AB Testing: "Auto-Target" VS. "Auto-Allocate"

[niocolasmeriel]

Hi there 

I would like to A/B Test the "Auto-Allocate" activity Type VS. the "Auto-Target" activity type for RPV.

Has anyone experience with this i.e. how this could be achieved?

It's for retail and we are wondering what would work better in terms of maximizing RPV?

(or should we assume Auto-Target to work better because it does not only optimize but also personalize and can be more aggressive due to a 90/10 split for example vs. a fixed 80/20 with Auto-Allocate?).

Client is not really intrested in the activity type nor optimization strategy but the business objective which is to maximise RPV.

Thanks for your feedback and guidance.

Nicolas

[alexbishop]

Could you split your traffic into two groups with a profile script and then assign Group A to the Auto-Allocate activity & Group B to the Auto-Target activity?

[niocolasmeriel]

Yes, I guess this should work. Thanks @alexbishop 

I just wonder if someone has done this before or has experience with this and if it is correct to expect that Auto-Target should perform better in increasing RPV?

[cristinel · Accepted solution]

@niocolasmeriel  compared to Auto-Allocate, which aims to find a single winner, regardless of the visitor profile, Auto-Target determines the best experience for a given visitor. Depending on your specific test conditions, it is expected that Auto-Target performs better than Auto-Allocate as it will maximize revenue per visit taking into account the different interests of your visitors.