Anything But Average
September 09, 2016
A recent episode of 99 Percent Invisible discussed the history of calculating the “average” and the implications of using average calculations. It is an interesting episode and well worth a listen, but it made me think about how many times I’ve personally relied too much on averages, as well as the many clients I’ve worked with who are relying too heavily on average stats.
On the one hand, averages are a convenient benchmark to know how a broad set of things perform. An average number gives you a rough idea of how to then evaluate the performance of a more specific thing. On a website, it is handy to know the average bounce rate of the entire site or the average time on site because then when you evaluate a particular page’s bounce rate or time spent on that page you can know, generally, if that page is doing better or worse than the average.
However, averages are rarely actionable numbers on their own. It is easy to look at your average bounce rate across various sources of traffic and worry that some sources are higher than others. You want to take action on the problem, so you jump in and start making changes to your website, trying to reduce the bounce rate based solely on this average number.
Of course, each page on your website is anything but average. Sure, the overall average bounce rate might be 84%, but some pages are going to perform considerably better than others and have a lower bounce rate and some pages are going to perform somewhat worse. However, as 99 Percent Invisible’s podcast made clear, nothing is average. The same is true with your website. Taking action on average is, at best, dangerous. Instead, you want to look at a much more specific number. In the case of your website, the better number to look at is not the average bounce rate across the whole website but rather the specific bounce rate of a particular page.
That page-specific bounce rate is actionable. You can use the site-wide average bounce rate as a guide to determine what pages have higher than average bounce rates. Knowing that, you can work to reduce the bounce specifically on those pages. Even better, you can learn what leads to a lower bounce rate by reviewing the pages with lower than average bounce rates. The same is true for any other metric across your entire website; nothing is average, but the average provides a rough benchmark that can help you group high and low performers.
The other place where averages can wreak havoc is when comparing your website’s performance to the average of all websites. It is interesting to know an average conversion rate across multiple websites. Knowing this can help you determine how you rank in comparison to everybody else. It seems like it is a rough approximation of whether or not you are performing okay. Here again, though, nobody is average and benchmarking your performance off these average numbers is, at best, troublesome.
Some websites will just naturally have lower conversion rates. Larger ticket items, highly competitive industries, products/services with longer buying cycles, newer/untested products, etc. will all generally skew toward low conversion rates. Sure, you can (and should) learn some ideas or tricks from high performing websites, but each website has its own collection of unique peculiarities that will make that website’s conversion rate anything but average.
Instead of relying on the average conversion rate across many different websites as a benchmark, you have two other, far more specific options. First, you can look at how your competitors are performing. Your competitors will probably have more in common with your website, sharing many of those same unique peculiarities. As a result, their metrics will offer a more specific benchmark to compare against your own website; “are you better or worse than your competitors” or “what is the average among my competitors” are much more actionable questions.
Even more specific than competitor performance metrics as a benchmark is to rely on your own historical metrics as a benchmark. As you make changes to your website, does your conversion rate skew higher or lower than it was before? If higher than before, keep those changes. If lower, ditch those changes.
The point here is that you’ll find more value in taking action based on specific metrics, as opposed to taking action on a broad average. None of this is to say you shouldn’t use an average, but just be careful when you do. Use the average as a guide to learn some general trends of what kinds of things tend to go above or below average, but when taking action on your website you want to use the specific numbers, instead of an average, to make decisions about what changes to make.