There are different approaches to prioritization :

[tasmih1234]

The following sentence can be used as a help: “If X is changed, then X happens, for reason Y (because).
An example hypothesis could be:

If the contact form is placed above the fold, we increase the number of contacts because it is easier to find.
If we color the CTAs green instead of gray, they will be clicked more often because the color green is friendlier and therefore more visible.
Once the hypotheses have been established, they should be prioritized , because several tests cannot be carried out at the same time. If this is the case, the results often cannot be clearly attributed to a change.

How great is the benefit? – What results are expected? Do we assume that an increase in the conversion rate is likely? These questions are central here and must be answered for this solution
Which pages have a lot of traffic? – Pages with a lot of traffic can produce portugal mobile numbers list faster and clearer benefits. If there is little traffic in general, the tests should definitely be carried out one after the other!
What resources are required? – What should be tested? Depending on the change to be made, more or fewer resources may be required (graphical change vs. technical change)
In addition, a null hypothesis must always be formulated, which follows the idea that no significant results can be expected. These hypotheses are used as comparison hypotheses .
Once an approach has been chosen and hypotheses have been prioritized, a clear roadmap should be created that captures the defined goals & outcomes.
4. Conduct tests
Setting up and executing AB tests - Testing
The time has finally come: the tests can begin! The implementation depends heavily on the test tool selected . It is therefore not possible to give a clear statement at this point as to how the tests will be carried out. In some cases you will need technical knowledge to implement them, but with other tools you will not. You should find out more about this before choosing a tool.
Once these questions have been answered and the test is running, the specified time should be waited until statistically significant results are available. The time required should therefore be determined taking into account the daily and/or monthly traffic, the number of variants, the expected improvement and the existing conversion rate.