When you create an A/B test, the proportion of users allocated to the original version and the variation(s) is set automatically.
When you create an A/B test the percentage of users is split 50/50 by default between the original version of your app and variation 1. This split allows you to compare the same number of users and have a point of reference to calculate the percentage improvement and reliability rate.
It is possible to modify the traffic allocated to your variation. You may wish to do this if you only want to expose a small proportion of your users to your variation for example.
Modify the percentage of traffic
To manually adjust the percentage of traffic allocated to your variation, please do the following:
- Click on the symbol.
This will display the traffic allocation window
- Move the slider to the left to select the required proportion to reduce the percentage of users allocated to your variation.
If you only have one variation, you will only have one slider to modify the traffic allocated to your variation.
If you are testing several variations, you will have several sliders giving you the possibility to split the traffic differently for each variation.
- Move the slider all the way to the right if you want to allocate 100% of the traffic to your variation.
If you choose to allocate 100% of the traffic to your variations, you will not have any visitors to your original page.
It is not possible to modify the percentage of traffic for the original version (the slider is deactivated by default). It adjusts automatically to the variation. It never exceeds the percentage of the variation even if it is very high.
When you reduce the percentage of traffic to a variation, the system automatically applies an equal split between the original version and the variation. This results in some users who do not see variation 1 being excluded. This percentage of untracked users is shown in grey on the traffic allocation diagram.
The percentage of untracked visitors is excluded from the statistics and therefore does not appear in the reports.