Our Guide To A/B Testing On An eCommerce Website

When it comes to building a prosperous eCommerce store, there’s one element that will make or break your success – conversions.

Conversions are the lifeblood of running an online business. Selling products and making money is what will keep your business going, so selling more is of course always beneficial. Conversions are the same across any industry and have a direct impact on the profit line of a business, so what can you do to increase the conversion rate on your store?

Well, one of the most effective strategies you can implement is A/B testing. Chances are you’ll have heard of it (possibly under the guise of ‘split testing’) but if you haven’t, here’s a brief explanation.

What Is A/B Testing?

A/B testing is a way of presenting your audience with two alternate versions of a single webpage, allowing you to compare the effectiveness of specific variables within that page. In terms of eCommerce A/B testing, this would be a way of testing which version of the page delivers a higher conversion rate.

Every page within an eCommerce website will promote some sort of action, be it signing up to a service, subscribing to a newsletter or (most importantly) purchasing a product. Testing different variants of that page give you the ability to understand how you can go about increasing the conversion rate of that action.

As an overview, performing an A/B test allows you to maximise the conversion rate of your existing audience for a specified action without the need for increasing your users. Simple.

Types Of A/B Tests

There are a few ways you can complete an A/B test on your eCommerce store, but here are some of the most popular:

• Altering one element at a time
• Altering multiple elements at a time
• Offering two completely different page layouts

All these methods can bring varying levels of success but recommended best practices involve changing just one element at a time.

Why? Purely because if you change multiple elements on a page and see a huge increase in conversion rate, you won’t know which of your changes delivered that increase. It, therefore, makes more sense to test and change individual elements separately.

Why Is A/B Testing Worthwhile?

In my opinion, you would be doing your own business a disservice if you failed to undertake any form of split testing. Their sole purpose is to increase both conversions on your site and improve the overall user experience, so why wouldn’t you want to give it a try?

It’s fair to say that the results can sometimes be surprising too – we’ve seen examples where even minor changes to text or wording have resulted in significant gains, so ignore this type of test at your peril.

In some situations, you may want to try something more significant. Offering your audience two completely different iterations of the same page can radically change conversion rate, both positively and negatively. This kind of test can also have an impact on other measurables like time on page and number of adds to a cart which can again fuel your thinking on future page designs.

What Specifically Should You A/B Test?

There is an endless number of different elements within a web page that you can A/B test on your eCommerce store, but here are a few ideas to get you started:

Calls to Action – this could relate to the button shape, size, colour and placement as well as the text included within it.

Images/Videos – this involves the addition, subtraction or movement of images and/or videos around the page to increase conversion.

Body Content – a change to the position of body content, either moving above or below the fold, can directly impact conversions; even changing one or two words can have an effect.

Navigation Structure – some stores have found that a reduction in the number of navigation options has led to an increase in conversion rate.

The ideas we’ve mentioned here are more significant, but you could also test smaller elements such as prices, promotional offers and free delivery to identify how dramatic the increase in conversion rate could be if you rolled these changes across the entire site.

Don’t forget that it’s not just what you present on your page, it’s how you present it. You can test the influence you have on your audience by making changes to font, formatting and styling as well. An aesthetically pleasing page is likely to convert at a higher rate compared to the same content poorly organised and lacking in structure.

How To Complete An A/B Test

Undertaking split testing is easier than you think, and there are a number of good paid and free solutions available on the market. One of the simplest to use is through Google Analytics and they posted their own guidelines on the do’s and don’ts of split testing.

Creating a test environment is very straightforward as long as you have the ability to duplicate a page within your eCommerce platform. Simply insert the code provided by Analytics into the head script of the original page and your split test can begin. If you’re not particularly coding savvy, you might need the help of a developer to complete this without causing other issues on your site.

If your store is built on Magento like many of our clients, we’ve found a great guide on how to set up A/B testing on your site courtesy of Amasty.

Step-By-Step A/B Testing Plan

Step 1 – Research areas that you feel could be improved, maybe using your closest competitors as a guide.

Step 2 – Hypothesise something to be tested. This could be a simple as ‘help more customers find the delivery information by moving it into the main navigation, therefore increasing the likelihood of conversions’.

Step 3 – Determine the variables, remember that this is a strict test that must be completed using the proper controls if the data is to be reliable.

Step 4 – Work out the length of time and number of visitors needed for a satisfactory test. You can use an A/B test calculator to help you with this.

Step 5 – Launch the test! Start recording the data and build your understanding.

Top Tips To Ensure Your A/B test Is Successful

• You’ll probably need a fairly considerable testing period to get some decent data, at least two weeks is recommended.
• Any page you test should get sufficient traffic – there’s little value in testing a page that only gets a handful of visitors each month.
• Don’t add any promotions on your site unless that is the sole purpose of the test. Any results you do get will be skewed if there was a sale running at the time of your test.