How to A/B test on Google Analytics?
Oh, it's so simple to A/B test, just like reciting the nursery rhymes. Design two versions of a page, divide the traffic between the versions and choose the one that gives more conversions. So simple, isn’t it? What if I tell you “No�. A/B testing is not that simple. In fact, the struggle of A/B testing starts even before you go for the testing. The struggle starts with the confusion in choosing the right A/B testing tool, knowing how to set up the test, and goes till "should I adopt the winner page? It's fishy" In this post, I will try to end your struggle for choosing a right tool by showing you how to use a free tool that is always available for youA/B testing on Google Analytics. I will also give some guidance on how to set up an experiment and when to stop the test. Hopefully, you will end up with a smart tool to start mastering A/B testing. So let's hop in.
Few words before starting A/B testing is always a great learning experience. You can learn so many things by testing different elements and variables. You can use
it to find out whether you should concentrate on niche conversions or go for the broader ones. You can understand which design elements are good for your conversion and which email templates are not so persuasive. So, no matter which element or variable you test, always be clear on your priorities and business objectives. The ultimate goal of improving your conversion rate is always to increase the revenue generation. However, you should also be aware of which conversion goals are more important. For example, if you are working on improving the conversion rate of your newsletter subscription form and even increased it with your CRO techniques. Did you check if it is hurting the sales rate? obviously, the sales conversation are more important than subscriptions. So, always keep in mind that businesses run on sales rate; not on conversion rate. Understand your bottom line and proceed for the winning design that improves your conversion but does not hurt the sales too.
So many conversion testing tools in the market The conversion tool market is full of tools being offered by various companies. Most of the best ones are paid in nature and out of reach for the newbies and startups. Well, there are some paid tools which offer affordable pricing but they are too limited. Google Analytics being an exception is a free yet one of the most popular conversion testing tools. From A/B experiment management to deployment, it offers everything you need to master the A/B testing. So, let's understand how you can use this tool.
1. Choose “Objective for this experiment� Google does not use the separate setup for A/B testing and Split testing. It has a combined setup under a single term- Content experiment.
To find the content experiments, go to Behavior<<Experiments. There you will find a button named “Create Experiment”, just click on it.
Next, you will end up on a screen where you will set up your experiment. ●
First of all, add the experiment name. Say you are A/B testing your account registration form, then provide a name such that you can easily identify which experiment it is.
●
Next, under “Objective for this experiment”, you need to define the metrics you will use to evaluate the results of your test. You can choose the metrics like Adsense, Goals, E-Commerce, Site Usage etc. You can choose any one or multiple metrics at a time. Depending on your objective from the test select the metrics accordingly. - Choose Adsense if you want to test the ad clicks or impressions. - Choose eCommerce if you want to boost the sales rate or
revenue. - Choose Goal if you already have predefined goals like session duration, page clicks etc. - Choose Site usage if you want to improve the user experience on your site by tracking the average page views, time on site and so on.
2. The web traffic division After setting up the objective, you can now divide the traffic for the content experiment. This entry will decide how many people visiting your site will see your testing version of the page. This is a percentage based control. You can decide the traffic depending on your website's traffic and nature of your test. For quick results, you may include your whole website traffic in the test. However, if the experiment is a bit complex and risky, include a small percentage of your website traffic for a safe play. At bottom of the page, you will see the "Advanced Options". Here you can control the division of the traffic by turning on "Distribute
traffic" toggle. By enabling this option, you will assign equal traffic to each variation till the experiment is live.
If you leave this toggle button disabled, then the traffic division will be done on default behavior where it will be adjusted based on the variation performance. Now, set a minimum time for your experiment to run. I would suggest a minimum of 3 weeks would be better to gain a confident result. The same advanced option also allows to you to fix a confidence threshold for your experiment to define the minimum level that must be reached by the page to become the winner. Caution: Don’t select a very high confidence level. Otherwise, your experiment can go very long as the tool will keep waiting to crown the winner and perhaps you experiment might never reach to that confidence level. So, be practical while deciding the level.
3. Configure your experimentThe next job is to add the URLs for all of the page variations in ‘Configure your experiment’ section.
The first URL you have to put is the "Original Page". Enter the URL of the original page and check if it is generating a Page preview. After that, enter the name of your page, whatever you wish. Next is the URLs of your variation pages. You will see multiple URL entry interfaces if you have created multiple variation pages. You can enter the URLs of all of your Variations. Similarly, you have to provide the names for your variation pages accordingly and hit the “Save Changes” button.
4. Setting up your experiment codeNow is the most crucial part- editing your page’s code. If you have set properly installed the Google Analytics tracking codes on your original and variation page, you will immediately see an experiment code in the box. It is a small script that is generated by the Google Analytics after setting up the experiment. You have to copy this code and paste in the head tag at the top of your original page.
5. Review and Start After adding the script in your page's code, the GA will validate it and report if there is any error in the setup. The most probable error that GA encounters is being not able to find the code. If GA is not able to find the code, go back to the page code where you have added the script and re-check if everything is fine. If everything is fine and there is no error message, you are good to go and start your content experiment.
6. Check the A/B testing resultsWhen the experiment completes its running duration, GA will analyze the results based on the set metrics and minimum confidence threshold and declare a winner page. Please note: Run the experiment for at least 3 weeks to get better results with high confidence level. Avoid the silly mistakes of A/B testing. When it comes to free A/B testing tools, Google Analytics is a game changer and reliable tool. Though it has some drawbacks like no multivariate testing and no visual editor it is still one of the most popular tools. For better visual editing and comprehensive targeting, you can try the MockingFish tool, which is available for free one year trial now. Read about "How MockingFish?â&#x20AC;?
to
setup
an
A/B
testing
experiment
on