Control group testing using impact Analyzer

[victor (Victor Roy)]

Hi

We are doing a PoC for our customer to implement control group testing. In the first phase we just wanted to do a control group testing to establish How is Next-Best-Action performing against a random relevant action. We are in 8.8 and leveraging impact Analyzer for this.

As we want to do only this test and not the other tests, we are disabling the other Impact Analyzer by making them “Inactive”.

My question is making inactive in impact analyzer will ensure that there is no control group entry(for any customer) for other experiments? or do we need to go to each experiment type and make the test group as 100%.

I checked the experiment type DDR and could not find any check box for Active/inactive.

Please suggest

[Otto_Perdeck]

Hi @victor , you can turn off Impact Analyzer experiments from the UI.

As long as you keep at least one of the experiments on with a non-zero percentage, that will continue to be used to give the ADM models some “exploration”. You do not need to change the percentages for disabled experiments.

Regards

-Otto

[Baibhav (Baibhav Raj)]

Hi @Otto_Perdeck I’m trying to do some POC on impact analyzer in my dev environment infinity 23. But on UI it is showing not enough data. So basically, is the data source is interaction history? We have IH record for past 20 days. Do we need to enable from somewhere to get data?

Please suggest!