The Beginners’ Guide to A/B Testing

Imagine you had one clever tool which could send your conversion rate soaring, banish your high bounce rate and win you more click-throughs than you knew what to do with. It might sound like some kind of crazy online Swiss Army knife, but this tool happens to exist – and its name is A/B testing.

Sometimes referred to as split testing, this is a process worth knowing for entrepreneurs serious about honing the performance of their websites and getting the most out of their limited online marketing spend.

If you’re new to the A/B testing game, here’s a quick beginner’s guide to what it is, how it works and how you can get started.

Can entrepreneurs do it themselves?

Yes – you can. It’ll take some time and deliberate effort, but there’s no reason a cash-strapped start-up can’t take on some A/B testing of their own. It might sound dry and off-puttingly scientific, but A/B testing needn’t be a laborious or overly complicated task.  As you learn more about the process you’ll be able to integrate ever more sophisticated aspects – but at an introductory level, A/B testing is easy and can reap great insights and results even at the initial stages.

When things are really humming and you’ve got the budget, bringing in other pros (like conversion-focused copywriters or consultants who specialize in conversions) is generally the better option. But if you’ve got the time, will and some basic technical know-how, this is something worth investing your limited time into (the dividends can be amazing!)

Alright, so what is A/B testing?

A/B testing is a testing method which pits one variable (or a group of variables) against another variable (or another group of variables) simultaneously in order to uncover the factors which have the most positive impact.

A/B testing is often used in conversion rate optimization, when testing ads, landing pages, calls-to-action or even design elements.  By identifying the factors which have a positive effect and the factors which don’t, you can gradually hit upon the perfect blend.

I like to think about A/B testing in relation to the story of Goldilocks; Papa Bear’s porridge (Test A) was too hot, Mama Bear’s porridge was too cold (Test B) but Baby Bear’s porridge (which took the best elements from both Test A and Test B) was just right!

You can use A/B testing to improve every aspect of your online marketing, from the success of your website to the effectiveness of your email marketing campaigns. As a general rule, tests can be split out into two types: onsite testing and off-site testing.

Onsite A/B Testing

This refers to any A/B test you apply to your website itself. Think about all of the different factors at play which could influence how visitors respond to your website:

  • Design elements, like colours and layouts
  • The prominence of sales, demos, testimonials and other social proof
  • The wording of headlines and calls to action
  • The length of the copy and the ways you go about trying to persuade a visitor to take an action

It’s clear that from the words you use to the colours you display and the images you choose, right down to the layout of your pages – every tiny part of your visible website affects how visitors react.  When you start to think about all of the elements at play, it’s easy to start feeling overwhelmed, but going in all guns blazing is not the correct approach.

Instead, start by focusing on the big, obvious factors – things like layout, call-to-action or headline. Before you begin worrying about whether your link text is more clickable in dark pink or magenta, hone in on the key variables that are easier to test.

There’s no need for a complete overhaul in layout – but what critical factors could you play with? Here are some ideas.

Key Onsite Variables:

  • Headlines – The first real sales proposition your visitor will see on your website. This tells them immediately where they are, whether they’re in the right place and can produce a gut positive or negative reaction.
  • Design and layout – Is it easy to find the next step in the buying journey? Is your look intimidating? Does it look too busy, with elements competing for customer attention?
  • Calls to action – Is your call to action clear and prominent? Is it too pushy? Does it look spammy? Is it appropriate for the context of the situation (the stage of the funnel the buyer is in?)
  • Buttons – Are the buttons you need visitors to click on to progress obvious enough to lead visitors through the buying process? Are they inviting with a clear direction? Have you used contrast to make these buttons stand out from surrounding elements and draw the eye?
  • Images – Have you used too many? Are they bland stock images? Have you used images of people to encourage a feeling of connection? Are those people common stock images (trust destroyers!)?
  • Pricing – Are your prices clearly displayed or hard to find? What colour are they? Are they modified with any qualifiers like “Just” or “Only”? Could you arrange the pricing to place prominence on the most affordable option (or the one that means the most to your bottom line)? Could you couple pricing with benefit statements or comparisons to other products/pricing plans?

