I am becoming concerned Google Analytics may not be able to track accurately in mobile devices.
Over the last year I've seen a huge rise in mobile visits to the websites I analyse. At the same time, Google Analytics has reported a massive change in how mobile users surf the web. A year ago the pattern of mobile traffic was fairly similar to desktop traffic - most people found sites in search engines, some came from other sites, and some came direct. Some sites saw more people coming from other sites than direct, and some saw more direct traffic than referrals from other sites, but all saw most of their traffic from search engines. That situation is now very different. In all cases, at least half, if not more, of the mobile traffic comes direct to the site. I see this trend in multiple countries, across multiple sectors. Since none of the sites I track are major, universally-known, brands, I have to wonder how on earth thousands of visitors are able to find these sites without search engines, especially as I know there's no major offline marketing to attract them. Most are first-time visitors, so they haven't bookmarked the site, which is the way most direct visitors get to a site. And why is it that a year ago most of these people used search engines, but now they don't?
If GA was telling me these people came from other sites I'd be less worried. The problem is that attributing a source as "direct" really means - "I asked for a referrer field and got a blank, or none." - It can just as easily mean the browser refused to provide the referrer, as some are designed to do, or that the GA referrer request wasn't processed by the browser.
So either the entire mobile community has radically changed its web useage in the last year, without anyone noticing or commenting, or Google Analytics can't track mobile referrers as well as it used to (which could be caused by changes in mobile systems, or in GA, or both).
I'm not the only person who's starting to get concerned. Social media people are starting to complain (http://econsultancy.com/ie/blog/62104-social-media-measurement-is-google-analytics-getting-it-wrong) and I've been working with some US telecoms people who are seeing really wierd stuff in their GA metrics, like mobile patterns completely out of tune with the market (eg: no iOS visitors).
Testing this can be done via an experiement, by someone able to get a range of mobile phones, access a site with them enough times to generate a valid sample size, then cross-reference what how they accessed the site with what GA reports. Such an experiment requires serious time and investment. Alternatively a bunch of us can pool our GA data in an annonymised fashion and subject it to analysis that way.
Have you seen changes in mobile visit sources shifting towards direct and away from search in the last year? If so, I'd like to know. If not, that'd be useful information too. Do you have concerns about whether Google Analytics is accurate on mobile? I'd love to hear them.
I guess we should have done this work a year ago. Mobile is a different set of operating systems and browsers from desktops, and it's naive to assume GA code would work exactly the same on all mobile systems just as it does on the limited range of desktop systems. When this data drives our understanding and decisions, naivity becomes foolishness.
So we need to know - how reliable is Google Analytics on mobiles?