Google Analytics is a great (and free!) site analysis tool. It’s likely already installed across your web properties. (If not, stop here and go install it!) Callis uses Analytics to monitor and report web activity as part of the SEO services that we provide to our clients.
If someone at your company watches Analytics closely, they should know about a few little-known issues (I’ll stop short of calling them “flaws”) that affect key reports. They relate to data sampling, veiled search phrases, and mobile organic traffic. These issues may lead to misinterpretations of the Analytics data – especially if it’s used to track changes in historical performance.
1) Data Sampling
For over a year, Google has been sampling data for high-traffic site reports. In other words, depending on the amount and complexity of data, Google many not look at every visit to compile some reports. It just requires too much processing power and storage space to provide that level of detail. Instead, Google may sample the data, delivering numbers that might not be spot-on, but should still be statistically accurate.
While you can’t avoid data sampling in Analytics, they do inform you when it’s being used. If your data is being sampled, you’ll see this at the top right of the page. Analytics then allows you the option to decide how many samples are taken: the larger the sample size, the more accurate (but slower) the tabulation. I personally always set the sample size as large as allowed.
2) Keyword (not provided)
A quick look at your top organic search keywords since Jan 2012 will show an explosion of “keyword (not provided)” organic search traffic. This means Google knows a visit came via a search, but can’t determine (or just won’t share) the specific search phrase. Today, if you do a Google search while logged into a Google property (think Gmail, YouTube, Picasa, Google+, etc.), you generate a “keyword (not provided)” Analytics entry on the sites you visit as a result.
Why does Google do this? It may be a privacy issue, or they may hope to monetize this info at some point down the road. Whatever the case, since Google veils a significant and growing percentage of search terms in your Analytics reports, you should understand the implications. If 1/3 of this month’s organic search visits show as “keyword (not provided)”, then for every two searches done on a keyword, there was likely a third search on that phrase. Though this number isn’t exact, it can get you much closer to reality if you need a monthly comparison of keyword traffic.
3) Missing Mobile Organic Traffic
Thanks to Apple’s mobile iOS 6.0 operating system, a chunk of your organic traffic has gone missing. If you were watching closely, you may have seen that your direct traffic spiked up in October, 2012, while your organic traffic took a similar hit. The reason? With the introduction of iOS 6.0, Apple Safari search began using SSL (Secure Socket Layer), and no longer passes on referral information with the search. The same issue has also become a part of the Android operating system side of the world.
When watching your organic traffic, you’ll want to recognize this. One viable approach: look at the percentage of direct traffic in relation to (direct + organic traffic). Assuming it should stay fairly consistent over time, you can use pre-Oct’12 percentages as a baseline. Apply that same percentage to more recent months for an estimate of actual direct traffic. The difference (attributable to the SSL from these mobile devices), can be added to your organic numbers for a more accurate view, and can easily add 5% or more (that number will continue to grow) to your organic traffic numbers. Yes, it’s a time-consuming process. But if you’re trying to show results of an SEO program, it is difficult to walk away from that much legitimate traffic.
These are just three areas where Google Analytics numbers may not be telling the whole story. If you’re looking for help digging through data to assemble good, solid, actionable information, give us a shout. We love doing that very thing!