There are many rules about website testing. I’d like to offer one more that is often overlooked by marketers when they evaluate the success of their campaigns: Don’t stop a test too early based on initial results. This may seem obvious, but the reason I mention it goes beyond what’s sometimes easily visible by looking at the top-line results of a test.
Most tests don’t move the needle one way or the other, and plenty of others end up being outright losers. It would be easy to simply disregard these tests based on analyzing top-level results and move on from there, but doing so can lead you to miss valuable opportunities.
When you test against an overall website population, it’s often the segments of traffic that show where there may be a potential winner out there. Looking at traffic broken down by behavioral, geographic, and technographic variances will often lead to new testing opportunities—and ultimately testing wins.
The trick is to allow your tests to run long enough for the individual segments to achieve statistical significance. A test might already be significant against the overall population, but smaller segments may need more sessions to achieve the same level of significance.
Take this case, for example: A marketer wanted to see the impact of hiding social media plug-ins from all traffic, only to find the initial results were flat. Further analysis, however, showed that this idea worked better for new visitors than return visitors. To confirm this premise, the marketer targeted the test to its new visitors only, and the positive results held fast.
Want to find hidden opportunities like this in your tests? Here are a couple of suggestions:
• Before calling a test, you really want to look at one or two additional segments to see if there are significant results against a niche that you might not have considered. Start with new vs. return visitors or mobile users, and work back from there.
• If the results are promising against a segment (or segments), you may want to allow the test more time to fully bake—or at the very least, test the new hypothesis on its own.
• Another benefit is the opportunity to gain new information that can inform your overall testing strategy. For example, you may discover that mobile users are a valuable segment to target distinctly.
Ending a test too early can often kill the wholesale value of running that test. With a little patience and the ability to look at a test through the filter of a number of visitor types, you can often pull a winner out of your hat—as well as develop the insight that helps you get a better return on your testing investment.
Spencer Lewis joined Monetate as a Client Success Director in January 2012. Since then, he has worked with a variety of clients to turn their website merchandising and testing fantasies into reality.
Leadership Potential image courtesy of Shutterstock.