Field based vs Graph based Identity sitching difference

The core difference between the two functionalities is that field based stitching only considers 2 types of identities for resolution while graph based stitching can use more than 2.

In practice, this means that in order for stitching to take place using field based stitching, there must be a row of data in which both the persistent ID (i.e. the ECID in most cases) and the transient ID (the logged in identifier) are present for stitching to connect the two IDs together. 

For graph based stitching, the persistent ID may be connected to another identity which is not the transient ID, but the identity graph can make the connection between the different logged in identifiers and return the transient ID you are looking for. Take the example of the persistent ID being the ECID, the transient ID being a logged in ID, and a third identity being a hashed email. If there is a row of data with both the ECID and the hashed email, field based stitching would not stitch this row since the transient ID is not present, but graph based stitching would stitch this row together if the hashed email and logged in ID are connected in the identity graph.

Here are a few slides I can share as a reference which may help explain visually as well:

Keep in mind that this is a simplified scenario only used as an example.

And here is a reference article that gets very in the weeds but is very helpful too:

Let me know if this cleared things up or if you have other questions! :)