Thursday, May 21, 2009

Why bounce rates are important...and not important.

There’s nothing like a solidly contrary title to start off a web marketing blog. One of the most fascinating and frustrating things about web analysis is that every metrics is relevant in certain contexts, and irrelevant in others. Depending on your business and the segment of traffic we are examining, metrics like time on site, page depth of visit, even goal conversion are totally different in terms of importance.

The example this blog will be using is bounce rate.

According to the Web Analytics Associations standards document, bounce rate is the ratio calculated by dividing single page visits by entry pages. In plain english this means that a bounce occurs when someone has a single page session. They entered your site on a specific page, and then left the site without going any deeper.

Your overall bounce rate number is the percentage of people who had a one page visit to your website.

I have had discussions with multiple firms in the last few weeks who were concerned about their ‘bounce rate’ numbers, specifically the overall number.

Overall bounce rate numbers are at best a health metrics for your web business. On it’s own it doesn’t really mean anything, but it can be indicative of areas of interest that actually do mean something.

Example 1#: One of the firms I am working with asked me to examine bounce rate issues as they were well in excess of 50%. I put together two quick segments, one for their ‘true conversion’ segment (they only sell in North America) and one for ‘rest of world’. The bounce rate for their true conversion segment was much lower than rest of world.

So when you look at it, high bounce indicated success. They were able to relay information to non-prospects that allowed them to move on and not waste their time. They were also able to relay information to real prospects to let them know they should stick around and engage the site.

Example 2#: An eCommerce company I work with asked me to examine bounce rates as they were super high, in excess of 80%. As bounce rates are a combination of pageviews and first page of session data, a good place to start to see real numbers is under “Landing Page” reports (which is what Google Analytics calls this type of report).

Over half of all visits started on the homepage, and the bounce rate was much better than expected at around thirty percent. The other half of their traffic initiated visits on something like 27,000 other pages within the site...and all of them had bounce rates in the high nineties! It turns out that this site generates a ton of natural search traffic from visitors searching for very generic non-branded terms (think ‘jogging pants‘ instead of ‘Roots Yoga pants’), which implies a lack of buying intent in the visitor.

So in example one, bounce rates showed that the website was doing a good job at speaking to its primary stakeholders. In example two, bounce rates helped identify a significant underperforming segment of traffic in unbranded natural search visitors.

Moral of the story: Bounce rates are important as early warning signals that allow you to start asking questions of your analytics. Bounce rates are not important as a success metrics for your website.



This week’s takeaway: Look at your overall year to date bounce rate in your analytics tool. Make a list of the top three reasons you think you have this rate, and see if you can prove/disprove with data.