Probabilistic Match Review and FHID Consolidation
Admins can now review likely duplicate customer profiles, decide how they’re linked, and merge them into one FirstHive ID (FHID), with every decision audited and reversible.
The problem
FirstHive merges records automatically only when they share an exact identifier, such as the same email or mobile number. So the same customer often ends up as two or more profiles: a Gmail alias, a new phone number, a typo in the name.
FirstHive already scored these likely matches, but admins couldn’t see them, had no say over whether matches linked automatically, and had no controlled way to turn a confirmed match into one profile. The result: duplicate sends, wrong frequency caps and inflated segment counts.
Brands also want different things. A regulated bank wants a person to review every merge. A high-volume retailer wants confident matches handled automatically.
How it solves the problem
- One review console: Unification Admins and Data Stewards see each likely match side by side with its match score (0.00 to 1.00), then link it, mark it as a household, or reject it.
- You choose how automatic it is: turn auto-linking on or off and set the score at which matches link on their own.
- You choose what linking does:
- Merge automatically into one FHID
- Admin review before the merge happens
- Keep separate FHIDs grouped under one ID, so Customer 360 and segments see one person
- Households: mark different people at the same address as a household instead of merging them.
- Safe and reversible: every link, merge and setting change is audited and can be undone. Retired FHIDs still resolve, so APIs and exports don’t break.
- Gets smarter: reviewer decisions are saved and show where thresholds should sit.
Recommended starting point: auto-link at 0.90 with admin review before merge. Banking and insurance brands should start with auto-link off.
Use cases
| Who | Use case |
|---|---|
| Unification Admin | Set auto-link at 0.90 for a retail brand, and keep every merge manual for a regulated bank |
| Data Steward | Work through uncertain matches quickly with both records side by side |
| Marketing ops | Get one profile per real customer, so frequency caps, suppression and segment counts are right |
| Compliance or DPO | Show who merged which profiles, why, and how consent was combined |