What is A/B Testing?

A/B testing, also known as split testing or bucket testing, is a method of comparing two versions of a webpage or app against each other to determine which one performs better. A/B testing is essentially an experiment where two or more variants of a page are shown to users at random, and statistical analysis is used to determine which variation performs better for a given conversion goal.

A/B testing can be used to test everything from individual site elements, like buttons and headlines, to entire pages and workflows. In fact, it’s often possible to run several tests at once.

How to conduct an A/B Test

To run an A/B test, you'll need to create two or more versions of a web page, app experience, or ad, each with its own unique URL or tracking parameters. CRM and tracking tools like HubSpot and Segment can help you keep track of these tags and events. You'll then use one of these tools to send a number of your visitors to one version, and a number to the other versions you've put together. You'll also want to create a AB test audience that is your "control group"- these visitors haven't experienced any change in experience, and will create a benchmark for you to compare your A/B test audience to.

Scoping out an A/B test also requires you to create a time frame or audience size to reach statistical significance. Whether you're testing until a certain time frame, or X amount of site or app visitors, this metric will indicate when your test concludes.

As your test is running, keep an eye on results. If one test is significantly underperforming, it may be worth considering turning off that version so that you don't lose out on visitors, or create a worse user experience.

Concluding the test means taking a look back and comparing your results between each other and your control group.

Value of A/B Tests

An A/B test allows you to make data-driven decisions instead of relying on opinions or guessing, and is a way for companies to experiment with different marketing, product, and content strategies to see which performs better. The goal is to get more traffic, engagement, and conversions.

When experimenting with AB tests, you're able to test out hypotheses on increasing user experience, engagement, or leads, without making a significant change to an already functioning and converting ecosystem. That way, you're not damaging the user experience or conversion rate for an entire audience.

A/B Testing Example

Say you're curious if changing the sign-up flow for a website will increase the number of qualified leads (people who might be interested in your product and signing up), but already have a 5% conversion rate for your existing sign up form. To run an A/B test, you'd use an AB analytics tool to send 33% of your traffic to the existing form- this is your control group. Then, you'd create a new sign up form with changes, say omitting the need to input phone number, and send the other 33% of your traffic there. If the other hypothesis is that asking people to input both their company and personal emails results in a better conversion rate, you'd create another sign up form, and send the final 33% of your traffic to that page. Finally, you'd want all three pages to reach "statistical significance"- so you decide to run the experiment until each page has reached 1000 visitors.