Marketers and the Fallacy of The Bounce Rate
I am in Toronto now for SMX Analytics, largely moderated by Danny Sullivan and Vanessa Fox.
I am so surprised to hear the consistent references to high bounce rate as problems. This seems irresponsible and short-sighted, as though a user who only accesses one page on your site is a failure of some kind.
Consider Google’s “I’m Feeling Lucky” Button
This search option is there for when you think you know exactly the destination in mind, or just want to add a bit of chance to the search. Regardless, I would argue that at least 25% of searches are users just looking for one page that has all the information they want.
One Page Visits That Should Not Be More
- “Definition” searches. This is a common motive (see my 10 Motives of Search).
- Contact info - no reason for two pages here. I’m feeling lucky!
- Wikipedia searches. The pages on that site tell a story from A - Z on one massive scrolling page.
- “First Touch” comparison, research searches. If you are price comparing, or skimming through options in order to round up a bunch of sites to consider for a purchase, you are less likely to go deep on the first touch. A bounce here just means you were considered.
The fact is that people want to get where they are going fast, and leave fast, unless you have the unique privilege of running a site like Facebook or ESPN that has a valid reason to engage people. I am not saying bounce rate has no value as a data source. I am saying it can be worse than no value, it can give you entirely the wrong picture. Be careful with bounce.
Measure Impact, Not Bounce
Instead of bounce, consider:
- Time on site
- Referring keyword (was it a “one night stand” keyword? Maybe they found what they were looking for and have no more reason to come back)
- Return visits - maybe the bounce was a first touch and the user is coming back.
Tags: analytics, bounce rate, SEM
March 31st, 2009 at 8:58 pm
Avinash wrote a great post about bounce rate:
http://www.kaushik.net/avinash/2007/08/standard-metrics-revisited-3-bounce-rate.html
I like his second definition, which is not visitors who visited a single page, but rather visitors who abandoned the site instantly (less than a few seconds).
Any metric can be problematic and should be viewed in context. Any metric can be valuable or lead you down the wrong path, depending on the context. Time on site for instance, can easily go up if the server slows down and things take longer to load, but in that case a time on site trend bar up and to the right wouldn’t be a great thing.
April 1st, 2009 at 6:04 am
Hi Vanessa,
Yes, I like his second definition as well and you are absolutely correct that all the indicators have to be taken in context, which is why I cited Wikipedia and their ONE long scrolling “entry page”.
However, I have a hard time believing that page load time is going to frequently affect an analysis of TOS, and I’ll bet you a Molson it would have the opposite effect. Sites that have high enough page load times to significantly affect TOS are going to frustrate users quickly, resulting in them leaving the site and therefore decreasing the TOS. Or maybe those that stay and those that don’t just zero each other out.