THOUGHTS & INSIGHTS BY THE TEAM AT FRANK DIGITAL
Share this
View our site

Split Decision: A Simple Look at A/B Testing
Share this
View our site

Split Decision: A Simple Look at A/B Testing

Recently, we’ve found ourselves in situations where opinions have differed. From whether to use a specific page payout, or a certain image or headline, the views of the agency and client will often clash. It’s your job as a marketer to make an informed decision backed by solid A/B testing data to optimise your website conversions.

A/B testing isn’t a new concept, in fact, A lot of clever marketers and designers have been using it for years to gain insights into visitor behaviour and to increase conversion rates.

Fortunately, A/B Split Testing takes the guesswork (and the bloodshed) out of many website decisions. Even if you’re not a tech guru, you can use real-life data to boost conversions and balance creative aesthetics with performance. That’s good news for your bottom line (and for your designer).

What is A/B Testing?

A/B, or Split Testing is a controlled experiment whereby two versions of a webpage are made available, and their effectiveness compared. Traffic is split evenly between the two, with site visitors seeing only one version each. Ever seen Sliding Doors? It’s a bit like that, with your visitors entering parallel universes. But unlike for Gwyneth, only one option delivers a better user experience, more conversions and a happier ending.

It’s as easy as A/B… see?

First, you need the numbers. There’s no point if your site only attracts a few visitors per day. To achieve statistical confidence, you need at least 1000 guinea pigs (or people – up to you) to visit your site during the testing period. Setting up the test sometimes involves fiddling with JavaScript or HTML code; it depends which tool you use.

Popular tools include Visual Website Optimizer, Optimizely and Google Analytics Content Experiments – which is free. For email campaigns, try Campaign Monitor or MailChimp. Once the test begins, monitor the results using metrics that reflect your business goals. Are you aiming for more clicks? Conversions? A longer dwell time? Testing can take days, weeks or months; it depends on your site traffic.

Typically, testing is most powerful when used to isolate a specific variant, such as the position of a ‘Buy Now’ button. That way, one option is usually revealed as the winner. If there’s no significant difference between the two, you can flip a coin – or let the boss decide. Then it’s time to update your site.

What can I test?

Start small, with something simple – like headlines, or button colours. Then once you get the hang of it, you can test almost any aspect of your webpage. If you have a large audience, you can test a larger number of variables. And for most businesses, it’s the bigger changes that have more impact on the bottom line.

You can use A/B testing to compare:

  • Simple vs complex page layouts
  • Shorter vs longer content
  • Single-page vs multiple-page checkouts
  • Copywriting styles
  • Graphic design styles
  • Different product descriptions, services or pricing

A/B testing is addictive

Yes, we’ll admit it: testing is pretty fun. For your marketing strategy, it proves what’s working. You may even decide to evolve your business model, adapting products or services in response to visitor behaviour. But a word of warning: don’t test just for the sake of it. Align your testing with your business goals. Keep your user experience in line with your brand values. And consider the big picture: a ‘FREE’ button may get lots of clicks, but is this good for your business in the long term? If not, give it a miss.

A/B testing enables smart decision-making at every level. If you could A/B split test all life’s choices, would you have still taken the same road? Perhaps. But either way, A/B testing is a tool that gives you hard evidence so you can change conditions and adjust your content with confidence.

You may also be interested in

Want to know more about us?

Why not check out our work, or view another article!