Reco Logics: Correcting reco algorithms by returned items
My client wants to correct the tracked conversions that impact algorithms (POP, AFF, DL, etc). by the items that were returned back to the store. It seems that other providers allow them to do this.
Is there a way
- for them to feed product return information back to DY and
- for us to in hindsight correct the conversion data, so that it does not distort recommendation logics or audience data?
CSM in Need
short answer: no
to do this you'd have to have 2 data sets:
- An export of transaction_ids (same sent to DY) which were returned, and the revenue amount that was returned
- Raw data containing the transactions_id attributed to any report you'd like to clean
Number 1 is easy in the sense that the data exists, but only the customer has it.
Number 2 is harder because:
- there are multiple things to correct (experiment results, recommendation attribution) hence multiple raw data sets we'd have to have available.
- Currently none of the above available to us contains a transaction ID
The additional issue is that even if we had all the ingredients above, we could produce at best tableau reports which are heavy in processing, and would not be exposed to customers.
I think this is a good idea, but it is a feature request.
Please sign in to leave a comment.