Entity resolution
Entity resolution is the process of determining which records across different sources refer to the same thing.
Category
Industry concept
See all terms
What is entity resolution?
Entity resolution is the process of matching records that refer to the same person, company, product, or other real-world entity, even when each record names or describes it differently. A CRM might list "Acme Corp," a support ticket might say "Acme Corporation," and an invoicing tool might use a tax ID without a name. Entity resolution decides that all three point to the same customer and merges what each source knows into one record.
The problem it solves is fragmentation across systems that were not designed to agree with each other. Without entity resolution, a company ends up with several partial, disagreeing versions of the same customer, employee, or product, and nobody can tell which one is current.
How does entity resolution differ from deduplication?
Deduplication removes exact or near-exact repeats within a single dataset. Entity resolution works across datasets with different fields, formats, and identifiers, and it has to infer a match from indirect evidence: a shared domain, a similar name, an overlapping address, a matching phone number. Deduplication asks whether two rows are copies of each other. Entity resolution asks whether two records, which may look nothing alike, describe the same thing.
How does entity resolution work?
Entity resolution starts by narrowing the search: comparing every record to every other record doesn't scale, so systems group records into smaller candidate sets first, by shared domain, similar name, or matching identifier. Within each group, a matching step scores how likely two records describe the same entity, using rules, weighted attribute comparisons, or a trained model. Matches that clear a threshold get merged into one canonical record, with conflicting values reconciled and the more reliable or current one kept.
A context repository applies this same process to the customers, people, and products it pulls in from every connected source. It resolves each entity once, keeps a single current record for it, and updates that record as sources change.