Customer data glossary
Deterministic matching
By Art Kukharenko · Published · Updated
Browse learn glossary resourcesDeterministic matching links records only when they share an exact, verified identifier, such as a customer ID, a login or a normalised email address.
Part of the Learn hub and the customer data glossary.
What it means in practice
Deterministic matching is rule-based: if two records carry the same value in a trusted field, they are treated as one person. Before comparing, values are normalised (lower-casing an email address, stripping phone formatting) so trivial differences do not prevent a match.
Its strength is precision and explainability. Every merge traces to a rule and a shared value that an auditor or a customer can follow. Its weakness is recall: if someone used a work email once and a personal email another time, with no shared identifier, the records stay apart.
Identifiers are not equally strong. An authenticated login or customer number is a strong key. A verified email is good. A phone number can be recycled or shared. A cookie or device ID identifies a browser, not a person. Teams therefore rank identifiers into a match hierarchy and add guard rules, such as refusing a merge that would join an implausibly large cluster, which usually signals a shared family mailbox or a shared device.
Hashing does not change the method: records can be matched on the hash of a normalised email. Hashed identifiers are still pseudonymised personal data: GDPR recital 26 says data that could be attributed to a person with additional information remains information on an identifiable person. Lawful-basis and deletion duties still apply.
Example (retail banking): a bank attaches web-analytics sessions to customer records only when the visitor is logged in to online banking, using the customer number carried in the session. The link can be explained line by line, and anonymous browsing is never attached to an identity.
Related terms
- Identity resolution: the wider process this method serves.
- Probabilistic matching: the score-based alternative.
- Identity graph: where deterministic links are stored.
- Golden record: the profile built from matched records.
- Single customer view: the outcome that matching supports.
Where it matters
- Identity resolution explained: how to combine deterministic rules with scoring.
- Customer 360 data model for retail banking: which keys banks can rely on.
Frequently asked questions
When is deterministic matching enough?
When most customers authenticate and you hold a trusted key such as a customer number. It suits banks, payers and other account-based businesses. It struggles with anonymous traffic and with customers who use several emails.
Why can a shared email cause false merges?
A family mailbox or shared work address can sit on several people's records, so an exact-match rule would merge them into one profile. Guard rules, such as capping cluster size or excluding known shared addresses, prevent this.
Is deterministic matching more private than probabilistic matching?
Not inherently. Both process personal data. Deterministic matching is easier to explain and often uses fewer attributes, which helps data minimisation, but the legal duties are the same.
Back to the customer data glossary or the Learn hub.
Further reading
- Customer Data Glossary: 50 Terms
- Customer 360 Banking Data Model
- Learn about CDP
- Golden record — definition
- Identity graph — definition
Published: · Updated:
Publication date recorded for the approved release. Editorial updates do not establish legal approval or a customer result.