Missing Analytics Data? It’s (Maybe) Not Your Fault
January 21st, 2013 by
With a recent, subtle Chrome update, Google has significantly curtailed the tracking of keyword data within its own Analytics. It’s been over a year since tracking of keyword data has been disabled for logged in Google users, who were taken to the secure version of the page for any searches. Keyword data from https://www.google.com/ searches (versus http://www.google.com/ searches) shows up in Analytics reports as “(not provided),” reportedly to protect the privacy of Google’s users.
With the recent Google Chrome update, searches from Chrome’s “omnibox” (its term for “address bar”) now all go through the SSL version of Google search, and thus, keyword data for these searches will not be tracked in Analytics. This is worth noticing, since Google Chrome is clearly dominating the desktop web browser arena at present. In short, this policy could be causing webmasters to miss out on between one-third and half of their data.
I casually happened upon this phenomenon last month as I was testing a client’s site for problems with cross-domain tracking for their Analytics. My usual procedure to bypass hidden keyword data was to go into Chrome’s incognito mode, search on Google for something that I believe will take me to the client’s site, then visit the client’s site from the results. On inspecting the Analytics cookies, none of the search keyword data was present! I fumbled through all of my Chrome extensions, said “UHHHHH,” a lot, and eventually resigned myself to the fact that that my browser was somehow broken.
Turns out that it wasn’t broken — this was intentional. The changes are obvious: comparing keyword data for our clients’ Analytics profiles between December 10 and December 18 of this year versus the same of last year shows a consistent increase in “(not set)” or “(not provided)” keywords. This isn’t to say that keyword data is not present, but rather that it’s unavailable from the leading desktop browser.
Google Chrome doesn’t alone hold the distinction of presenting difficulties for achieving accurate visitor data. For the same client, I also noticed a significant presence of “direct” traffic, which is an apparently misleading term. After a bit of research, I found that the issue was due to security settings in Safari counting a significant portion of traffic as “Direct.” As you can see by the screenshots below, Safari traffic (a majority of which is through iOS – Safari being the default browser) accounts for a sizable portion of all direct traffic.
Google’s official description of direct traffic reports in Analytics is that they provide details on “which of your URLs are the most popular destinations for direct traffic: which URLs people can easily remember (e.g., google.com), which addresses appear most often in auto-completion, or which of your pages are bookmarked the most.” As we’ve seen, this is misleading. The depicted client had a significant amount of direct traffic to URLs laden with query strings: something unlikely for users to generate on their own.
So how do we respond to this decrease in data available? As the Google blog post linked above mentions, Webmaster Tools still provides a list of the keywords that take users to your site. By pairing this with other information from Analytics, we still know what drives traffic to a site, but don’t have as precise a pool of data available, and for understandable reasons.
We do use Google Analytics to give us insight into how users interact with clients’ sites and to help us plot courses of action, but Analytics is just one of the tools we use. Besides, SEO is a rapidly growing field! What worked two years ago or one year ago — or even a month ago — is no longer applicable. Constantly staying on top of changes made to the technology used within your industry, and specifically knowing what these marked changes in data mean are both necessary tasks to stay afloat.
Just a quick note that w3schools is not an accurate sample to conclude the most popular browser from. In their own words, “Note: W3Schools is a website for people with an interest for web technologies. These people are more interested in using alternative browsers than the average user. The average user tends to use the browser that comes preinstalled with their computer, and do not seek out other browser alternatives.”
I imagine Google’s argument (and it’s true to some degree) is that individual search terms don’t matter, rather categories of searches. As long as that “(not provided)” doesn’t creep towards 100%, we can still draw conclusions based on combining Analytics data with Webmaster tools data. The problem is that now it’ll take longer to reach a sample size that is statistically significant.
[…] Missing Analytics Data? It’s (Maybe) Not Your Fault, http://www.searchinfluence.com […]