Off-Site A/B Testing

There are a number of online marketing methods you might be using which can benefit from A/B testing. For instance:

  • Email marketing campaigns (Keep in mind – you often need on-page elements to help you collect e-mails for remarketing, newsletters and follow-up campaigns)
  • SEO (though this is controlled on an on-page level, elements like your title tags or your meta-descrptions will be viewed in a search engine before people click through)
  • Pay per click advertising (paid search)
  • Display advertising (also paid, but different in format – this format relies on advertising that stands out on the pages of other websites and in search)

The split testing process is exactly the same as with on site testing, but you will find that there are fewer variables at play than when someone is on-site, while the offers and pricing you suggest have a greater influence on whether or not someone comes back to your site.

Key Off-Site Variables:

  • Headlines – Do they suck the reader in with a clear hook? Are they personal or salesy? Do they communicate a benefit? Do they stick out among the competition? Do they intrigue, or create curiosity/urgency?
  • Offers – Are you using BOGOFs? Discounts? Coupons? What offers are people most responsive to? Are those offers hitting people at the right stage of the funnel? How might you tweak them to elicit a better response?
  • Calls to action – Are CTA’s clearly displayed in e-mails and off-site communications? Do they suggest an obvious next step (is it clear what will happen when someone clicks through?)
  • Layout/format –Only applicable to visual mediums like banner ads or e-mail marketing, but the way you lay these out to place prominence on calls to action or to draw in the eye can have an enormous impact.


How to Get Started

Having a clear idea of what you want to achieve is essential before you get started with your first test.

I recommend starting out by changing something very simple but very prominent. The colour of your call to action button could be a good, simple starting point – but may not yield the biggest impact. Larger elements like headlines are easy to change and can have serious impacts on conversions, so consider starting there.

Segment your traffic into two streams: A and B. Ensure each stream only sees one version of the button and measure the results to discover which is the most successful. You can use tools like Visual Website Optimizer and Optimizely to make this process easy and smooth.

Don’t get bogged down in making more than one change at a time. Once you start to change more than one variable you are in the realms of multivariate testing which doesn’t always offer such readable or easy-to-implement results. A/B testing is designed to give you clear evidence that X is better than Y.

You might choose to try out two entirely new versions of variables and use your existing site as a control, or you could keep your existing version as A while you pit it against a new contender: option B.

What Should I Measure?

When the results come in, there are a few metrics you should be looking at to give you the information you need. What you measure will depend on your goals and the medium you are looking at, but a few ideas for places to start:

  • Bounce rate (Are people coming in and immediately leaving? This could point to several problems: Ambiguous copy, poor or untrustworthy design, slow load times, unclear branding/messaging)
  • Conversion rate (How many people are taking the desired action? If the number is low, you can look into other elements like layout, calls-to-action and even the copy that’s driving the action)
  • Sales (How many people are buying? In what frequency? Has the buying cycle been shortened? Where are people dropping out before the sale?)
  • Click through rate (If you are running an off-site test, how many people are following your titles/headlines/advertising back to the site?)

The version with the most powerful conversions and sales and the lowest bounce rate should be the clear victor. I recommend implementing the changes as indicated by these results before moving on to split testing new variables.

How Long Does it Take?

That depends.  You need numbers large enough that the changes can be statistically significant – changing your variations every hour on the hour will only lead to confusion over poor data.

For brand new start-ups and entrepreneurs, this can be a while. That may actually work to your favour, as you won’t need to frenetically set up test after test after test – just pick a few key elements, and roll with it.

For most sites with steady traffic, around 2 weeks should be more than enough time to harvest enough data to be statistically significant. Timing is very important in this process. An A/B test which is too short risks not having enough data to be representative. An A/B test which is too long is vulnerable to lots of different variables, from seasonality to changing trends.

An A/B Tester’s Work is Never Done…

A/B testing may provide empirical data that you can use to hone your website, but the results are never finite. The digital world is a fast-paced place where chance can occur rapidly. That’s why it’s important to keep on testing and re-testing to ensure you are on the right track.

And there’s always more you can tweak. In fact, after running a few A/B tests you might find yourself hooked. Once you are happy with the big, headline variables, there are no end of small tweaks you can try with lower level variables to really start sweating the small stuff until your marketing materials and website are pure perfection.


Leave a Reply

Your email address will not be published. Required fields are marked *