← All resources+PRACTICAL GUIDE / 002

Different accounts. The same underlying actor?

A practical introduction to entity resolution, evidence quality, and the limits of a match.

Why identity fragments

The same person or business can appear through different account identifiers, devices, emails, and payment methods. Systems also disagree about formatting and timestamps. Entity resolution is the process of connecting records that may refer to the same underlying entity while preserving the uncertainty in that connection.

Distinguish exact and inferred relationships

A stable customer identifier can support an exact association. A similar email pattern or overlapping address is weaker evidence. Combining these without distinguishing their strength can make a graph look more certain than the source data allows. Record the type, origin, timestamp, and confidence of each relationship.

Test legitimate overlap

Shared office addresses, family devices, corporate payment instruments, and recycled contact details can create false matches. Evaluate these cases explicitly. A graph should allow analysts to inspect and challenge a relationship, rather than silently collapsing all related records into one asserted identity.

Evaluate with reviewed examples

Build a representative evaluation set that includes confirmed matches, confirmed non-matches, and unresolved cases. Measure incorrect merges as well as missed connections. Revisit the evaluation when sources change or new fraud patterns emerge.

Keep correction possible

Investigation outcomes should be able to correct a relationship and preserve an audit trail. Entity resolution is an ongoing process, not a one-time cleanup. The aim is a useful, inspectable account of the available evidence.

Examples are illustrative. This guide is educational and does not provide legal, regulatory, or financial advice.
← Explore all resources
+MAKE THE CONNECTION

Find the connections
your fraud stack is missing.

Bring your risk challenge. Let’s map a path forward.

Book a meeting