Entity resolution is the process of figuring out when different records — spelled differently, entered separately, or deliberately disguised — actually refer to the same real-world person, company, account, location, or asset. It takes "Roberto Gómez," "R. Gomez," and "Robert Gomez S.A. de C.V." and decides, with evidence, which of those are one entity and which are not — then links everything that entity touches into a single view.
If you've ever pulled a customer, vendor, or suspect and found five near-duplicate versions across five systems, that's the problem entity resolution solves. It's the difference between counting records and knowing who and what you're actually dealing with.
Why does entity resolution matter?
Most organizations don't have a data shortage — they have a data fragmentation problem. The same person exists in your CRM, your onboarding system, a sanctions list, a contract PDF, and a spreadsheet an analyst kept on their desktop. Each copy is slightly different. Until those copies are resolved to one entity, you can't answer basic questions:
- Is this new customer the same individual we rejected last year under a different spelling?
- Do three "unrelated" vendors actually share a director, an address, and a bank account?
- Is the beneficial owner of this account on a watchlist under an alias?
- How much total exposure do we really have to one counterparty spread across six subsidiaries?
Without entity resolution, the answer to all of these is a slow, manual investigation — or a miss.
How does entity resolution work?
At a high level, entity resolution runs in stages:
- Ingestion — pull records from every source: databases, warehouses, APIs, case files, contracts, registries, structured and unstructured alike.
- Standardization — normalize formats (names, dates, addresses, tax IDs) so they can be compared fairly.
- Matching — compare records using multiple signals (exact keys, fuzzy name matching, shared attributes like phone, address, or RFC/tax ID) to score how likely two records are the same entity.
- Resolution — cluster the matched records into a single resolved entity, keeping every source record traceable underneath.
- Linking — map how resolved entities connect to each other: ownership, shared addresses, transactions, family, employment.
The last step is what turns a clean list into intelligence. Once entities are resolved, the relationships between them become visible — and that's usually where the risk lives.
Entity resolution answers "who is this, really?" Link analysis answers "and who are they connected to?" You need both to see a network.
Entity resolution vs. deduplication: what's the difference?
They sound similar and often get confused:
- Deduplication removes redundant copies of a record inside one system so you store each thing once. It's a housekeeping task.
- Entity resolution decides whether records across many systems refer to the same real-world entity — even when they don't match exactly, and even when someone is actively trying to look like two different people. It's an intelligence task.
Deduplication makes your database tidier. Entity resolution makes it truthful about who's in it.
Where is entity resolution used?
Wherever the cost of confusing two entities — or failing to connect two — is high:
- KYC / AML / onboarding — resolving customers to real beneficial owners and screening against sanctions and PEP lists under aliases.
- Fraud — spotting rings that reuse addresses, devices, or bank accounts across "separate" identities.
- Due diligence & third-party risk — seeing the real ownership and connections behind a vendor before you sign.
- Investigations & anti-corruption — mapping how people, companies, and contracts actually connect.
- Supply-chain risk — finding hidden concentration when "different" suppliers trace back to one parent.
What makes entity resolution trustworthy?
A match is only useful if you can defend it. Good entity resolution is:
- Source-traceable — every resolved entity and every link points back to the original records that support it, so it holds up in audit, procurement, and court.
- Explainable — you can see why two records were merged, not just that they were.
- Reviewable — analysts confirm or split uncertain matches; the machine proposes, a human decides.
- Secure — sensitive and regulated data can be processed on-prem, sovereign, or air-gapped, not shipped to someone else's cloud.
How Axentra Sherloc does entity resolution
Sherloc is Axentra's entity-graph intelligence and search platform, and entity resolution is its core. It ingests everything you already hold — databases, warehouses, APIs, contracts, PDFs, case files, registries, structured or not — resolves the real people, companies, accounts, locations, and assets inside (across duplicate, misspelled, and aliased records), and maps how they all connect into one graph.
From there you search a single name and get the whole network back: multi-hop link analysis where every node and edge is source-traceable — defensible in audit, procurement, and court. You can search in plain English or Spanish, and because it's built for sensitive data, Sherloc runs on-prem, sovereign, or fully air-gapped. It's used for investigations, anti-corruption and procurement, organized crime, financial intelligence, and enterprise due diligence, KYC, AML, fraud, and supply-chain risk.
The point isn't to replace your analysts — it's to hand them the resolved, connected picture in seconds instead of weeks, with the evidence attached.
If you're trying to turn scattered, messy records into one accountable view of who's who and how they connect, let's talk.