You've done a thorough job checking the obvious dependencies! The fact that those two columns are still checked and greyed out in the "Select tables and columns" for that data source is the key indicator that a dependency still exists at the system level, even if you can't see it directly in segments or measures.
Here's a systematic approach to uncover these hidden dependencies and then remove the data source:
1. Focus on the Greyed-Out Columns in "Select tables and columns":
These greyed-out columns are the most likely culprits. The system believes something still relies on them. Since you've checked segments and measures, the dependency is likely within the data unification process itself, even after you've seemingly removed it from the steps.
2. Investigate the Mandatory Matching Rule:
You mentioned a mandatory matching rule that allowed you to save unification. This is a prime suspect. Even if it appears to only use the unified Customer entity, it might have been initially configured (or the system might still internally track) using columns from your obsolete data source during its creation.
- Edit the Mandatory Matching Rule: Go back to your Matching rules and open this specific rule. Carefully review all the rule conditions and the fields being used. Even if it currently points to the unified entity, see if you can identify any lingering references or configurations that might have involved the columns from the obsolete source.
- Try Recreating the Mandatory Rule: As a test (be cautious, as this might impact your unified profile quality), try deleting this mandatory matching rule and then recreating it, ensuring you only use fields from the unified
Customer entity. If you can then proceed with removing the data source, this was likely the hidden dependency.
3. Examine Table Relationships within Unification Again (with a Different Lens):
You mentioned two relationships that seem like unification artifacts. Let's revisit these with the greyed-out columns in mind:
- Identify the Tables and Columns in the Relationships: Note down the exact tables and columns involved in these two relationships.
- Cross-Reference with Greyed-Out Columns: Do the greyed-out columns from your obsolete data source participate in either of these unification-related relationships? Even if it seems like a system-level relationship, the system might still consider those columns as being "used" due to their involvement in defining the relationship, even if the relationship's logic currently doesn't explicitly leverage them.
- Try Deleting These Relationships (Carefully): If these relationships involve the tables from your obsolete data source and you suspect they are the cause, carefully consider deleting them. Back up your Customer Insights - Data configuration before doing this! After deletion, try removing the data source.
4. Use the API to Inspect Dependencies (More Advanced):
If the UI isn't revealing the dependency, you could try using the Customer Insights - Data API to inspect the metadata and dependencies more directly. This requires some technical knowledge. You would look for references to your data source or the specific greyed-out columns in the API responses for entities, matching rules, and unification configurations.
Steps to Remove the Data Source:
Once you've identified and removed the dependency:
- Go to Data Sources.
- Hover over your obsolete data source.
- You should now be able to click the "..." (ellipsis) and select Remove.
In your specific case, the greyed-out columns in the "Select tables and columns" section of the data source are the strongest clue. Focus your investigation on any matching rules or unification-related table relationships that might still be referencing or be configured based on these columns.
Start by carefully re-examining the mandatory matching rule. Let me know what you find!
