Deduplication
Deduplication is the ongoing enforcement that one real person or company is represented by exactly one record, across every object that references them.
Duplicates are not a tidiness problem. They split engagement history so scoring under-counts, they route the same buyer to two owners, they inflate funnel counts, and they make account-level reporting wrong in a direction that varies by segment, which is worse than being wrong by a constant, because it cannot be corrected for.
The hard part is matching. Email is reliable and often absent; company name is present and unreliable; domain is good until subsidiaries and shared free-mail domains appear. A defensible dedupe strategy states the match rules, the merge precedence for conflicting fields, and what happens to the history of the losing record.
Where it breaks
Merge rules keep the newest value on every field, so a merge silently discards the enriched firmographics on the older record.
Related terms
- Revenue data model
- The revenue data model is the set of objects, relationships and required fields that every revenue-facing process reads from and writes to.
- Routing
- Routing is the deterministic assignment of a record to an owner based on registered rules, segment, territory, coverage, and existing account relationship.
- Data quality
- Data quality is the measured fitness of records for the decisions made from them, completeness, accuracy, consistency, timeliness and uniqueness, each measured against a stated requirement.
- Account-based marketing (ABM)
- ABM is a demand model in which the account, not the contact, is the unit of targeting, measurement and qualification.
Field notes on this
- The MQL debate needs an acceptance record
A recorded acceptance decision makes lead-quality disputes inspectable. It is a signal, not a complete diagnosis of demand performance.