Hi All,
I m working on merge policy and I would like to understand when we need to use ID Stitching None OR Private Graph and when to use merge method Dataset precedence OR Timestamp ordered, mostly how to combine them. Assuming that I have this use case:
1) I want to create 5 segments that qualifies profile from all of my dataset that are enabled for profile. These segments will be used for advertsing destinations
2) I want to create 1 segment that qualifies profile from DS1, DS2,DS3 that are enabled for profile. This segment will be used for email marketing/survery destination.
Any suggestion?
Thanks
Solved! Go to Solution.
Views
Replies
Total Likes
I would look at the data, its quality, and which source is sending it to decide the right merge policy to pick. The intention should be to get the right/best data to build the 360-view of profile for the activation. I would not look for activation/segmentation to decide which merge policy to pick.
If data is coming from 3rd party then the preference will be to pick 1st party data and in this case, dataset precedence is the best choice. For other cases, I would go with time preference.
Regards,
Chetanya
Hi @Luca_Lattarini,
1] ID Stitching None : Multiple profile fragments are not stitched together to form single view of the customer though profile fragments may belong to same customer. This could result having multiple audience for same profile during segmentation.
2] ID Stitching Private Graph : Multiple profile fragments are stitched together to form single view of the customer. This could result only one audience for same profile during segmentation.
In following document, these concepts are explained in more detail.
I would look at the data, its quality, and which source is sending it to decide the right merge policy to pick. The intention should be to get the right/best data to build the 360-view of profile for the activation. I would not look for activation/segmentation to decide which merge policy to pick.
If data is coming from 3rd party then the preference will be to pick 1st party data and in this case, dataset precedence is the best choice. For other cases, I would go with time preference.
Regards,
Chetanya
@ChetanyaJain @vikash4 Thanks guy for your suggestion.
Assuming that i want to create a segment by using qualities profile data from ALL datasets that have been enable for a profile.
If I use ID Stitching None method and Data precedence with ALL my datasets ordered by quality data (First 1st party data and last 3rd Party data).
Am I able to have a good qualities of segment and use ALL datasets OR it is more accurate Stitching Private Grapgh method and Data precedence?
Views
Replies
Total Likes
@Luca_Lattarini This is tricky and will depend from case to case and is heavily dependent on the data that is flowing in. Always do data profiling before you ingest and gain better confidence.
Views
Replies
Total Likes
So @Luca_Lattarini my POV is u should use id stitching : none, given its for Ad targeting. ie. topline growth, and not nearly an email campaign.
Views
Replies
Total Likes
Views
Likes
Replies
Views
Likes
Replies