I was doing some soul searching over this site’s Google Analytics data which as Chris told you put Windows device traffic at 3.7% , now 2.7%, iOS 53%. Hard to believe given how alone all you WP enthusiasts make me feel here. So I ran the numbers myself, came up with some figures that roughly 40% of you will appreciate. I’ve let this slide for too long with Google, time to blow the whistle.

You probably didn’t notice but in addition to having Google Analytics javascript I have a 1×1 pixel transparent blank GIF in the footer of all pages hotlinked to my own server which for a few reasons struck me as a good idea, largely because relying on Javascript made the Google Analytics data too vulnerable to doubt whereas just about every browser will load that hotlinked GIF in the site-wide footer on a normal non-RSS pageview.

So every time someone does something on the site, it leaves a mark on a big text file I’ve been collecting for some time. Included in that file is whether you’ve loaded the GIF before, so can differentiate between landings and multiple pages per visits. As for RSS and Google Reader traffic, I’m in the habit of tagging my articles and sometimes David’s in order to get a peek at that data as well. Between these little invisible GIFs and Google Analytics, also the server-side logs which are kind of useless with all the bots inflating them like crazy, though not 100% air tight, I’ve got a pretty good view of the viewers. One might call it obsessive or sort of creepy. Pic related, by the way. :)

So Google says that yesterday, of the mobile device visits to any page on the site, only 2.7% of them were “Windows” while 49% were Apple (iPod, iPhone, iPad) and 45% Android. As much as I want to believe that, I sought a second opinion from my server. I dumped Monday’s blank GIF-serving data into two different files, one counting only landings while filtering out cached hits, another counting all pageviews, both including hits to the mobile theme.

I’ll give you the details of exactly how I produced these figures if you want, but for pageviews for yesterday I came up with iOS at 30%, Android at 27%, Windows Mobile at 6%, Windows Phone at 34% — so 40% for Microsoft-based mobile platforms, not 2.7% as Google’s showing. Visits almost the same give or take, though I scooped up 69% more mobile visits out of the pile than Google tallied while using the same array of platforms – even though our total figures for phones and computers for Monday were on par with each other, and that’s also peculiar. I did this meticulously.

If you’re inclined to believe my data over Google’s, or at least somewhere in between, it’s probably true that Google is telling a lot of other people out there checking out the mobile hits on Analytics to their site that Windows devices basically don’t exist even more than actually is the case.

And that in turn leads people like Chris to write an article about the stunningly low figure on a very WP-heavy-looking website like this, and that activity occurring here and probably elsewhere definitely doesn’t help the platform attract more consumers or hang on to the few it already as. Nor does it reflect well on Google Analytics and my overall confidence in all of its data, or yours now that you’ve read this.

Here, look, the most recent hit, and guess which platform it is:

mobile-[redacted].mycingular.net – – [16/Aug/2011:13:24:10 -0400] “GET /blankmd.gif HTTP/1.1” 200 358 “http://mobilitydigest.com/windows-phone-7-tethers-you-can-do-it-now-heres-how/ “Mozilla/4.0 (compatible; MSIE 7.0; Windows Phone OS 7.0; Trident/3.1; IEMobile/7.0; SAMSUNG; SGH-i917)”

I’d say the deck is stacked against WinPho enough as it is and that by now this should have been addressed. What’s up with that, Google?

Doug Simmons


  1. “Hard to believe given how alone all you WP enthusiasts make me feel here.”

    Maybe if you didn’t take a shit on Microsoft, Windows Phone, and anyone who uses and likes a lot of the stuff they make, you wouldn’t be made to feel that way. But your irrational hatred of anything and everything Microsoft comes through loud and clear in every article you write.

    But I appreciate the numbers. I don’t know what to believe or what’s accurate. When these services try to extrapolate Windows Phone data by including Windows Mobile, they lose all credibility. Windows Mobile is no longer made and no longer supported. It’s obvious that it’s market share is sinking, and yet some of these services claim that means Windows Phone is sinking. From now on, they need to forget about Windows Mobile, and just give us the numbers on Windows Phone.

  2. But they don’t identify that as being Windows Mobile data, they call it Windows data under mobile device traffic. Even if it were Windows Mobile with no Windows Phone in it, that’s still less than half of what WinMo traffic appears to be on my end. As for combining the two, not sure why that upsets you. They don’t divide all the different Androids and believe it or not but Windows Mobile 6 phones began identifying themselves to servers as Windows Phone.

  3. I’ll give you an example, a hit to my server: – – [22/Jun/2011:18:42:47 -0400] “GET /images/touchpro2batterystack.jpg HTTP/1.1” 200 41096 “http://batteryboss.org/” “Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; Windows Phone 6.5)”

  4. Nice sleuthing, Doug.

    I’m sure there is a reasonable explanation here, but what a difference.

