I’ve been using Google Analytics heavily for about six months now (was using Omniture previously) and have come across some quirks after seeing data that didn’t make sense to me when trying out different things. Here are some tips that I wish I had been aware of when I was starting out.
1) GA source/medium override – search > referral > direct. We have a section of our site highly optimized to rank for first+last name keywords. I was seeing these keywords show up as top organic keywords to our homepage. Clearly this wasn’t correct. It was only after looking into the issue further that I found out that if a user returns via a direct visit to the site after their initial organic visit, that direct visit will be reclassified as organic and assigned the keyword from the initial value stored in the utmz cookie. If you want to address this issue, you can set the utmz cookie to be deleted when a user session ends.
2) Data sampling issues. The more you drill down into your data, the more at risk you become for sampling issues to surface. If you start to see several items all sharing the same metric value, you’re probably running into this.
3) Custom reports vs. segments vs. advanced filters. Segments only segment at the visit level, so if you want to know how many pageviews a particular section received based on the rules you define, you’ll need to use a custom report. For example, if you’re trying to define a segment using some RegEx on the Page parameter that matched a certain category of your blog, say www.someblog.com/categories/google_analytics/, you’ll be getting data that you may not have expected.
4) Profile best practices. You can setup filtered profiles, but always have a profile containing raw, unfiltered data for each of your accounts in case something goes awry. There have been several occasions where I’ve needed to go back into our data to search for a certain query string parameter or other item that was stripped out of the data in our default profile.
5) Metrics definitions. Google Analytics visits are really entries, and visits to particular site sections or content in the sense that Omniture uses the term are more akin to unique pageviews. Time spent invested in understanding the GA data model and metrics definitions will pay large dividends.
6) Cross-domain tracking. As my friend Kashif Zahoor pointed out, this is a biggie. I also had a nice chat with Loren Bast from the Cheezburger network about this issue. It involves tracking users properly as their move across websites you own, or even subdomains within one website. If you’re seeing a large number of self-referrers, there’s probably an issue with your cross-domain tracking. At my company, I noticed something wasn’t right when we updated a subdomain and suddenly bounce rates jumped up for that particular section – users were being seen as exits when they traversed from the subdomain to our main site. There are several articles on the subject, but a good start would be the Google Analytics Dev Guide.
7) Goal limitations. Goals can only increment once per user session. Why this is, I have no idea. But if you see events not lining up with conversions, this is the reason why.
Have any thoughts or additional tips? What do you wish you had known when you were just starting out with Google Analytics? Please share in the comments below!