I’ve heard many people suggest that “bounce rate”, which is one of the standard metrics you see in most visitor analytics tools, is one of the most important performance indicators for your website. That can be true. But I happen to think that if used without some serious thought (which it rarely gets), bounce rate is one of the most misleading and dangerous statistics around. Unless you’ve considered exactly what you’re looking at, I’d advise you to ignore it completely and set up some more meaningful ways of assessing the quality of traffic to your site.
The usual definition is that bounce rate is the percentage of visits in which users view only a single page of your site (if you want to know more about what bounce rate actually is, read Google’s explanation here). Within that explanation are some understated warnings. Google suggests some improvements you might want to make if you’re “seeing a high bounce rate”, but doesn’t say what a high bounce rate actually is. That’s because it can – and should – vary hugely from page to page. It’s easy to think up examples where a 100% bounce rate is exactly what you want. For example, one of the most common reasons for people visiting your website (perhaps the most common reason) is simply to look up your telephone number. If they search Google for “[company name] telephone” or “[company name] contact”, they’ll probably be presented with a link to a page with your telephone number. They click through to your page, which gives exactly the information that it’s supposed to. Brilliant. Except that your bounce rate has just been increased, because who would want to look any further?
The same could probably be said of really good product pages. People who are obsessed with bounce rate are saying that somebody who enters a website at the home page, clicks around a bit, then leaves without doing anything, are good visitors. They didn’t bounce. But somebody who’s sent straight to a product page, says “that’s what I want” and gets straight on the phone to you is a bad visitor, because they only looked at one page and therefore “bounced”.
The only use for bounce rate, in my opinion, is to compare true like-for-like performance, and to be fair, the Google page above does gently suggest as much, if you read it fully. Maybe you’ve made a change to a particular page, and want to see if it’s affected the response of visitors from a certain source. But there are so many potential hazards, even with this, that I don’t know where to begin.
Measure the positive outcomes from your site: the sales, the calls, the downloads. Measure the “quality” of the visitors: those who are from the right part of the world, and perhaps who are new to the site and look around. But never say: “visitors from website A have a higher bounce rate than visitors from website B, so they aren’t as good”.