We touched upon A/B Testing in our previous post, where we looked at some of the reasons for shopping cart abandonment and what you can do about them – stressing the importance of testing, to make sure you know what changes led to the improved performance.
What is A/B testing?
Very simply, it’s about taking an existing web page and modifying it to create a second version of the same page. This can be done by changing a single variable, like the colour of a button (we’ve listed a few more below), or a complete redesign. Half of your traffic is shown the original page, and the other half the new version. The analysis then determines which version of the page performs better in achieving a set goal (i.e a purchase).
Here’s that list of variables – this is by no means everything though!
Call to action (CTA’s):
These are the buttons users press or the forms they fill out – the last step in the behaviour (conversions) you want users to take. Here you can test the copy you are using for the CTA’s, size and colour, placement on the page, etc. Doing this is vital to work out which combination of variables has the greatest influence on conversions.
Headings, subheadings, body copy, email subject lines, etc. Here you can try testing fonts, writing style/tone of voice, formatting, and so on.
Design and layout:
The main thing A/B Testing will help you identify when it comes to design and layout is what are the essential items to keep on your web pages and site. We see it all too often when businesses struggle to identify essential info and so pages are cluttered with too many elements that only distract, confuse, and ultimately drive users away from the site.
This really is about keeping things simple, predictable, and matching users’ expectations. This is where you’ll have to do a lot of research not only on how your site is best structured and how all the pages and content are linked together to ensure the user easily finds what they are looking for, but also looking at other leading websites. This leads us nicely to our next section, the steps to take when performing an A/B test, the first of which being, research…
Performing an A/B Test
We’ll keep things brief.
1. Research – Google Analytics, Heatmaps, and session recording tools, are all useful resources to draw upon here.
2. Hypothesis formulation – here you’re making sense of the data you’ve collected.
3. Creating a variation – identifying the changes you want to make based on your hypothesis.
4. Testing – presenting the different variations to your audience.
5. Analysing results and drawing a conclusion – be sure you are using the correct metrics to measure success. If the test is deemed a success, you can then move on to implementing the changes.
It’s important to point out that A/B Testing isn’t a do-it-once-and-done activity, it should be an ongoing, constant process of refinement.
However, there are some drawbacks to traditional A/B Testing…
As we previously mentioned, with an A/B test your testing two pages against each other and so you need to run several sequential A/B tests (for all the different variations) on the same page with the same goal. As you’d expect, this takes time, and when conducted by agencies it can become expensive – resulting in it being weeks and even months before you see any positive effect on conversions and ultimately profit. But there is another way…
Harnessing the power of artificial intelligence
A more advanced A/B testing method is what’s called Multivariate Testing.
In simple terms, this is where variations of multiple-page variables are simultaneously tested to determine which combination of variables performs the best out of all the possible permutations.
Hopefully, the below breakdown will help make that easier to understand.
Let’s say you wanted to test:
- Headline copy
- Main product image
- Colour of the checkout button.
You want to test two versions of each of the above variables – so two versions of the headline copy, two versions of the main product image, and two versions of the checkout button. Therefore, you’ve got 8 different variations to test. With traditional A/B testing, you’d have to run several sequential A/B tests to identify the best-performing versions.
However, with Multivariate Testing, all the tests can be run concurrently, saving you a lot of time and therefore money too. Allowing you to get answers and implement the changes your website needs, in a much shorter period.
Now, harnessing the power of artificial Intelligence can supercharge this testing method (Shameless plug coming).
Our AI-powered conversion engine is fed up to 92 inputs based on conversion optimization best practices and creates multiple dynamic variations to test in real-time.
Traffic is evenly distributed between the variations and the signals are analysed and scored to quickly identify the highest converting combination for your web page.
This entire process is managed by your dedicated conversion specialist throughout the duration of the experiment to ensure your business has the greatest chance of growing in revenue.
Thankfully, we’re now seeing a rise in evidence-based marketing. Where marketers are basing decisions on real-world results and data, instead of gut feelings, preferences, and assumptions, with those employing this approach reaping the rewards.
This is clearly the huge benefit of A/B Testing, and specifically, the Multivariate (AI-Powered) kind, in that it takes the guesswork out of the optimisation process. As you have well-defined ends/goals with decisions on how these are being met, made by using data. Simple.
If you want to improve the performance of your website and grow your revenue, then let us show you how our AI-powered conversion testing tool can help you achieve your goals.