Adjusted Bounce Rate from Google Analytics
Google Analytics announced today a new metric, which is really an adjustment of one of their standard measures. In addition to bounce rate, users can now measure Adjusted Bounce Rate. The full blog post announcing the addition can be found on the Google Analytics Blog.
While the old bounce rate had its uses, its real impact was quite limited for many reasons. One of the principal reasons comes from a limitation with Google Analytics itself (and almost any other analytics software). Namely, that the analytics software cannot tell when you leave their site. This has implications for a number of Google Analytics measures, but it has a direct impact on bounce rate.
The unadjusted bounce rate is sometimes used to indicate uninterested visitors – namely, those who end up on your site, but the site is not what they were actually looking for. It has long been recognized that there is a problem with this interpretation. The bounce rate does not distinguish between those who leave your site after visiting a single page because they are unhappy, and those who leave your site after visiting a single page because they have found exactly what they are looking for. These are two very distinct populations, but they were lumped together by using bounce rate.
The Adjusted Bounce Rate adds the element of time to the data interpretation. Essentially, all it does is recognize visitors who spend a certain amount of time on your site. The website owner can set the time. But then Google infers that visitors who stay on the page for a certain amount of time are successful interactions, and those who stay on the page for less than that time are unsuccessful interactions.
There is nothing really new in this approach, as it is a solution that has actually been implement in various Google Analytics hacks. The difference is that Google is now putting it into their reporting, so that it is easier to find without further customization.
It is somewhat limited in that a success is now defined as inactivity, where it is much preferable to interpret user activity.
Not a perfect solution, but it is a step forward.