One interesting way to look at CDF on Materialized Views is the difference between source-level change and business-level change.

CDC might tell us that five banking transactions changed. But after those transactions pass through our transformations, what a downstream consumer may actually care about is:

Risk Score: 42 → 67
Monthly Spend: $8,500 → $11,200

This raises an interesting architectural question: can our data products tell consumers not only their current state, but what changed since the last processing cycle?

I explored this using a banking Customer 360 example, including potential patterns for downstream processing and reconciliation, as well as why CDF should not be treated as a replacement for persistent audit history.

Full article: From Materialized Views to Change-Aware Data Products.