Your experiment report is your tool to understand which variation was the most effective.
A variation is assigned by the server as soon as you run the initialize function. The system will add one view or one conversion to the report for each event. One user can record more than one view or conversions.
If the Arise client could not reach the server, the client will fallback on the default variation value and no events will be recorded (views or variations).
Variation A | Variation B | |
views | 1290 | 1300 |
conversions | 130 | 340 |
conversion rate | 10% | 26% |
This table means that:
The conversion rate is the total number of conversion events divided by the total number of view events.
Keep in mind that:
In this example, variation B was better than variation A.
Variation A | Variation B | |
views | 2340 | 2251 |
conversions | 2100 | 3450 |
conversion rate | 89% | 153% |
Seems like a bug in our system? No, it’s a feature! This experiment could have been implemented in a very addictive game.
A screen is shown to the player when he has no more life (game over):
Variation A:
Variation B:
In this example, the results means that a player who loose purchases in average 0.89 life with the variation A and 1.53 lives with the variation B. The revenue is higher with the variation B than with the variation A!
This is why in some cases we can have more conversion events than views! The goal here was to optimize the number of lives purchased.