Synthetic test users make end-to-end QA dramatically easier.

Ad hoc test data makes QA harder than it needs to be.

When every release creates new test users on the fly, teams spend time reconstructing states instead of testing behaviour. One person tests an opted-in customer, another uses a profile with incomplete identity, and a third unknowingly reuses a record whose history already changed the result.

I prefer a small, named test set.

The exact profiles depend on the program, but the structure usually covers the important branches:

The profiles need operating rules.

• Give each profile a stable purpose and owner.
• Document the expected state before each scenario.
• Reset only what must be reset, and record when the state changes.
• Keep identifiers obvious enough that nobody mistakes them for real customers.
• Use the same profiles across source systems when end-to-end lineage matters.
• Pair each profile with an expected outcome so screenshots and logs have context.

The value is shared evidence.

Synthetic users make debugging faster because everyone is talking about the same record. Marketing can point to the expected message, engineering can inspect the event and payload, data teams can trace the source record, and QA can compare actual behaviour with a pre-agreed scenario.

They are especially useful for negative tests. It is easy to prove that a message sent. It is more important to prove that the customer who should not receive it did not.