    When you’re being objective, you’re very good at it.

    When you’re being subjective, you leave lots of near-naked women so we know it’s you.

    You remind me a little of this guy, and that’s a compliment: http://www.politicalmathblog.com/?p=1590

    You’re true to the data, wherever it leads you. Thanks.

  5. Windows phone browsing is the best of them all… Android browsing is just sluggish… Any android phone out there can’t compete with IOS and wp7 when it comes to browsing the net….

  6. Maybe this is an across the board thing and since so many of the calculations as to phone sales are based on web hits maybe WP7 is really 40% of the market currently…if only it weren’t for those bum analytics which Google is leading the charge on then the truth would be told.

    What? We’re still single digits? crap…

  7. “As for combining the two, not sure why that upsets you. ”

    It only upsets me because it doesn’t tell us anything about what Windows Phone is doing, good or bad. It skews the data. For example, I’ve read a recent article that lumped Windows Mobile with Windows Phone, and said that Windows Phone sales crashed. In truth, it was Windows Mobile that crashed, and Windows Phone remained stable.

    In fact, here’s a link to a perfect example of what I’m talking about. You can’t tell what Windows Phone is doing if you don’t look at it specifically. Windows Mobile is no longer made or supported. So why even bother with collecting data on it. Of course it’s collapsing.


    “They don’t divide all the different Androids”

    Android is a single OS. Windows Mobile is not Windows Phone. They are two entirely different and separate OS, one of which is no longer made or supported. Tell us what WP is doing and leave Windows Mobile alone. Does anyone really care what Windows Mobile’s sales and market share is?

    Pretty strong analysis.

  8. “What? We’re still single digits? crap…”

    I expect it to stay that way for a while. But I saw a recent article that said WP is at 2% market share in the U.S. While that is not good, it’s better than the 1% WP used to claim. That same article stated that WP was at 7% market share in Germany.

  9. Thanks Ike. I’ll read that link next.

    I think the arguably reasonable explanation is that Google’s got a pretty big to do list, and just like making WP versions of all their apps isn’t on it (just as it isn’t even always on Microsoft’s), tweaking whatever code that’s in Analytics that parses user agents to be aware and take not of Windows Phone’s strings still hasn’t been enough of a priority to maybe even occur to anyone to deal with it. The concern is elevated here obviously because, well not obviously actually because I think some of us tend to project what’s in our heads onto the non-nichers out there, because we’re a phone blog, chiefly Windows Phone news. So I’d expect it to strike people here as excessively negligent, objectionably negligent or intentional.

    My best guess is that they’re just counting Windows Mobile phones that identify themselves as Windows Mobile, not Windows Phone, which would help explain why the number of Windows Mobile hits I got are much smaller than the whole Google Analytics Windows figure combined.

    Otherwise why would there be any data at all.

    If the explanation is technical, like something to do with Javascript an IEMobile, then sort it out with Microsoft or cook up some math to extrapolate from the data you do somehow get and try to produce a more accurate number, or indicate that the number you do report for Windows-based mobile devices has a double digit margin of error. And if it’s something funky about javascript, then maybe it’s time to let people do page tagging instead of just javascript with Google Analytics.

    I’d wondered when I first discovered this if we were doing something wrong on our end, depriving Windows Phone phones an opportunity to snatch the GA javascript and crunch it, so I spoofed the user agent to make sure they were getting the code in whatever theme this site dished out to them, the answer’s yes it’s there.

    Note that the data I came up with does include RSS and a healthy chunk of Windows Phones get this site’s content through RSS, that Weave thing specifically which mobilizes the page (and probably doesn’t hit Google in the process). The other platforms have such software, but after this article drifts further down from top post I think it would be worth while to see how our RSS and Google Reader traffic compares across the mobile platforms. I bet it’s even more flattering to Windows Phone than the on-site viewing data I came up with.

    Forrestall: Happy to help. Not used to all this lack of exclusive negativity.

  10. Joe: While I appreciate that it’s very cut and dry to you that Windows Phone 7 has nothing to do with the thing I pasted that reported itself as Windows Phone 6.5, maybe you should follow Microsoft’s lead and try to fluff that number up as much as possible by not excluding anything. Even though they appear to be missing most if not all data on Windows Phone, Google does reference the figure as Windows and mobile devices, so in spite of the rebranding, it seems sufficient for almost everyone’s purposes. If you want to pull up every single individual browser that touched your site, like if you’re XDA or whatever, Google can show you that if you don’t mind relying on them (and javascript).

    I could go through Google Analytics and dig out pretty much everything — except full IPs. Pesky privacy fanatics.

