Ken Kovash has an update on the Net Applications Market Share data for the month of April that reported an across the board drop in not-IE market share and a pretty substantial spike for IE 6.
Long story short, the report was wrong, Firefox did not lose share, IE didn't have a massive spike, and Net Applications will be posting a corrected report.
The April, 2008 market share data has some significant variations from established trends. The following major anomalies occurred on April 18th:A 25% increase in visitors A 3% drop in Firefox share A 4.9% increase in Internet Explorer 6.0 share A 3.4% increase in Windows XP share (with a small drop in Mac share) A .7% drop in Windows Vista share
Since April 18th, all trends have returned to expected values, including an expected uptick in Vista share due to the release of SP1.
Once we discovered the extent of the variations, we have worked diligently to discover the cause. The variations were coincidental to the release of Vista SP1 to automatic updates, so we initially thought there might be a connection. However, our investigation showed Vista SP1 had nothing to do with the problem.
What happened was a distributed collection of sites inadvertently caused the problem. We can't identity the sites responsible, but the nature of the problem is that all the millions of new visitors we saw were part of a massive marketing campaign that only worked on Internet Explorer. This glitch caused respective drops in Firefox, Safari and Opera share.
We are in the process of removing the skewing data. It should be completed by May 7th.
So, one anomalous day that was enough to skew the monthly report. Now, had I subscribed to Net Application's super cool upgraded version, I'd have seen this as a one week anomaly in a specific geography (weekly reports and geo location data being the key parts of the paid upgrade,) and a return to normal trends before the month was over, saving me a few gray hairs. Live and learn :-)
A big thanks to Ken Kovash and Net Applications for getting to the bottom of this.
Posted by: Jesse Ruderman | May 6, 2008 3:58 PM
Jesse, I expect they know but don't say.
Posted by: Asa Dotzler | May 6, 2008 4:29 PM
Only works in IE6, eh? Wonder what it could be... massive popup spam? Drive-by infections?
Posted by: ant | May 6, 2008 6:25 PM
Apparently, it doesn't take too much to skew statistics. A "massive marketing campaign" on one single day made a significant impact for the entire month - I wonder how skewed the statistics already are by things that run all the time.
Posted by: Wladimir Palant | May 7, 2008 10:01 AM
Posted by: Name | May 8, 2008 10:47 AM
http://marketshare.hitslink.com/report.aspx?qprid=0&qpmr=15&qpdt=1&qpct=3&qptimeframe=Q&qpsp=36
well, even after those "skewed data" are removed, it still shows an increase in IE, and a decrease in Fx.
And it only makes you sound sour when you care about this little "anomalies" so much. I don't see any IE fanboy celebrating this "event", complaining like this just make the Firefox fanboys look bad :rolleyes:
Posted by: Waleof Suous | May 12, 2008 4:57 AM
A "massive advertising campaign" with "millions of visitors" and yet they don't know (or won't say) which sites were involved? Weird.