Insight model score feed layout overview

Overview

This document provides additional detail related to the insight model score feed layout, including the use of the feed layout, how the data is reflected in the database, and how to use the data from RPI.

This section provides the following detailed information:

Feed layout intent

The intent of this feed is to allow for business unit/entity ID level insight information to be loaded to the Redpoint CDP. This feed is used to ingest model scores/segments data for both externally and internally generated data sets related to individuals, households, emails, and party profiles. This data could be used in selection criteria based on model score or segment for both targeting and personalization.

This feed does not have an extension feed like many other feeds in Redpoint CDP.

Feed layout areas of consideration

This feed enables the ingestion of model score/data segmentation information at a variety of entity levels. The identity ID field is a mandatory attribute that must be included within this feed. The CDP confirms that this ID exist in the CDP, and if it does exist, then the data will get loaded to the insight data model within the CDP.

The identity ID that is expected needs to originate from the CDP, with the exception of the Party Profile ID. The party profile ID is actually the source party profile ID, so it can be constructed outside the CDP if you know the appropriate value to use for a given record.

Party profile ID has a unique processing behavior that is different than the processing of records with the other IDs:

• The other IDs are confirmed to exist in the CDP, and then they are loaded to the appropriate table within the insights data model, specifically the individual, household, and email score tables.

• When the party profile ID is provided:

  1. First, that value is verified to exist as a source_party_profile_id in the party profile table.

  2. Then individual_id and individual_business_unit_id are selected and used to load the score information to the individual model score table.

  3. Then there are two additional checks that occur:

     • Is household_id populated? If so, then the scores will be loaded to the household model score table using that ID and the scoring information from the feed layout.

     • Is personal_email_id populated? If so, then the scores will be loaded to the email model score table using that ID and the scoring information from the feed layout.

There are additional email ID fields in the party profile table besides personal_email_id, but personal_email_id is the only email ID field that is checked and a score loaded for when party profile ID is provided in the feed layout. You can load emails related to the other IDs by providing that ID in the feed layout and not party profile ID.

These are the two fields used to look up the entity within the Redpoint CDP and load the score/segment information to the correct table within the insights data model.

• Identity ID: These IDs are sourced from the Redpoint CDP, with the exception of party profile ID, which is really the source party profile ID. This could be one of the following IDs:

  • individual_id

  • household_id

  • email_id

  • source_party_profile_id

• Identity Type Code: This id the name of the ID that corresponds to value of the ID being provided in the identity ID field:

  • individual_id

  • household_id

  • email_id

  • party_profile_id

Assuming that the ID exists in the CDP, then the data will be loaded to the appropriate model score/segment tables. Given the unique behavior of party profile, a record that is confirmed to exist in the CDP will generate 1-3 records in each of the respective score tables.

Identity matching

The identity feed does not go through the matching process.

Use in RPI

Insight data can be used in RPI to either select or suppress records from an interaction using model score, decile, or segmentation information. This information can also be used to support personalization of content.

Use in aggregates

Currently, model scores data is not used in the creation of any aggregates.