Managing duplicates and merging records
How Dalil detects duplicate contacts, how confidence levels and match reasons work, and how to merge or dismiss duplicate groups safely.
Duplicate records fragment your CRM: two versions of the same person split the conversation history, confuse scoring, and can get a prospect double-messaged by a sequence. Dalil detects duplicates automatically and gives you a dedicated review workspace to merge or dismiss them.
You'll find it as Duplicates in the sidebar (with a badge showing pending groups), which opens the Manage Duplicates page.
How detection works
Dalil scans your People database every night at midnight UTC, and you can trigger an immediate Re-scan now from the Duplicates page after big imports.
Matching runs a series of rules, from strongest to weakest, and every duplicate group shows its match reason and a confidence level:
| Confidence | Match reasons |
|---|---|
| High | Exact same email address · exact same phone number · exact same LinkedIn profile |
| Medium | Very similar names (fuzzy matching) · similar name at the same company · near-identical emails on the same domain (typo-distance matching; different domains are never matched) |
| Low | Weaker signals, only when no stronger rule fires |
Records that match are clustered into duplicate groups; one group can contain more than two records.
Reviewing duplicate groups
The Manage Duplicates page lists pending groups with their contacts, confidence, match reason, and when each record was last seen. For each group you can:
- Review the records side by side, with one record marked as the primary (the one that survives the merge)
- Merge: combine the group into the primary record
- Dismiss: mark the group as not-a-duplicate; it stops appearing
To move fast on a clean backlog:
- Filter with Show only no-conflict entries to see groups where merging loses nothing
- Select multiple groups and use Bulk merge / Dismiss to process them together: the preview shows exactly what will be merged
What merging does
Merging combines the duplicate records into the primary record: conversations, activities, and relations follow the surviving record, so history is preserved in one place.
Merges are permanent, but restorable for 30 days. The Audit log on the Duplicates page records every merge, and you can Restore an individual merge within 30 days. The "Merged - 30d" view shows recent merges.
When records are merged, a Contact merge notification is sent so the team knows the data changed.
Why this matters
Clean, deduplicated data isn't cosmetic; it keeps the rest of the system honest:
- AI summaries and scoring read one complete history instead of two partial ones
- Sequences don't enroll the same human twice under different records
- Analytics count each contact once
Review the Duplicates page after every significant import, and let the nightly scan handle day-to-day hygiene.
Was this article helpful?
Your feedback helps us improve our documentation.