vector - share icon

We’ve been asked a few times recently to explain the concept of split testing and why someone would consider it.

Here we’ll delve into the murky world of systematic visual changes and tracking…

Have you ever wanted to work out what works best for your clients? What images or content instigates the most responses from your clients?

Well that’s where split testing comes in.

Split testing is a way to refer to the practice of sending one version of something to a segment of a target market and sending another version of it to another segment.

Each segment should be proportional and ideally representative of the whole target market to ensure that the results are worth listening to and the feedback will tell you which version of the content (or which set of differences) has been more successful – providing you valuable information going forward.

If that’s a mouthful which is difficult to understand, let’s break it down a bit more:


Proportional Representation

For anyone that has studied data analysis or statistics, this should be easy.

A ‘proportional’ segment of a target market (in this instance) would be a segment which is big enough to be considered representative of the whole market.

So, if we had a case study of 100 individuals that we wanted to understand which stimuli instigated the biggest response, it would be silly to consider the response of a single individual. The segment would be considered too small.

Ideally, you would want to record the responses from all 100 individuals in 2 categories – a 50:50 split.

That’s easy in this example because we know that the target market is 100 people and it’s quite a manageable number.

In reality, when you are working with much bigger numbers, you:

  • wouldn’t know the actual size of the whole target market and
  • be able to record the reactions of every individual within the target market.

So, you have to use a ‘best guess’ scenario.

In our example, we could consider the responses of say 20 of the 50 people in each segment as representative, because 20 would constitute a large chunk (40%) of the overall tracked market.

It’s more manageable to be able to get 20 responses than 50 and the sample is much bigger than working with the response given by a single individual.

In a real-world situation, you could be working with samples of 100, 1,000 or 10,000 for example, which will give you good feedback on what really works in the situation you are trying to test.

You can’t improve things you don’t test and measure.

By split testing, you gain valuable feedback which you can use to influence future decisions – for the betterment of the campaign(s).


What sort of things can I test with split testing?

There are all sorts of scenarios where split testing could be beneficial.

However, we will work with digital marketing relevant examples – which are generally based around getting a reaction or some sort of engagement from your target market.

This could be testing the subject line of an email campaign, to see how much of an impact that makes on the open rate of the email, or tracking the content or layout of a web page to see which entices visitors to engage more – either by sharing, commenting, bookmarking or filling out a contact form for example.

You can even test different types of call to action to see which type or text generates more contact requests for instance.


How do I split test something?

In its most basic form you would simply identify your target market, interact with half of it in one way and interact with the other half in a different way and record the outcomes.

However, things aren’t always that straight forward. Fortunately, marketing tools are making it easier for people to carry out split testing and here are 3 useful tools to help you:

1. Email Marketing – Mailchimp

Mailchimp is the undisputed king of email marketing for small business. Available for free, Mailchimp gives you access to a whole host of powerful marketing tools which will help you make the most of your email marketing for your business or venture. These include (but are not limited to):


2. Web content / Layout – WordPress

When using WordPress, it’s easy to implement split testing with plugins. A free one we’ve started using is:

  • which allows content to be displayed to X number of visitors in one fashion, then the same content (or different content depending on what’s being tested) to be displayed to another X number of visitors. The information is then able to be used to confirm which format works best for the target market of the page.


3. Online Ads – Google Adwords

Ultra important because it actually can save / make you £1000’s, split testing the effectiveness of your Google Adwords ads is essential to maximizing the ROI of your campaigns.

By split testing your ads, you compare them and chose the most effective* in order to then compose new ads and compare those against the previous winners – constantly looking to improve the results of the ads.

*most effective can be measured in a variety of ways. It may be most clicks, least cost per click, best customer lifetime value as a result of the click or a million other things.

Split Testing Summary

Split testing is a powerful marketing technique to be used to improve the effectiveness of your marketing overall and in specific areas.

It can take some time, so should be included within your overall marketing strategy rather than as a standalone project, because each item split tested should then be split tested again against new, improved versions to constantly improve content and its results.

Marketing tools such as Mailchimp are making access to this kind of technique ever more available and if you are not using this to improve your results, you can bet your competitors are and will soon be leaving you behind if they are not already.