For a company with such a public focus on mobile browsing Google has repeatedly failed to cater for anything mobile in their analytics product. Indeed, independent comparisons earlier this year showed that Google Analytics only tracked 12% of the mobile page views to a website that Bango Analytics identified (see http://www.brysonmeunier.com/mobile-analytics-with-google-analytics for details). So after much waiting and unknown quantities of inaccurate analytics reports later, Google announced a new version of Google Analytics and this time the long list of features includes [sound of fanfare] “mobile tracking”. Among a big list of new features on their website there is a whole, but small, paragraph dedicated to mobile.
And this complexity is only the start of it. Google themselves warn about the additional load this new tracking code will add to your servers and instructs you to test this out on a small, less important part of your site to begin with. They also tell you to ignore any location information because the connection originates with the wireless carrier gateway which is not co-located with the mobile user. There is no proof that Google has any solid knowledge about carrier gateways – they certainly cannot detect gateways that are shared between Mobile Network Operators (MNO) or differentiate Mobile Virtual Network Operators (MVNOs) from the host MNO (very different demographics). There is also no proof they can correctly tell you which phone your visitor is using. Even other legacy analytics solutions, like those from Omniture, CoreMetrics or Webtrends manage to provide basic handset identification through integration with DeviceAtlas.
Which brings me to the most important consideration – the accuracy of information recorded. Accuracy is the number one factor for selecting an analytics solution according to Forrester Research. Accuracy of any analytics solution is depended on its ability to identify each individual visitor. We all recognize the limitations found in traditional solutions that depend on cookies, IP address and user agent strings – you cannot track everyone reliably because people delete cookies or turn them off for example. Well on mobile it is many more times as complex, what with all the differences in browsers, devices, settings, connections, transcoders, operator gateways and so on. Traditional PC centric systems that still rely on one of the standard visitor identification techniques all end up struggling to deliver any sensible level of accuracy on mobile. It is a common problem found in all legacy PC analytics solutions like Google Analytics and something they are all keen to hide.
Bango significantly improves on accuracy and quality by using a server based user identification technology that leverages many mobile industry relationships to obtain a precise user ID for each visitor. Only when this ID is not available does Bango fall back to more traditional techniques. This intelligent, tiered approach allows Bango Analytics to deliver up to 8x higher accuracy, giving you mobile results you can depend on. But don’t take my word for it, grab a trial of Bango Analytics, install the simple tracking code alongside Google or any other legacy analytics solution and see the difference for yourself.
So what has Google achieved with their latest analytics release? Well it looks like they have suddenly woken up to mobile measurement and rushed something very basic out. The new Google mobile analytics solution now works without breaking some sites, but still lacks the accuracy delivered by true mobile solutions like Bango. It is a shame that after all this time their attempts to catch up have only managed to land them in last place.