There are many articles and ebooks on the virtues of A/B testing out there. What isn’t talked about so much are the hacks that experts use to get the most out of optimization tests. This information shouldn’t be kept secret, so here are a few helpful pointers.
As you are planning a landing page on your website to test, remember the letters A and B.
The first key, “A,” stands for *A*mple time. Depending on your regular site traffic and/or the campaign that’s stimulating traffic, your test may take a while to reach a statistically significant outcome. Run some numbers through an online Sample Size tool and ensure that you’re testing something that’s worth the wait.
Statistics wasn’t my favourite subject in school, but it taught me one helpful lesson for gauging conversion optimization success. Every data set comes with a margin of error, and you can’t declare the results of any marketing as a success or a failure until you know how wide the margin is. Why? Let’s say that you think a lead generation effort failed & netted only 3% in additional business when you expected it to yield 6%. But let’s say that the tallying of results was sloppy, so we actually have a +/- 5% confidence around our campaign results. This means your actual campaign result may be as low as 0% (the lowest it can be) or as high as a whopping 8%, beating your 6% target. The point is that until we have tight numbers with a small margin of error, it’s difficult to use the results we get to make meaningful business decisions. Next time you’re tempted to cite a statistic & make a snap decision on your sales or marketing, ask how carefully that statistic was collected and whether it’s solid enough to put your full confidence in. In an A/B Test, this means waiting long enough for a large enough dataset to collect so that results are statistically significant.
Beyond the Obvious
The second key, “B,” involves test variation (the image/text/layout change trying to beat status quo). In this case, “B” stands for going *B*eyond the obvious. You don’t need to test against an awful page, you need to change an awful page. A/B Tests are best saved for the elements that aren’t obvious improvements. After you’ve taken care of things that are obviously awful, look at the page for things that could help convert visitors, but that you aren’t sure about. Those are the best elements to put through an A/B test, because they need a large sampling of people to verify their merits.
So when A/B testing, remember A) to give Ample time and B) to go Beyond the obvious.