Split Testing Best Practices For Website Optimisation

August 5, 2021

Split testing can be a very valuable tool when looking for solutions for website problem areas. When performed well, the results can drastically improve the conversion rates and yield maximum returns on investment (ROI). This is because the results are dependent on the actions of real users. Thus, you will also be able to gather deeper insights on the behaviour of your audience. Split testing is also a great tool for optimisation work by continually implementing positive changes to the website. As split testing is very problem-driven, any time is a good time to conduct a split test experiment. In this article, we will share with you the split testing best practises to maximise results.

Split Testing Best Practises

Step 1: Conduct Research & Collate Data

Firstly, you would have to determine which areas of your website are not performing as well as you had anticipated. To do this, you would have to collect data on all aspects of the website, such as traffic to each page, bounce rate of each page and what are the conversion goals of each page. Consider utilising website analytics tools such as Google Analytics is valuable to furnish you with the relevant data.

To add on to the quantitative analysis of your website, you may also want to conduct qualitative assessments. Both qualitative and quantitative assessments together will provide you with a comprehensive outlook on your website performance. Qualitative assessments include using a heat map tool to understand scrolling behaviour and determine where users are spending the most time on. Another qualitative way of assessment would be to conduct surveys to uncover user problem areas of the website.

Step 2: Determine Problem Areas

It would be easier to determine your website problem areas once you have collected all the data necessary. Here, we are trying to see if the website goals match up to the website performance. Make this comparison on a page by page basis to find out which pages have the largest disparity.

Then you can start making a list of the problem pages, from highest to lowest priority. Within these problem pages, highlight the problem areas and start listing out the elements of each page which is the most problematic for you in terms of conversion.

Step 3: Create New Versions

Once you have identified the specific problem element on a particular page that you want to address first, work on creating 1 to 2 new versions of that page, but with only one small change. You would only need to create 1 to 2 different versions so that you can evenly divide enough traffic between them to provide you with conclusive results for your split test.

Step 4: Conduct Split Test

Once you have decided which website element you would like to test and your tracking goals for the element, you may commence the test and wait for the results to come in. Eg, if you are testing the success of your CTA button, clicks could be the most effective goal to track.

Usually, the test will run for 1-2 weeks in order to be able to extract definitive results. There are various A/B Testing Tools that can help you with all your testing needs and find that optimal solution to the issue that you are facing.

Step 5: Analyse The Results

Most of the A/B Testing tools will come with a reporting feature that collates the results for you and systematically presents the data for your review. Based on your audience size, determine if the winning results are statistically significant. If it is, you may either choose to use that version, or you may optimise it further by tweaking another element of concern and conducting the test again.

In other cases, your test results may be inconclusive, which means that the changes made were not striking enough to produce a remarkable difference in behaviour in your target audience. You may want to run another test but with bolder changes. Eg, changing the CTA button from light blue to dark blue may not result in significant changes. You may want to create a version with orange or green CTA button instead.


What Are Some Of The Things That You Can Test?

i) Headline & Body Ad Copy

Headline Copy

First impressions are crucial in marketing. Your ads are the very first thing that your visitors will see. Hence, this sets the stage for what your website is all about. Your ad copy could be a determining factor that decided if the user wants to continue engaging with you or not.

  • Make sure you headline is representative of who you are.
  • Keep your headline concise
  • You may want to test various font types, font size, copy and messaging

Body Copy

This copy should resonate with the headline. Make sure that your headline and body clearly tells your prospective visitors what is in store for them. Use the right tonality for your messaging and ensure consistent formatting for an easy read.

ii) Design & Layout

Have a clean layout and only keep the most essential elements to be displayed.

  • Pay special attention to critical pages such as your home page and landing page
  • Always display high definition images
  • Showcase relevant video to capture user attention

If you are an e-commerce store, the performance of all your product pages are crucial to securing conversions. Make sure that all the elements on your product page is orderly and the product description is easy to understand and does not confuse your visitors.

  • Provide accurate and clear information, especially when it comes to stock availability, sizing guidelines, etc
  • Feature some customer reviews
  • Keep the content the simple – Do not confuse your buyers
  • Create a sense of urgency so buyers will make the purchase immediately – Have tags like “Only 2 left in stock” or “Offer ends in 45 minutes”

iii) Navigation

Have a definite user journey for your users. Create a pleasant user experience where visitors can complete the user journey most successfully.

  • Be clear on your website structure and how each pages are linked to one another.
  • Ensure that there are no broken links and users are able to easily find what they need
  • Match user expectations by having a predictable structure, such as have the horizontal navigation bar at the top and vertical navigation at the left
  • Stay consistent with your theme for all pages

iv) Forms

Optimise your forms to your target audience. Most businesses utilise the forms as a means for their customers to get in touch with them. If this is a critical part of your sales funnel, place some emphasis in optimising your form. Split testing best practises would be to only ask for information that is most necessary. Visitors objective is to complete the form as fast as possible. You may then request for more information once you are able to get in touch with these prospective customers.

v) CTA Button

This button is integral to any website. It is where much of the action takes place. The performance of your CTA button has a direct correlation to your conversion rate. Conduct tests on the copies, button colour, button placement and even button size to ascertain the best version for your website.


Avoid these Amateur Mistakes in Split Testing

1. Referencing Other Website’s Results

Just because you know of a website which experienced a drastic increase in conversions by changing their CTA button does not mean that you will experience the same result. Each website has unique elements that are specific to that particular website such as unique target audience or specific conversion goals. Take note that no two websites are the same, and that what works for one website might not necessarily work for yours. Study and analyse the data for your website to make the pertinent changes to your website.

2.Testing Too Many Elements At The Same Time

Testing too many elements of a website at the same time makes is rather difficult to precisely figure out which element was the most significant in influencing the results. Furthermore, the more elements that are tested, the more traffic is required to produce valuable results. It is essential that you pick out your problem areas and prioritise them before conducting split testing.

3. Disregard For Statistical Significance

Split testing is conducted as a way to eliminate guesswork. If personal opinions were to be incorporated in the test, the results would automatically be deemed as skewed and biased. Therefore, it is important that you take into consideration the statistically significant results in every test. If the results yielded are inconclusive, Consider Step 5 again.

4. Conducting A One-off Split Test

Split testing is meant to be conducted regularly, in an iterative manner. This is because the insights derived from previous tests are meant to influence the direction of any future tests. Test each element repetitively to ensure that your website it running at its most optimised version of itself.


Split testing is an easy way to optimise the performance of your website. Our guide to split testing best practises may seem overwhelming and time consuming in the beginning. However, with consistent effort the work will eventually pay off and you will have an easier time managing a truly optimised website.

Split testing is a really effective way to determine how best to connect with your audience. Let us provide you with deeper insights on how to conduct split tests that is suitable for your campaigns. Reach out now as we love to identify critical patterns that may be hard to uncover. 

The Truth About Vanity Metrics

The Truth About Vanity Metrics

Vanity metrics are metrics that make your campaign look good.In truth, vanity metrics can be misleading and ineffective in determining the ...
Social Media Advertising For Small Businesses

Social Media Advertising For Small Businesses

Social media advertising helps your customers to discover, learn and purchase your products & services. It is integral to your business ...
In-Depth Guide For Manual & Automatic Ad Extensions

In-Depth Guide For Manual & Automatic Ad Extensions

Ad extensions help you to provide more information about your business. We show you how to set up ad extensions to benefit your business.