Ab testing articles

A/B testing is a way to measure two different options to see which one performs better.

Commonly also called split testing, it can be a powerful way to make data-backed decisions.

How A/B testing works

In a nutshell, A/B testing takes two variations of something and shows one variation to half of your visitors/customers and the second variation to the other half.

Then over time the interactions are counted until the end and the better option is choosen.

It has to use special statistical formulas to make sure the better option is actually better or was it just a fluke. That's why an A/B test calculator or tool is needed to verify the results.

Almost any can be A/B tested...

A/B or split testing can be used to test almost anything. All you need are two or more variations, time, and the ability to track the end results.

Time is the most common problem though.

You need enough time to collect and measure behavior to the level the statistics mean. Depending on the test, that could take months, years, or decades.

That's why many split tests are done on high-traffic pages or by large companies. They collect enough behavior to rapidly test the variations.

... but most things don't matter

Some newsworthy split result results report improvements just by changing button colors or a small amount of text.

They might work but the majority of A/B tests are failures. The new variations were worse or just about the same as the prior version.

This means the most value in split testing is found with changing major elements. New headlines, calls-to-action, whole new pages, or hold-out tests are common variations tested.

A/B and split testing can be powerful optimizations when given enough time.

Articles for further reading

The newest articles are at the top.

Would you like a daily tip about Shopify?

Each tip includes a way to improve your store: customer analysis, analytics, customer acquisition, CRO... plus plenty of puns and amazing alliterations.