Option to re-include previously unmatched records in matching rules
C
Chelsea Strong
started a topic
3 days ago
We've hit a recurring limitation with how unmatched records are handled. When an admin first dedupes a dataset, they manually unmatch records they don't believe are duplicates. But once a record is unmatched, DataGroomr treats that decision as permanent and global. It never surfaces that pair again via any matching rule, in any dataset, even when matching on completely different values.
The problem is that "not a duplicate" isn't always a final answer. Later, when an end-user works the list, they sometimes find that a previously unmatched record *is* in fact a duplicate. We have a process where end-users can flag duplicates for merge, and they will automatically get merged by DataGroomr. But given the "unmatched" limitation, there's no way to automatically have DataGroomr merge those records.
**Suggested solution:** Add a dataset-level toggle to include previously unmatched records in matching. Ideally it could be scoped to only re-include records unmatched from a *different* dataset than the one being processed, so admins can re-evaluate cross-dataset matches without reopening every decision made within the current dataset.
Chelsea Strong
We've hit a recurring limitation with how unmatched records are handled. When an admin first dedupes a dataset, they manually unmatch records they don't believe are duplicates. But once a record is unmatched, DataGroomr treats that decision as permanent and global. It never surfaces that pair again via any matching rule, in any dataset, even when matching on completely different values.
The problem is that "not a duplicate" isn't always a final answer. Later, when an end-user works the list, they sometimes find that a previously unmatched record *is* in fact a duplicate. We have a process where end-users can flag duplicates for merge, and they will automatically get merged by DataGroomr. But given the "unmatched" limitation, there's no way to automatically have DataGroomr merge those records.
**Suggested solution:** Add a dataset-level toggle to include previously unmatched records in matching. Ideally it could be scoped to only re-include records unmatched from a *different* dataset than the one being processed, so admins can re-evaluate cross-dataset matches without reopening every decision made within the current dataset.