Last fall, Google began shielding logged-in users’ search terms from analytics. Plenty of sites found that their top keyword became “not provided.” Though it didn’t prompt a wholesale shift to Clicky or Bing’s Webmaster Center, many of us are wondering what that change did to our level of accuracy.
You don’t hear so much about Internet Explorer’s InPrivate browsing, but it does much the same thing. Users of IE8 and up can choose to browse privately without leaving any traces. So can Chrome, Safari, and Firefox, though only Safari seems to have many users of their version of private mode (one source claims that 14% of Safari users say they use private mode). Google also has an opt-out add-on for any browser that lets people avoid leaving any trace, and of course any machine without Java enabled won’t be tracked for Google analytics.
The big question is: does this affect the value of your web analytics?
Your analytics give you lots of useful information about who visits your site and how they behave once they arrive. You can find out where your visitors are geographically, how they found your website, what links they clicked to arrive at your site, which pages they visits, what keywords they used to find you, and much, much more. This helps you with usability issues for your site, gives you valuable information about the effectiveness of your marketing, and lets you know something about who you’re reaching.
If you have a fair amount of traffic, this probably doesn’t matter. It removes some of your sample, but not necessarily any particular group. For example, the typical visitor to our lab site is using a commercial, school, or government network and comes from an English-speaking country. If it were the case that thousands of people from German speaking countries and using corporate networks all chose privacy options, we’d have a false idea of our typical visitor. However, we think it’s likely that our private visitors are similar to the ones we can see.
When might that not be the case? Well, researchers say that 8% of people visiting porn sites use private browsing. If yours is an adult content site, you are very likely not getting accurate analytics. You may also be getting more visits from what How-to Geek calls “tin foil hat types.” This could be significant if you have a political site, or even an ecommerce site for some types of products.If you’re running a website for privacy advocates, you probably don’t use analytics anyway, but there may be related issues that would be affected.
If you have a small amount of traffic, it’s more of a problem. Small amounts of data are less accurate than larger amounts in any case, and they can easily be skewed. I’ve read that 90% of all websites get fewer than 10 visits a day. If your site has a handful of visits, losing data on any visitors might make the information you have much less accurate.
On the other hand, if you have just a few visitors, you probably aren’t making good use of your analytics anyway.
If you’re concerned that your analytics are being thrown off by privacy features, you could ask your visitors for information. Explain why you want to know and set up a poll asking them to check off some harmless data such as what browser they’re using. You could also offer them a white paper or something in exchange for information you can see in analytics — that’s the only way to get email addresses now, so you could ask for your visitors’ zip codes at the same time and compare that data with your analytics to check . Unfortunately, this is self-selecting: people choose whether to share their information with you, so you may only discover that visitors from Orem love whitepapers. In other words, you’re back to the same problem.
To my mind, any data is better than no data. We know that a lot of the information we work with is imprecise, but we are usually watching for changes over time, so “pretty close” can still provide actionable insights.