How to Create
an A/B Test

Using Google Analytics
An A/B test is a way of analysing a web page's performance. Essentially, you create page A and copy it, creating page B. In page B you modify an element such as a headline or a button (it must contain at least one element that is modified from the original). Then you divide traffic equally between two webpage and test which page achieves the goal (order, purchase, subscription) faster.
Before you can start testing, you need to have a Google Analytics account. How to connect Google Analytics.

Go to your account on Google Analytics > Behavior > Experiments. Click on Create experiment.
Fill in the following fields:
— Name of an experiment;
— Choose a target (How to submit the goal completion rates to Google Analytics).
— Specify the percentage of your visitors to include in your experiment. To speed up the results, increase the percentage. However, if the variants are too wide-ranging, you ran the risk of getting negative results, so keep the number of experiment participants low.
— Turn on/off email notifications.
Enter the URLs of the web pages you'd like to test. Click Next.
Click Insert code manually and copy the code.
Go to Tilda, go to Site Settings → Other → HTML code for Head zone → Edit code. In the window that opens, paste the code, save changes and publish the site.

You can also add the experiment's code for specific pages only. To add a code in the Head for a specific page, go to Page settings → Additional → Insert code to Head.
Return to your Google Analytics account, click Next. The experiment code will be turned on. After the code is found, click Start experiment. From now on, all traffic to the page will be redirected, as a percentage, to the original page or other variants of the page.

You will get the first results after a few days. To track the results, select the experiment you'd like to track from the list and go to the reports page: Google Analytics> Behavior> Experiments.
For reliable data, tests should run for at least 7 days, since user behaviour differs from day to day.

Additionally, for a successful A / B test, you should make sure that the audience is homogeneous and that the indicators are stable. To do this, conduct an A/A test (this is when the site visitors are shown the same page but the traffic is distributed; this tracks the consistency in user behaviour).
Made on