Think of Conversion Rate Optimization (CRO) as a toolbox that today’s businesses use to improve their websites and turn visitors into customers. One of the most powerful tools in that box is A/B testing, which is exactly what it sounds like: an experiment that tests two variations of a web page to determine which gets better results. If all other things are equal, the winning variation should have users doing what you want them to do more often. Done effectively, A/B testing can raise your website’s ROI by decreasing your cost per acquisition.

The benefits speak for themselves, but not every business is quite ready for A/B testing. If your business has less than 1000 transactions per month – purchases, signups, leads, etc. – your test groups probably won’t be large enough to be statistically sound and, hence, the results might not be reliable.

If you are ready for it, however, A/B testing needs to be dialed in to make an impact. Here’s how to use A/B testing to make your business stronger. 

1) Cover the basics first.

Many smaller businesses are better off investing time and effort in other areas until their numbers increase. But just because your site may not meet the threshold doesn’t mean you can’t improve it.

Optimization, for instance, isn’t only about getting more conversions, it’s also about understanding why you’re missing out on them in the first place. Understanding your users and what they are trying to accomplish will help align your business goals with their objectives. Dig into the data to discover where users might be dropping out of your process – and fix the leaks.

If you look into Google Analytics and notice a lot of people are dropping out of your conversion funnel at a certain page, dive deeper into analyzing what may be causing doubts or hesitations on that page. Ask users, or observe session recordings, to try to discover why that might be happening. If something is clearly broken, don’t waste time trying to test – just fix it and stop the bleeding.

2) Know what to test.

The short answer to “what do I test first?” for almost every business is the unique value proposition (UVP) – the top-of-page experience that should quickly convey what a business does and why it’s better than the competition. In short, it’s the compelling reason why a user should stay on the page and convert.

Seems simple, right? Not exactly. There are a number of UVP elements to test, including the primary image (or “hero image”), titles and subtitles, the intro paragraph, and any benefit bullet points. Other elements include calls to action, visual media, navigation, personalization, and offers/pricing.

3) Have a hypothesis.

An effective A/B test starts with a hypothesis; basically a theory about what isn’t working and how to fix it. If you have a number of theories and fixes to test, prioritize them based on their likely impact on conversion rates.

To dial in A/B tests, consider factors both on and off the page that affect conversion. Aside from basic data gathering and analysis, analyze customer interview (“what they say”) against website session recordings (“what they do”), compile qualitative data with on-page surveys, and use both heat maps and user testing to gather as much information as possible about customers and prospects before forming and prioritizing hypotheses about how to best motivate behavior.

4) Know what success looks like.

The indicators of success should be laid out clearly before the test. Each hypothesis should be designed to influence a measurable variable. A typical hypothesis might look like this:

If [variable to be manipulated], then [expected measurable result], because [justification of expected answer from research].

Or, a real-world example: “If the unique value proposition is adjusted to mention our 100% guarantee, then more users will convert online because this will reduce user anxiety – since our customer interviews have indicated this is the No. 1 concern with online leasing.”

In this example, the online conversion rate would be the No. 1 metric; but other secondary/correlated metrics – such as call rate, bounce rate, average time on site, pages per visit, and other considerations – can be observed as well.

5) Know how to turn lemons into lemonade.

If proper research has been done, the hypothesis should be strong enough that a new variation should improve the conversion rate. If it doesn’t, however, the opportunity to research why your hypothesis failed may be worth the gut punch to your ego and bottom line.

Why did users respond opposite to what was expected? What did you learn about their motivations and behavior? How can you better match their expectations and objectives? Go back to the drawing board and start interviewing your customers.

Like any other effective tool, the important thing to remember with A/B testing is to use it wisely and as often as necessary to keep getting better. A/B testing is an ongoing process, and there is always room for improvement in the rapidly growing and evolving digital marketing landscape.