’s possible that a higher percentage of dynamic sites run on IIS servers vs. Apache. One big reason to run an IIS server is to run the ASP applications that run on it. From my own experience, most ASP/IIS sites I’ve worked on have duplicate content issues, and we spend a great deal of time and money cleaning up things that we could fix in a few minutes with a PHP/Apache site.
  • With a higher percentage of dynamic sites (why run IIS unless you’re using ASP?), you could see a bias if MSN’s engine is better at handling dynamic sites, or dealing with the specific issues that come from IIS servers. It all depends on how they rank pages. :D
  • Actually, you could see a bias if MSN is *worse* at handling dynamic sites, because they may be counting all that duplicate content and giving sites credit for links there, etc.
  • Does MSN have a bias for dynamic sites in general? Because of their technology, MSN may be giving templated pages more of a chance due to the way they break a page down into smaller pieces for analysis.
  • If you buy the idea that more large sites run on IIS (I have no idea if this is true), then even a slight additional bias in favor of large sites would account for some shift in SERPs as well.
  • Maybe, just maybe, more IIS sites are part of MSN’s BCentral Submit-It service, and that’s where the real bias is, some sort of ‘authority bonus’ M$ confers on their own directory.
  • If M$ is really good at filtering out doorway domains, link farms, and spam sites, this would probably appear to bias against the low-cost Apache hosting environments where so many domains live. After all, if you’re paying for a license to run the server, it’s much more likely that you’re actually running a site.
  • Apache market share numbers may be overstated anyway, because they are based on the # of hosts and domains. I’d love to see a look at the Alexa Top 1000 and see what the market share breakdown is there. That would be a lot more telling… has anyone done this?

In any case, it seems very unlikely that there’s a variable in MSN’s ranking and retrieval algorithm that is driven by the site operator’s choice of a web server platform.

  • drakke

    MSN "biased" to favor IIS servers?

    Take five minutes and fix this problem please.

  • The research results aren’t shocking. I wouldn’t take it serious

  • I took that Alexa server stats challenge :) Here are the results – http://www.phpied.com/alexa-500-server-stats/

    In brief: Apache – 54%, IIS – 23%.

  • Brilliant, ssttoo, thanks! I had no idea anyone could slap code together that fast and actually run the test, but I guess that’s why you’re a SitePoint Zealot. :D

    This actually shows about the same ratio (2:1 for Apache) as Ivor found in MSN’s search results. Makes sense, since MSN is using a much smaller index than Google and you’d expect their SERPs to be more biased vs. larger sites. When it comes to crawling the web, nobody’s keeping up with G just yet.

    Now for one million “programming mojo points,” someone take the Alexa Top 500, extract the META keywords tags from all of them, and see whether they have more top rankings on Google or MSN.

    Netcraft’s latest numbers show 69% of domains on Apache, 20% on IIS, lending credence to the notion that these raw numbers don’t really reflect what the most popular sites are running. More popular sites are likely to have more and stronger incoming links, which of course makes them more likely to appear on any given SERP.

  • Thanks, Dan! I’m not quite clear on the what you want to do with the meta keywords, but I’ll post the code and the data from my test, so if anyone’s interested, they can take it from there.

  • I rank best in MSN and I’m on a LAMP server.

  • Here’s the Alexa stats collector

  • What I was looking for, and it would be a BIG project, would be to compare SERPs from Google and MSN as Ivor did, to see which engine has more of a “bias” in favor of popular sites. My money’s on MSN.

  • saif

    Well the question is , is this bias deliberate or accidental? doing for a search for browser wars yields different results for google and MSN. Interesting the yields for MSN pick more sites that favor IE6 than Firefox

  • Now that MS is offering tabbed browsing in IE (actually it’s been available for some time, just not well known) I wonder if Firefox has already had its day in the sun. Packaging it with the MSN search toolbar seems like a shot at Google too – if Windows users are willing to tolerate a browser that often crashes into oblivion with each minor update (Firefox) in order to get tabbed browsing, will they also be willing to dump the Google toolbar for MSN’s to get the same?

Get the latest in Entrepreneur, once a week, for free.