An Introduction to Split testing/A/B testing

One of the often overlooked aspects of website ownership is establishing whether your pages are getting the best returns and conversion rate possible. You may already have pages that are successfully collecting email addresses or perhaps making sales. But, can these assets be improved to increase your results?

Thankfully, we can analyse our site\’s performance and optimise it further towards our conversion goal with the use of split testing or A/B testing. In short, split-testing is a method, which allows you to run multiple variations of a page side-by-side to see which one performs the best. 

A well planned and executed split-test or experiment can dramatically improve your marketing efforts, leading to a more profitable and successful business. To do this, we must first break down what makes a good test. Here are the main things to consider when first getting started.

1 – Have a Strong Offer – What is it and what can it do for me?

One of the first things to consider is to start from the beginning and to understand exactly what your customers want. Is what you\’re offering relevant to your customer and is your message clear? 

In short, a bad offer will not improve your results, regardless of how optimised your page is. But, a well-optimised page with a good strong offer will dramatically increase your bottom line.

If you\’re currently stuck on what to offer your customers, spend some time understanding your market; what their pains and problems are, and how you can offer a good solution to their issues. Places to look at:

• Forums
• Facebook Groups
• Q&A Sites
• Competitor websites/sales pages
• Your website\’s analytical data.

2 – Create a Hypothesis

Conducting an experiment without a hypothesis is simply a guess. One of the biggest errors often made by business owners when running an experiment is to test a specific thing because “They think it will work“.

The best testing will come when you have an open mind and test a variety of factors, not just the things you think will work. 

Your test hypothesis is the presumption that forms the basis of your test. It details the element on your page you want to test and what outcome you want to achieve. 

So for example, I want to see if I can improve the number of subscriber conversions on my landing page, by testing the current button colour against four other variation colours. 

Once you have run the test, you can then examine the results to determine if an alternative button colour presented an uplift in conversion. 

You can learn a lot from this valuable data in addition to the successful result. For example, you might discover which colour doesn’t work for your chosen niche if a button colour performs much lower than others. 

A clear hypothesis for each experiment will help you to achieve the best results for the tests you run and you will be better equipped at conducting focussed and smarter tests with little guess work.

3 – Testing Types

When starting, you will also want to identify the type of test you will want to run. This will help form the basis of which elements to test moving forward.

Offer Tests – This type of test will help you determine if you have a strong offer for you website visitors. For example, are you testing, the opt-in offer or the price point of your product? You may decide to test multiple variations at different price points as the first stage in the testing process. This test is ideal for confirming whether your offer is the perfect fit for your market as covered in Point #1.

Structural Tests – Are the elements of your page targeted to better serve your audience? Typically, this type of test will include testing of headlines, images, sales messages, colours and other on-page elements, which we will cover shortly.

4 – Decide on a clear and focused goal – test one thing at a time

The main aim of any page is to \’achieve one goal\’, whether that goal is to opt-in to a web form or click on the buy button. If the action you want your visitor to take is confused this will have a negative effect on your conversion rate.

The same single focus applies to creating your test. You must establish the goal of your test. The purpose of identifying your test goal is to provide you with the most definitive answer.

For example, if you tested multiple elements on a page in a single test, button colour and headline copy, you would have 2 metrics being measured in the same test. Testing 2 or more metrics in the same test would make it hard for you to attribute the page conversion to see whether it was affected by the headline copy or the button colour.

A clear and focused test in this instance would be 2 separate tests.  For example, button colour on a page in the 1st test, which button colour converts best on the page, a green button or a blue button? Then, once you have determined that result you could create a 2nd test that would test variations of the page headline.

4 types of testing goals that a page can focus on are:

– Audience Engagement – How long do visitors spend on your site?

– Opt-ins – How many visitors complete your web form?

– Downloads – How many people download a product?

– Sales Conversions – How many people visit a shopping cart or complete a purchase?

As you begin testing, your results and overall success will generally determine the percentage of these goals. The higher the percentage, the more successful your experiment and ultimately, your business will be.

5 – Things To Test

Testing different elements of your page is crucial if you want to better optimise them for your audience. Since no audience is created equal and is dependent on the market you\’re targeting, many factors will need to be considered in order to determine the success or failure of your business. For this, various elements will need to be tested. To help you get started, we\’ve listed the main ones for you below:

Please Note: In order to gain the maximum benefit of your experiment, it\’s vital that you \’only test one thing at a time\’.

1) Headlines – Your headline is arguably the most important element of your page as it\’s the first thing your visitor and potential customer will see when first landing on your page. To get the most out of your headline, you\’ll need to ask yourself 2 important questions from the perspective of your customer: “What is it?” and “What does it do for me?“. If you\’ve followed the advice given in step #1, this should be very straightforward.

2) Call To Action – Like your headline, your call to action has to stand out and make an impact as it\’s the key thing that will get your visitors to take an action. Test out different colours, button styles and words. A few examples could be “Continue“, “Buy Now“, “Download\’ etc…

3) Images – In short, your images need to compliment your page and should not distract your visitors. Ideally, pick an image that will
encourage your visitors to take the action you require. Perhaps, a face looking at the opt-in form or an arrow pointing to your call to action.

4) Colours – Your choice of colours can make or break your page and may or may not fit your potential market. Test different colour variations to see which combinations make the best impact.

5) Sales Message – Does your message identify with your customer? Again, the better your understanding of your market and what your customers want, the easier your sales message will be. Spend some time experimenting with different sales messages to see which ones stick.

6) Guarantees – A guarantee is crucial for building trust with your customers. Test different guarantee variations: \’30-day Guarantee\’, \’60-day Guarantee\’, \’90-day Guarantee\’, \’100% Guarantee\’ etc…

7) Video/Text – Test different page styles. Your potential customers may prefer a video sales page instead of a long form sales letter. 

6 – Test Periods

When testing you need to allow for a fair and constant test period. This means allowing the test to run for a set period that will allow enough visitors to iron out any anomalies such as, public holidays, weekends, seasonal events, summer holidays or religious celebrations.

These events might impact the test and provide unusual test data and mean an unusual or non-typical result.

7 – Analyse Your Results

Once you\’ve performed a few experiments, you will inevitably be faced with decisions depending on whether your tests have succeeded, failed or flatlined. It\’s important that you keep a note of your stats to help you understand why your experiments have given you that particular result. Your visitors and customers are your best research tools and will often tell you what will need to be improved on your website.

If a test produces a negative result, it is not a failure. It\’s valuable data that will help you better understand your audience. 

Conclusion

Whilst we have covered what makes a successful experiment, it is important that you\’re constantly testing new things and keeping a finger on the pulse of your business. The more experiments you conduct, the more of an understanding you\’ll develop for your market, leading to more conversions, more sales and ultimately, the long-term success of your business.