Earlier, we used CDTs for data handling, but we moved to Record Types because they provide better features, performance, and scalability.

Use Record Types When:

Building a Modern UI: Records allow you to create lists, filters, and record views with minimal coding.
Speed is Critical: Synced records provide faster query results because they pull data from memory rather than executing a full database call every time.
Handling Complex Reporting: Use Custom Record Fields to aggregate data (e.g., total sales per customer) directly on the record.
Working with Multiple Sources: If your data comes from both a database and an external API, records can unify them into a single data model.

Use CDTs When:

Unsynced Data Needs: If you are using an Unsynced Record Type, you still need a CDT and a Data Store Entity (DSE) to map the database table.
Legacy Requirements: Older process models or specific smart services (like "Export DSE to Excel") still require CDTs for their configuration.
Process Variables: CDTs remain useful for storing temporary, strongly-typed data structures within a single process instance.
External Integration: Some web service integrations or Java-based plugins require a CDT to define the incoming/outgoing data structure.

Appian now recommends using synced record types for most new development due to their increased flexibility, performance, and built-in functionality.