We have defined our adaptive model with “Clicked” as positive and “Rejected” as negative response. Our App does not capture impression yet. How does the model calculate the success rate which is present in PDC. I feel like it is calculated as (Count of Clicked/(Count of Clicked+Count of Rejected))* 100. Does the model really consider number of requests sent through Container API?
You’re absolutely right in your assumption. When impression data is not available, Pega calculates the success rate using the following formula:
Success Rate=(Count of ClickedCount of Clicked+Count of Rejected)
This is based purely on recorded responses, not on the total number of recommendations made.
Also, No, the adaptive model does not directly consider the number of requests sent via the Container API unless those requests result in a response event (e.g., Clicked or Rejected).
The model learns only from explicit feedback—i.e., when a response is captured and mapped to the model.
If impressions (i.e., recommendations shown but not interacted with) are not tracked, the model has no visibility into how many times an action was presented but ignored.
@RaviChandra Thank you very much for your detailed response. I do have a follow up question. Once impression is built in the app and we configure Impression as a negative response in the models, will Click through rate formula would be as below?
(Count of Clicked/(Count of Clicked+Count of Rejected + Count of Impression))* 100
The Click-Through Rate of an Adaptive model is always tied to the Outcomes configured in an Adaptive model. So if Impression is captured and configured as a negative response in your adaptive model, the Click-Through Rate (CTR) formula would indeed change to reflect the broader set of outcomes.
Updated CTR Formula:
CTR=((Count of Clicked) / (Count of Clicked +Count of Rejected + Count of Impression))×100
Clicked → Positive response
Rejected → Explicit negative response
Impression → Passive negative response (i.e., shown but ignored)