A/B tests are a great safety net when you can use them

I've been working on a new design for Repeat Customer Insight's pricing page.

It's been functional and describes what each level of account receives, but it was designed a long-time ago when the app had fewer features and only needed a paragraph or two.

Now that the app has matured some, the design isn't working for it and sowing confusion.

Nick D from Draft who does a lot of conversion optimization often recommends A/B tests to make sure design changes end up improving your store. But A/B tests require a level of traffic that my app's pricing page doesn't meet.

Instead I've used another tool Nick's given me for low-traffic websites:

Heuristics

Or in easier words, rules-of-thumb and expert opinions.

I've been using my own experience as a developer (and halfway designer) along with the opinions of designers I trust to help guide the design. The design will probably be better and measurably better, but it's a risky change without the safety net A/B tests provide.

It'll be what I call A-then-B test... first I see how one version works (old version), then I replace it with the new one and compare.

(and yes I know that's cheating statistically-speaking but at least it's giving a cool sounding name to what a lot of people end up doing)

If you have the traffic levels though, real A/B tests can work very well. Back when I used ads, I used A/B testing and got my conversion costs way down (50-100+% ROMI for the ad-geeks out here).

But the real power with them comes from testing parts of your store and having the hard data to backup design decisions. Which is something Nick D has done quite well with.

Soon I'm hoping I can share the new pricing page design and outline the major conversion elements that I changed. Some might directly apply if you sell subscriptions but even if you don't, there are many standard conversion principles at play.

Eric Davis

P.S. In case you're wondering, this is purely a design change for Repeat Customer Insights. The actual prices aren't changing at all, just how the page looks.