It is quite usual in digital marketing the need of finding the right strategy or option to start a marketing campaign in a website. Since there are many alternatives, it’s not easy to make up our mind about which colours, slogans or images are the best for it. One way to optimize results and check the different options is by using A/B testing.
What’s A/B Testing?
In the online world, A/B Testing is a tool or method that allows the final user to have different versions of the same website. This way, he/she can check which one is more attractive. The goal is to know, with a simple experiment, which version makes the user spend more time browsing, which one makes him/her to fill a web form. This testing is broadly used for digital marketing campaigns. And most of the times, the reason to use it comes from the disagreement between the responsible of those campaigns.
Defining the experiment
When proceeding to do an A/B test, first thing is to know what we want to test: the colours of the CTA (Call to Action) button, the website’s slogan, the header… We’ll need to offer to the final user a different version for each item.
Also, it’s important to know how to show the variations. We cannot show two completely different versions for the same website. The more different they are, the harder will be to find what is really attractive for the user.
A/B Testing tools
There are countless tools in the market, some of them free, to do the A/B testing in a website. There are some that allow you to completely design the HTML content with custom editors. Others, however, provide you measurement tools with performance reports.
One of the most well-known tools in the world is Google Analytics. This free tool from Google allows us to tailor the experiment, defining the goals in order to check for the website’s owner the web user interaction.
Of course, you’ll need a Google account and some experience with the tool. Let’s move to the experiment configuration.
First step is to access the Reports page and then the Behaviors > Experiments section. In this section we’ll be able to configure as many experiments as we want. We’ll just need to register by clicking the Create Experiment button.
Step 1: Goal
Besides from naming it, the first step of the experiment configuration is to know what we’re measuring. Google gives some default options:
- Duration of the session
- Visits per page
- Bounce rate
In case these options aren’t enough, it’s possible to create new metrics by clicking Create a new goal. There we’ll be able to specify many different interactions between the users and the website: a form to be filled, video views, pages saved as favorites, orders made…
It’s very important as well to define the percentage of visits that will be analysed (usually it’s a 100%) and the duration of the experiment.
Step 2: Variants
In this step we’ll just need to specify the urls of the different website variations (obviously, each one must have its own address). Many variants can be specified, but mind that the more different versions we offer to the user, the more hesitant he/she will be to choose the final version.
Step 3: Tracking code
At this point, the configuration has ended. Google collects the information we’ve sent and provides us a script that we’ll need to embed in the website. To do so, we just copy the code and paste it within the HTML code of the original website and all its variations, right after the <head> tag.
Step 4: Revision
In the last field, we can check if the tracking code has been correctly embedded in each of the web pages. If everything is correct, then we can start the experiment. In case the code is not correctly embedded, we’ll need to check the HTML code for the different URLs specified in step 2.
Step 5: Results
Once the experiment configuration is finished, Google will provide us the information we need after the first 48 hours.
This information will be available for each of the Analytics experiments configured, showing in a graphic the goals specified for each of the web variants provided (always within a range of time), as well as a summary table.
Below each graphic we’ll see a table with the results ranked, so that we can see which the best choice is.
To sum up… It’s pretty simple to use Google Analytics to measure different criteria and choose the most suitable option for a real experiment with users.
If this tool is for free, what would be the benefits of the payment ones? We’ll see soon…