How to Make the Most Out of A/B Testing?

Landing Page Academy » Chapter 4: Landing Page Optimization » Lesson 6/7

We've talked about image and copy optimization. We know what generally works better, but as I mentioned at the end of the previous lesson, even if your landing page now seems perfect, that doesn't mean it will stay that way forever.

Making changes to your landing page based on a feeling would probably get you nowhere – intuition hardly ever works in cases like these. To properly improve your landing page, you need data. Make way for A/B testing!

What is A/B testing?

A/B testing is a method of experimenting with landing pages (or ads, posts, and so on), in which you go through a three-step process:

• Create two or more versions of your creation,
• Make a single change to one of the versions,
• Launch them simultaneously,
• Compare their results and pick the better one.

It looks simple, and it is, but to make the most out of A/B testing, we need to go through each of these steps.

How to create two variants of a landing page?

Usually, it's as simple as hitting the "duplicate" button – almost every landing page I've enjoyed using had such a feature, and after duplication, we've got two identical landing pages.

Now it's time to make a change. Yes, a single change. It's crucial for getting precise results. If you decide to change two elements at a time, you won't know which one caused the difference in conversions.

Pick a single element and prepare a variant. If the headline seems to be the main problem, rewrite it and do nothing more. If you're not sure where to put the form, create a second version of the landing page with the form placed elsewhere, and that's it.

Hint: if you've got many visitors (a few thousand rather than a few dozen), you may try creating more than two variants, but still – stick to testing a single element.

You'll be launching the test in this step, but you need to set up some options before that.

One: how do you want to divide the traffic? The most popular option and the one that makes the most sense is to split it equally. If you have two variants, show them to two halves of your visitors. If there are three variants, you may want to limit the risk and show the "old" version to 50% of people while displaying the test ones to 25% of visitors each. It's up to you.

Two: when will the test end? If you run one test and your priority is to get the best result out of this one, or the change you've made is a huge one, I recommend setting this up to a certain amount of conversions. That way, the test won't end until X people convert.

On the other hand, if you're running A/B tests regularly, you may want to finish this test and run another on schedule. If that's the case, set the timer for a few days or a week, depending on how many visitors you expect, and then pick the winner.

Launching the test

Comparing the results and picking the winner

That's the most interesting yet the easiest part of A/B testing. After some time (or after achieving a certain amount of visits or conversions), you'll get the results.

They are unforgiving for the losing variants – the data is in percentages and quantities, so there's no room for discussion. This means you may just announce a winner and apply the change to your landing page.

Keep in mind, though, that even if variant B wins the test, it’ll be wise to monitor the results from time to time. The sole fact that it was better than the previous version does not necessarily mean that it will be the best solution for your landing page in the future.

Before you launch a test, you should set up a goal. Usually, it will be the same as a landing page goal, but they may differ sometimes.

The most popular A/B test goal is increasing the conversion rate. But how to determine which part of a landing page is responsible for the underperforming conversion? You can go back to Lesson 1 of this chapter – I've written about finding the pain points there.

The same goes for the bounce rate. If it's too high, you should optimize the first section of a landing page or take a step back and focus on an advertisement that directs to the landing page – it, too, can be A/B tested.

If you run an eCommerce business, your focus is most likely on more purchases. Optimizing a landing page for sales may be more demanding than just A/B testing its headline. There might be a problem with the form, but also with payment gateways or the offer itself. Try different approaches, integrate your landing page with more payment options, focus on benefits – do they fit the target audience's expectations?

Another goal of the A/B test may be getting better leads. It's usually all about the form. You can quickly test the form by adding a field and getting more data. Why shouldn’t you add it without the test? Because asking people to share more information about them may hurt your conversions.

A/B testing goals

Ground rules

To make the most out of A/B testing, you need to follow a few additional ground rules:

Take some time for testing – you won't get reliable stats after two days of testing, but on the other hand, there's usually no point in having the test enabled for a month.

Test one element of the campaign at a time – if you're A/B testing a landing page, don't make any changes in the e-mail, advertisement, or banner associated with that campaign.

Get some feedback – if there are no clear results from the test (they are similar or highly unexpected), ask the users how they like the version they've seen. It may seem tricky to execute, but hey, you can always send them an e-mail and see what happens!

To make the most out of A/B testing, you need to follow a few additional ground rules:

Take some time for testing – you won't get reliable stats after two days of testing, but on the other hand, there's usually no point in having the test enabled for a month.

Get some feedback – if there are no clear results from the test (they are similar or highly unexpected), ask the users how they like the version they've seen. It may seem tricky to execute, but hey, you can always send them an e-mail and see what happens!

Test one element of the campaign at a time – if you're A/B testing a landing page, don't make any changes in the e-mail, advertisement, or banner associated with that campaign.

Takeaways

1. Set up a goal before starting an A/B/x test.

2. Test one element at a time.

3. If you're not sure about the results, you can always run another test.

4. Keep in mind that there's no such thing as a perfect landing page. It can always get more conversions and bring better leads.

What's next?

In Lesson 7, we'll focus on checking how users behave on landing pages. What they like to see, which elements are more "clickable" than the rest, and how to use this knowledge in landing page optimization.

Privacy Policy

Terms & Conditions