What is the difference between networks forwarding individual SKAN postbacks to Branch, rather than providing data via an API? Is one option better?

There are two main ways ad networks are sharing SKAdNetwork data with MMPs:

1. Forwarding the raw SKAN postbacks. This is the most common flow, and involves the network resending the individual conversion reports that iOS devices send to them. When ad networks forward these postbacks directly, MMPs are able to process and cryptographically validate the individual reports (as recommended by Apple), filling their traditional role of independently verifying data and ensuring trust by all parties.
2. Providing pre-aggregated data via an API. This approach is being used mostly by the largest walled garden networks (often referred to as SANs, or Self Attributing Networks). In this flow, the network collects individual SKAN postbacks, cryptographically validates them internally, and then provides only pre-generated results to MMPs via an API endpoint. This means that MMPs won't be able to 'check the homework' of networks using option 2. In addition, not every network sharing data via API is providing all the data dimensions reflected in the SKAdNetwork postbacks.

To see which method is being used by a given network, please refer to this list.

Note: While individual SKAdNetwork postbacks are not technically 'aggregate data' (because each one corresponds to a single install — for more about the difference, please see What's the difference between device-level and aggregate data?), the design of the SKAN system ensures that the data inside each postback is already at an aggregate level of detail.