    Joe, if you want the truth and can handle the truth, use that Facebook scraping data, that’s as close as you’ll get. Even Surur’s with me on that.

  11. Doug, all I can say is respect… at least until you decide to go on another Windows Phone/iOS rant.

    There’s nothing wrong with being an Android fanboy: just don’t go on crazy bashing parties.

  12. My question is if Google is throwing out/ignoring about 37% of Windows Mobile/Phone traffic on Mobility Digest, where is it being shown? Those hits have to be tallied somewhere. Did they give you some ambiguous “Other” column accounting for 36% of your mobile traffic? That would seem suspicious to any webmaster tracking his/her stats.

    Regardless, I’m glad to know somebody’s checking up on these things. It makes me wonder about the accuracy of those market share statistics that come out every quarter based on mobile web traffic.

  13. Mark, here’s what the user agent of Windows Phones through HTTP GET requests for something like a blank gif, and I’m guessing given that Google’s javascript collects so much more information than I do, that they at least get something along the lines of the following:

    “Mozilla/4.0 (compatible; MSIE 7.0; Windows Phone OS 7.0; Trident/3.1; IEMobile/7.0; HTC; 7 Trophy)”

    Google Analytics can give you a list of actual individual user agents, so let’s see if GA is at least counting IEMobile at all in the full dump (IE not separated by mobile and not mobile), gimme a sec. I gotta do this in a way that you math wizards can’t try to figure out the rest of our traffic data, those books are still closed.

    Okay, so telling Google Analytics to give me a dump of the most popular browsers (also split up by browser version) for Monday, here’s what I get, that Chrome Stable is the most popular browser (though everything with IE in the name overall is the most popular), followed by IE 9 which could either be IE 9 that Brianna runs on her desktop or it could mean this guy as well:

    [redacted…].available.mfnx.net – – [17/Aug/2011:15:46:39 -0400] “GET /blankmd.gif HTTP/1.1” 200 358 “http://mobilitydigest.com/blake-griffin-and-his-brother-get-windows-phones/” “Mozilla/5.0 (compatible; MSIE 9.0; Windows Phone OS 7.5; Trident/5.0; IEMobile/9.0; SAMSUNG; SGH-i917)”

    My guess is that the above hit is getting counted by Google not as a device hit but as a regular desktop MSIE hit.

    So here’s the top clump of browsers (specific to the version number but not extremely specific to the full user agent) for you, a little screen shot in descending order, then I’m going to see what’s going on with resolutions which I think may explain this whole thing (what’s happening technically that is, not why Google didn’t get off its ass to tweak the code, surely they do it all the time with other browsers… argh, okay screenshot, then I’ll come back to you with the resolution and browser breakdown if I can get it which is what I should have started out by doing:

  14. Okay folks, now we’re getting somewhere, the resolution/browser data. Boy, this is pretty revealing in terms of what we’re after. Can’t give you the visit figures, though the top entry is twice that of the second.

  15. In my experience, all web analytics software needs heavy tweaking in order to filter out the crap and get some real, useable information. Google analytics wouldn’t be my first choice. Google is great at search algorithms… the rest…meh.

  16. Which is why I look at the actual server’s logs (and not parsed shit, not webalizer, but the actual logs), my server’s logs for the vhost I run just to serve all these blank gifs from pagetagging this site which is what I used to come up with the figures in the top as well as Google Analytics which, by the way, for most people is really the only game in town.

    Regarding the fun data you can get out of Google’s javascript, like screen resolution (which is, if you look in the table, revealing to us that the most likely screen to view this site is 480×800 and runs IE, which affirms my own findings and my guess that Google hasn’t bothered to discern between Windows Phone MSIE/IEMobile hits from desktop MSIE hits, consequently making people who glance at mobile devices think that Windows Phone is much less popular than it is because not only is it not counting Windows Phone, it’s apparently not counting all the Windows Mobile 6.5 phones that report Windows Phone as their platform in their user agent, even though they’re not WP, hence the “Windows” figure under mobile devices on Google being less than half what I picked up for Windows Mobile. That stuff can be picked up to, though not as well, with javascript plugins, for example, for something like awstats, a server-side cruncher.

    But that was me getting very sidetracked, a lot of coffee. Returning to what you said, what would be your first choice? I’ve got one screen devoted to tailing httpd access logs of my various sites, parsing them like Rain Man in my head.

    I said earlier I was going to run the RSS data I collected for this article.. little later.

    Hey I was expecting a bigger reaction to that second screenshot, that’s kind of big information right there. But I guess we already knew the WP presence on this site would be in that neighborhood… but up top? And double that of the next resolution/browser?

Comments are closed.