Segment Your Way Out of Sadness
You don't need to be an analytics ninja to set up a few helpful segments. Remember, metrics based on averages produce average results.
You don't need to be an analytics ninja to set up a few helpful segments. Remember, metrics based on averages produce average results.
Last week, my team of analysts and I were discussing a client that wasn’t getting the types of results we had expected. This particular client was implementing changes well and clearly should have seen the conversion rate needle move based on those efforts. Instead, the numbers were flat. Something was clearly wrong.
We began to dig deeper. We learned that the client, an online software service, was pushing current customers through a key selling page we were optimizing. Because these customers had already converted and were simply on their way to sign in, it was causing noise in the data, canceling out any movement we might have seen otherwise.
I instructed one analyst, a competitive type who takes great pride in delivering results for our clients, to segment his way out of sadness. The analyst was able to shortly report some very positive results underneath the noise.
He segmented out current customers and was able to analyze optimization results using cleaner data. Simple. And overlooked, more often than you’d think.
Of course, segments can be used to filter out visitor noise, but they can also be used to keep a close eye on specific campaigns and events and to hunt for opportunities. Let’s look at a few ways you can use segments to get a better idea of why visitors do what they do on your site.
First, for the sake of simplification, I’ll refer to segmenting in the free Google Analytics tool, but know that segmentation is available in almost any major analytics offerings. While viewing visitors by segments isn’t new or novel, it’s becoming easier to do.
What Are Segments?
Google Analytics calls them advanced segments. Here is its official definition:
Ideas for Happy Segmenting
Let’s start with a few ideas courtesy of ROI Revolution. Here’s one specifically for e-commerce sites:
And another:
In our practice, we look at several segments by default. The new vs. returning visitor segmentation is always helpful. This segmentation allows us to see how well a site is doing with first-time unique visitors compared to returning visitors. It gives us insight into the length of the buying cycle and can help us determine if a site has credibility problems.
We also do a lot of persona-based segmenting and segmenting by buying stage. But that is another, much longer column.
You can also set up segments to track how visitors might be behaving based on traffic source, such as why one traffic source, such as affiliate traffic, is outperforming your own PPC (define) campaigns. The possibilities are endless and fun, and can be very profitable.
For lead-gen sites, segments can be set up to cross-analyze conversion rate among content pieces that people visit. You can then drill-down and learn why Whitepaper B converts at a higher rate than Case Study A.
Maybe you just produced a few new product demos and videos to sell your goods and you’d like to know if they’re working. In Google Analytics, you can easily set up event segments and isolate the traffic that has actually watched your new video events. Are those visitors indeed converting higher than those who don’t see a video?
Get some more ideas directly from Google and Omniture.
Last Words and Caution
One caution: Segments can be used to torture data into saying whatever you want. Resist the urge to segment and segment and then segment again until you see the positive result you’re looking for. Simply use it as a way to either zoom in on a specific visitor behavior compared to a control of some kind or as a way to filter out unqualified, or noisy, traffic.
If you’re a little bit of an analytics junkie but have hesitated to jump into segments, you’re doing yourself a disservice. Jump into segments, the water is warm. You don’t need to be an analytics ninja to set up a few helpful segments. Remember, metrics based on averages produce average results.
Have any interesting lessons from your segments? Tell us all about it.
Meet Bryan at Search Engine Strategies San Jose, August 10-14, 2009, at the McEnery Convention Center.