Accuracy and Google Analytics

Exact figures are hard to come by, but some experts estimate that on average, almost one site in every three has Google Analytics tracking code on it. Even if only a small percentage of those webmasters pay regular attention to the information GA provides about visitor numbers, popular pages, and traffic sources, that still amounts to millions of sites relying on GA for their analytics data.

There is no doubt that Google’s package is the industry leader. It’s thought that Google has an 80% market share when it comes to web analytics. When it comes to free analytics, the figure is probably even higher. But just how good is the data provided by GA?

From a mathematician’s perspective it’s not that good at all. All professional web analysts know that Google Analytics significantly underestimates traffic in the overwhelming majority of cases. Anyone who has ever compared the stats from their web server with the stats from Google Analytics knows that the two sets of numbers rarely match up.

There are several reasons why, but the most significant factor is that the GA tracking code only fires if the person loading the page has Javascript enabled. That means that search engine crawlers and other automated web crawlers don’t get picked up, which is great- for the most part, web analytics is only concerned with the activities of real people- but it also means that anyone browsing with Javascript disabled will be completely invisible.

Somewhere between 10 and 15% of us regularly use the internet without Javascript. For GA and other tag-based analytics solutions, those 10 to 15% just won’t show up. As a result, Google Analytics usually significantly underestimates overall traffic. There are other minor accuracy issues, but that’s the big one.

However, it’s not the same as saying that the figures you get from GA have a 10 to 15% error. As long as the percentage of invisible visitors stays roughly the same and they behave much the same as other website users, the stats returned by Google Analytics are still representative of what’s going on. If it says that average page views went o up 20% from May to June, that figure will still be about right.

Although they are presented as real and exact counts, Google Analytics statistics like Visits and Pageviews should be considered metrics, not measures. They reflect the visitor numbers and total pageviews, but they don’t provide precise figures. It’s the change in each metric over time that provides the most valuable information, not the exact figure itself. Knowing that your site had exactly 16892 visits in June is one thing, but in most cases it’s more useful to know if that’s more or less than you had in April and May, or June last year.

As long as the mix of visitors stays roughly static and measurement methods are kept the same, GA’s figures can provide meaningful insights. It would be a mistake to assume that they are perfectly accurate down to the last visitor but if used with care, it’s still possible to get excellent information from Google Analytics.