A/B tests tell you how different versions of your product are performing in the wild.
A/B tests measure one version of a product against another.
Let’s say that you have a great idea for a new onboarding process that will help people get started using your product. Instead of replacing the old onboarding flow with the new one and hoping, you decide to make sure that the new one is really better.
You set up an A/B test where 50% of the new users get the new onboarding process, while the other 50% get the old one. You then measure your key metric—in this case, probably engagement—to see whether more folks in the new on-boarding flow engage than in the old on-boarding flow. As discussed earlier, you might also want a few health metrics to track, like revenue or early retention, just in case your new on-boarding flow is great at getting people on board, but bad at getting them to come back and pay you.
Once you determine which version is better for moving your key metric, you turn off the experiment and move everybody over to the “winning” version.
It is possible to do this with more than two branches at a time, but the more you have, the more users you’ll need to send through the system to get statistically significant results.
Test your website for
- “relevance” to visitors. Does it meet their expectations?
- “clarity.” Examine your text, graphics and design. Designing for clarity creates an unimpeded eyeflow, an unobscured message and a prominent call to action.
- to remove “distractions.” Focus and simplify
- to remove any factors that increase users’ anxiety.
- to increase visitor “urgency”