Hey @Gecofer 

I’ve seen similar cases before where the SQL logic was perfectly valid, the data unchanged, and the only difference was who was running the query. Sometimes, discrepancies like this can be related to runtime context—specifically:

• The identity used to run the dbt model (e.g., a service principal or job user) might not have exactly the same permissions as the one used in the SQL Editor.

• If your workspace uses Unity Catalog, there could be data masking policies or fine-grained access controls in place that behave differently depending on the caller.

• It’s also worth checking if the field in question is subject to any dynamic filtering or masking, especially in sensitive datasets, where visibility might be altered by design.

If the results differ, the root cause could lie in policy-based behavior rather than the query logic itself.


Hope this helps, 🙂

Isi