Split testing is a term that you may be aware of if you’re tech savvy. But, if you happen to have no idea what it means, don’t worry because we’ve created a blog that tells you everything you need to know about split testing and why you should be aiming to increase conversion rates in 2017.
What is Split Testing?
Split testing is when you create multiple versions of something and test them against one another in order to see which one gets the best results. Ultimately, it’s about improving the user experience in order to increase your sales performance.
Many things can be tested during this process including on-site content, search ads and sales emails, but it’s worth making a note of the crucial things you want to test so you’re not spending months testing every little thing.
Things like headlines, calls to actions, images and website content are great starting points as these are the elements that are most likely to have the biggest impact overall. In addition, testing various offers and being vigilant with your methods will ensure you are conducting a worthwhile experiment.
A/B testing is the simplest way to split test. This is where you create two different versions of something to see which version performs best. You can even add extra variants to your experiment which is known as multivariate testing (but we’ll touch on that later).
A/B testing is a controlled way of gathering accurate data and if done right, it’s a lot easier for you to make decisions based on the results you have found. It’s worth testing regularly as well since results can change overtime and it’s handy to keep up-to-date with your research.
What Can You Test?
On-site Testing: If you want to run an on-site test, look at pieces of content about sales, such as your services and/or products.
Off-site Testing: Running a test off-site refers to things like sales ads and emails.
Testing A and B against each other is one thing but testing both of these against your current results will allow you to work out where your campaigns need to improve, which one is the most effective and why one surpassed the other.
These tests need to run simultaneously in order to generate the most accurate results. For example, if you run test A today and test B tomorrow you will be unable to factor in any variables such as the text used and the location of the call to action which could potentially change between the two days.
Testing a variety of offers is also an important factor. For example, if you are offering a discount to people in test A and a discount to test B, you need to make sure that test A always contains the same visitors and test B contains the same visitors. (So that a visitor would never see both offers).
It’s also possible to test things alongside one another. An example of this might be testing landing page A with newsletter A and landing page B with newsletter B. Following this you might then want to mix things up a bit and test landing page A with newsletter B to get even more concrete results.
Multivariate testing follows the same process as A/B testing but instead you are comparing a higher number of variables. Just like A/B testing, multivariate testing measures how effective each version is in order to reach your desired goal.
The site must receive enough traffic in order for the test to be carried out. That way you will have a sufficient amount of data to be able to make your comparisons. These comparisons could be which design is the most successful or which elements have a more positive response from users.
The most common form of multivariate testing is when the experiment consists of various different page elements such as headings, footers, sign-up forms etc. Instead of creating a different design like in an A/B test, you might create three different headings or a few versions of a sign-up form that might be different in length.
From this you will then funnel visitors – known as factorial testing – and this is ideal for sites that receive a substantial amount of traffic on a daily basis. In short, the more variations that are being tested, the longer it will take to gather significant data.
How Does Split Testing work?
Below is a guideline for split testing which you can adopt in your own experiment:
- You can use ‘Content Experiments’ which is part of ‘Google Analytics’ to carry out your own split test. The most useful initial tests should focus on a key goal; this could be a form being filled in, a phone number being clicked or a product being purchased.
- Try and focus on an area that is high in traffic and look for pages with low conversion rates that are in need of improvement. Low traffic pages will take longer to show results.
- Once you have discovered what you’d like to achieve from the experiments, rank them in order of importance so you can prioritise the impact and implementation.
- Create your two variations before conducting the experiment such as changing the colour of a button, the navigational order of something or even hiding certain elements from your navigation menu.
- Begin your experiment and wait for visitors! These visitors will be assigned to either test A or B and how they interact with each test is then measured and compared to deliver the end results.
Whatever your results may be, you can use this experiment as a learning curve for any future experiments you wish to conduct. For example, if your variation works, fantastic! If not, you can apply what you’ve learnt to your next set of split tests.
So what have we learnt?
Split testing might seem a little tricky to get your head around but you don’t need to be a statistician to understand the concept behind the method.
Using a split testing method allows you to make targeted changes in your marketing campaign. This will help you to distinguish between different sets of results and which of the two will ultimately drive the most sales!
Creative site value for your customers is crucial in today’s technologically advanced world, not only for a better user experience but also for building trustworthy relationships. After all, increased sales and loyal customers is at the heart of every brand.
Our Top Tips:
- Make sure you are using Google Analytics (or an analytics platform)
- Find out your current conversion rate for you website and key pages.
- Create a list of changes you think could improve your key pages
- Create the test – i.e. use a Content Experiment.
- Evaluate the results.