Our thoughts are with those affected by the bushfires across Australia. If you have been impacted and need help with your account, please call your Account Manager.
About the author
Share the love


A Basic Framework for Conversion Rate Optimisation (CRO)

We all know most websites, landing pages or pop-ups can be improved and optimised. Of course, there are the simple, ‘low hanging fruit’ options such as creating a stronger call to action, making contact details visible etc., but it’s important to follow a simple structure in order to get the best results from running a split test. After all, how do you REALLY know that these things will work and make a difference? That’s why we need to test.

1. Research and gather data

Before running a split test, it is vital to gather information about the website or business you want to test. Whether that is your business or someone else’s, think about the information that is the core of what the business stands for: the ‘why’. Where does the company want to go, what is the purpose of the site and importantly what are the USPs (Unique Selling Points) of the business? (These will come in handy later.) Jump into the Google Analytics account to determine where the traffic is coming from – you may be surprised to learn that most of your traffic is coming from a mobile device. Hopefully your goals will already be set up so you can see where different areas of the site are converting. You are looking for patterns in the data and any trends that can provide clear areas of improvement within the data. If you don’t already have these set up, you can learn how to add these here.


2. Create a hypothesis or goal

Now we have a foundation and basis for the website and customers, we can work towards a goal of what we may want to split test. Think about what you would like to improve: perhaps you feel customers could benefit from a safer payment area in your checkout? We can even segment this traffic to a specific user that we may want to target; for instance, perhaps new users and subscriptions for unique sessions is something you want to improve upon. Most testing software such as VWO and Optimizely allows you to do this and much more.

Where do you want to test? A specific page such as a contact page or landing page or even two pages where you may want to test the titles of the tag line or the price and benefits of the products against each other.

3. Propose an alternative design/flow

This is where the fun begins. Now that we have an idea of how to test, it’s time to design a test or set of tests to run. We have a team of designers here that can do all this work for you and liaise with you in order to implement design changes that suit your brand. Some testing tools may have restrictions in making drastic changes to the designs so both you and our team can bear this in mind when creating a split test to ensure it can actually be tested. For instance, perhaps we can test a blue ‘Add to Cart’ button instead of a red one – trust us, we have seen huge results from small incremental changes.

4. Create and build the new test

Now to build the test, you just need the software code installed to get started. The more complex the design, the longer this may take for our team to create for you. Set up the targeting to your channels you have decided on. If you have high traffic numbers, we can set traffic distributions accordingly – for example, enable 30% of your traffic to see the test while the remaining 70% will continue through the test as normal and therefore not be picked up as testing participants. Once a browser testing has been conducted and completions and designs have been previewed and checked, we are then ready to activate!


5. Review results to clarify hypothesis

After receiving a solid number of sessions via the testing, we can review whether our test was successful in increasing conversions or not. Most testing tools come to a confidence level for you, however, Google Analytics will also show the number of people via each test and the conversion rate on both. It can be hard to predict statistical significance in testing, but the more the better. A few thousand unique visitors through each is a good start; however, there are online calculation tools to help you to arrive at a decision that best suits you.

Once the test wins, we can roll out the test design to a higher number of visitors, if not all traffic, and test again.

If the test loses, which will happen, we digest our learnings and create an alternative design as long as you are keeping in mind what failed last time.

Conversion Optimisation is an ongoing process as we can always optimise to increase to conversions and the look of your site. Follow a simple process and good things will come from your online users!

By Charlotte MacInnes

Get in touch!




PayPal Certified Developer
Facebook partner
Adwords qualified individual
Adwords partner
Google partner
Australian web industry association