Question
Tata Consultancy Services
GB
Last activity: 8 Feb 2021 17:50 EST
How to input to campaign run first 250k customer from a sorted list on model score
Hi All, We have a requirement to input to campaign run first 250k customer from a sorted list on model score. Is this achievable in Pega Marketing with out compromising performance. What is the best approach to this requirement. Shall we have a processing outside of Pega. may be in the db side to find out first 250k customer and input those to the campaign run. Please advise. Many thank.
Updated: 8 Feb 2021 17:50 EST
Pegasystems Inc.
US
Did you consider using the OOTB capabilities with Next Best Action Designer (NBAD) to get your modeled results instead of a segment driven approach you are trying here ? If not NBAD is highly recommended, please try.
Assuming you have other constraints for this decision, here is a possible thought: Functionally and technically you could add additional "associated" data to your dictionary. For your case see if this would help:
Let's assume you are able to create a VIEW in the underlying schema that does a ORDER BY on your model score. You should also probably add a sequencing column for your requirement of (250K) Let's call that class/view : ModeledData
1. Define association to the customer dictionary to create a association with Customer to : ModeledData - Assuming 1 to 1 to Customer, but 1 to many could also work.
2. Now you can construct a segment/audience rule that queries against the ModeledData to join with customer data.
(the sequencing column would limit to 250K volume and the VIEW's ORDER BY would give you the SORT you need)
3. This segment rule could be the source of your outbound run.
You probably need some orchestration of when the underlying table for ModeledData is refreshed and when the segments are re-run. Segments could also be refresh at the beginning of an outbound run.
Did you consider using the OOTB capabilities with Next Best Action Designer (NBAD) to get your modeled results instead of a segment driven approach you are trying here ? If not NBAD is highly recommended, please try.
Assuming you have other constraints for this decision, here is a possible thought: Functionally and technically you could add additional "associated" data to your dictionary. For your case see if this would help:
Let's assume you are able to create a VIEW in the underlying schema that does a ORDER BY on your model score. You should also probably add a sequencing column for your requirement of (250K) Let's call that class/view : ModeledData
1. Define association to the customer dictionary to create a association with Customer to : ModeledData - Assuming 1 to 1 to Customer, but 1 to many could also work.
2. Now you can construct a segment/audience rule that queries against the ModeledData to join with customer data.
(the sequencing column would limit to 250K volume and the VIEW's ORDER BY would give you the SORT you need)
3. This segment rule could be the source of your outbound run.
You probably need some orchestration of when the underlying table for ModeledData is refreshed and when the segments are re-run. Segments could also be refresh at the beginning of an outbound run.
I also assume once the 250K ordered items are obtained in your segment, you don't mind in which order the strategy is executed and the actions are sent out to those customers.