Attributes can be used to create Segments using the Segment Rules. ShiftForward DMP provides two sources of attributes: event-driven and machine learning.
Event-driven
Attributes based on the user’s collected events. ShiftForward DMP comes with a standard set of Attributes but custom Attributes can be created as well.
Standard event-driven attributes:
Attribute name | Type | Rule description |
events.eventType.last | Time | Time of the most recent event of the type eventType. |
events.eventType.count | Number | Number of events of the type eventType performed. |
events.timestamp.first | Time | Time of the oldest event. |
events.timestamp.last | Time | Time of the most recent event. |
events.count | Number | Total number of events. |
mkt.lastcampaignId.name | String | Name of the most recently seen campaignId= in the event stream. |
mkt.lastsource.name | String | Name of the most recently seen source= in the event stream. |
ecom.cartLeaver | Boolean | True if the most recent addToCart event happened after the most recent Purchase event. |
ecom.purchase.value | Number | Sum of the transaction.price value from all user events. |
ecom.purchase.last.value | Number | The transaction.price value from the most recent Purchase event. |
Optionally enabled standard Attributes:
geo.city.last | String | City name in the most recent event. |
geo.country.last | String | Country name in the most recent event. |
geo.location.last | Lat/Long | Lat/Long in the most recent event. |
Example of custom event-driven attributes:
name | Type | Rule description |
retarget.campaign148374.last | Time | Time of the most recent event that contained campaignId =148374. |
retarget.geoNearShop.count | Number | Number of events where geo.location was less than 500m from a specific list of location. |
retarget.funnelStage1.last | Time | Time of the most recent event that contained custom key “funnelStage” with value “1”. |
Machine learning (ML)
Machine learning identifies patterns in captured data, enabling the creation of recommendations or behavior predictions for specific users. The application of machine learning greatly increases the usefulness of the captured data as it helps to better segment users.
ShiftForward DMP leverages machine learning to identify high value users from their historic behavior. Given the required information sources, ShiftForward DMP provides the following additional details for each user:
name | Type | Rule description |
ml.clv | Number | The expected spending for the user for the next 12 months. |
Create your own using your custom ML models, see more on the machine learning in ShiftForward DMP.