Split testing, or A/B testing for your ads’ landing pages is an advanced but pretty reliable method of scrutinising and improving how your website interacts with visitors and achieves your visitor goals. Split testing essentially looks to monitor and improve the % of visitors that accomplish goals. Goal conversions that sites can optimise for could be:
- Completing a product or service purchase
- Downloading further information such as a white paper or E-book
- Newsletter sign up’s
A home page or landing page that is correctly tested will essentially lower the bounce rate, increase the amount of time a visitor spends on the site and ultimately boost revenues.
So what is split testing?
Split testing is a system that lets you create two or more variants of a particular page and allows you to test them to ascertain which is the most popular. Essentially we want to test a variety of page combinations and see which one delivers the greatest conversion improvements with regards to conversions i.e. activities that are competed and measured as a goal or metric value. The key thing to bear in mind when doing split testing is there are two types of testing that can be done with regards to ad campaigns. The first is Adwords Experiments and the second is Content Experiments.
- Adwords Experiments These are centred around A/B split testing with regards to keywords, ad and ad groups within your adwords account.
- Content Experiments For A/B/C or multivariate testing of specific web pages and landing pages where we you use Google Analytics and synchronise it with your Adwords account.
Content Experiments are the type of split testing we will be looking into for this article. So when you test landing pages using Google Analytics you are able to test up to 10 variations of a page, each one containing different content and layout. When you launch the experiment each page will be shown in equal proportion to the other and rotated through to people clicking on your ad. Then impression share can be optimised to ensure the page with the highest performance gets more impressions over time. Then monitor your landing page experiment for around 4 weeks and Analytics Content Experiments will let you know which landing page has completed the most goals and achieved the most campaign metric improvements.
Before you start you need the following:
- Set up conversion tracking goals in Google Analytics and make sure its synced with your AdWords account
- Have different versions of your page to be tested each set up each with its own URL
- Get E-commerce tracking set up (if required)
Which features to test & measure in Content Experiment split testing:
- Use different headlines, copy wording and general writing tone
- The layout of your copy
- Experiment with different background colour schemes, text colour etc.
- Placing images in different places and using different images
- Style and size of various page elements such as enquiry form, call to action button, videos etc.
- Add different contact details e.g. general Contact us, more specific details of the in house person to contact directly
Step by Step
So first of all you need to embed the tracking code for Google Analytics and set up all the goals. These goals can vary from which URL they visit an event goal e.g. call back request, enquiry form fill out, overall visit duration goal etc. Once you decided upon these elements (URL visit goal, event goal, visit duration goal), it’s time to get split testing.
Next to get access to the Content Experiments feature, go to Standard Reporting – Content – Experiments
Next up, enter the URL of the original landing page that you want to test then click on ‘Start Experimenting’. You will be taken to a page where you will be required to name your experiment and add the URLs of all the home or landing page that variations that you want to test.
The next step is ‘Set your experiment options’. Here you’ve got to confirm the type of goal that you want to be improved and measured. At this stage you’ve also got to state the percentage of visitors that you want to run the experiment on. You may find it handy at this stage to write notes to remind you later about the experiment’s purpose especially if you are running a few at the same time.
Next you will be provided with some important code to insert after the <head> tag of the control page’s HTML. This bit of code informs Google Analytics how to display all the landing page variations to your visitors.
Finally this control page complete with the code script must be uploaded to your server. The simply check all the landing pages are working correctly and start your testing!
Testing and measuring your landing page experiment
It takes Google Analytics a couple of days to show you the first results from your experiment – now you are on the way to optimising your conversion rates. There is nothing worse than running a test over a prolonged period and finding at the end it wasn’t set up or reporting correctly so make sure you check the data is coming okay.
So to test this, enter the url of your control page into your web browser and you will see one of the page variations will appear. If you are running an experiment with a lower than 100% of visitors, refresh the page a couple of times to see if the page rotates around.
If you want to double check the URL will have a different ID number indicating it’s a variation. Luckily for you the whole testing and measuring process is predominantly done by Google Analytics. However don’t take this as an invitation to do nothing whilst it’s running
You need to regularly check in and see what data is coming back in. Obviously the amount of data you are able to aggregate depends upon the overall visitor traffic from your ads. So make sure your budgets are set to send some visitors through your funnel.
So what are the key metrics to track?
- The experiment summary on the right hand side – this overviews the total visits across all landing pages, the running period and experiment status.
- Conversion Rate graph – this compares the conversions across each page over time
- Individual page visits and the impression share percentage – this provides a good indication of which page is performing better than the others and getting greater exposure by Analytics
- Probability of Out-performing Original – this tells you which ads are most likely to replace your control landing page by the time you reach the end of the split testing experiment.
At the end of the experiment Google Analytics will arrive at an overall winner based on all he performance metrics. Armed with this information you can select this as your new landing page confident that it is guaranteed to increase your conversion rate and the all-important return on investment.
From this point forward you can use this as the starting point or control to conduct more landing page split tests if you think more improvements and increased ROI can be achieved.
If you feel you have achieved as much as you can currently achieve you may consider having a look at other high traffic ads that may help you improve conversion rates and run Content Experiments on them.
Generally your campaign landing page quality and conversion rates contribute to improving your Quality Scores which can reduce budget spend and increase the all-important profit.