Develop a hypothesis

[Reddi2]

At the hypothesis development stage, you need to determine what exactly will change (on the website page, in the application, in the advertisement) and what you are going to test. Let's look at an example of how to put forward a hypothesis when preparing an A/B test of an advertisement.

To conduct an advertising campaign, you need to develop a new design for the call-to-action button. There are several parameters that you can test: the phrase itself, the font, the color of the button. Let's say you first decided to change the color of the button from green (option A) to red (option B). In this case, you can put forward three hypotheses:

changing the color of the button will increase the number of clicks;
changing the color of the button will reduce the number of clicks;
The color of the button will not significantly affect the clickability of the ad.
To make the experiment as accurate as possible, it is better to test each what are the benefits of botim? parameter separately from the others. If you prepare two very different versions (with different calls to action, fonts, visual design), it will be difficult to understand what influenced the users' choice, and it will be difficult to interpret the test results.

Read also: Calls to action (CTA) - examples of call-to-action in advertising
Prepare an experiment
The accuracy of your experiment depends largely on how carefully you prepare it, so pay attention to this step. To create an A/B test, follow these steps:

Prepare two versions of the product or ad you will test—version A and version B.
Select your audience for testing by user type, geography, platform and split it into two equal groups.
Determine the minimum sample size using a special calculator . Please note: a sufficiently large sample size is required to obtain statistically significant results.
Set the testing period. To calculate it, divide the total sample size by your daily traffic. The result will be the number of days needed to run the test.