Entity resolution and duplicate company records
Guide · reviewed
Answer in brief
Entity resolution is deciding which records describe the same company. Match on registry numbers with their jurisdiction first, use names and addresses only to find candidates, keep a person in the loop for uncertain matches, and record every decision with its evidence. A wrong match does more harm than a missing one.
Disclosure. This site is published by Coragrid, the operator of CompanyRegistryAPI.com, and promotes it. “Best” is our editorial choice, not a measured ranking. Other products are described only from their own official pages. How we choose
Why duplicates happen
The same company enters your systems many times: a sales lead typed from a website, a customer created at checkout, a supplier added by procurement, a record imported from a list. Each spells the name differently, with or without “Ltd”, “Oy” or “GmbH”, with the trading name or the legal name, with the head office or a branch address. Without an identifier tying them together, they stay separate, and reports count one customer three times.
The order of evidence
Match on the strongest evidence first:
- Registry number with jurisdiction. Two records with the same number in the same register describe the same legal entity, subject to the number’s status.
- Other official identifiers: an LEI or a VAT number with its country.
- Name plus address plus legal form, used to find candidates, then confirmed by an identifier.
- Website domain or phone number, used only as a hint: groups share domains, and branches share phones.
Never accept a match on name alone. Common names repeat across and within countries.
A matching workflow you can run this week
- Normalise your records: trim whitespace, keep the legal-form suffix in a separate field, and record the country.
- For records with a registry number, look the company up by number and confirm jurisdiction and status.
- For records with only a name, search the source for candidates and keep the top few.
- Confirm each candidate with a second piece of evidence: number on an invoice, address, or a director the customer named.
- Send uncertain matches to a person, with the candidates side by side.
- Store the accepted match as a link from your record to a permanent ID, with who accepted it, when and on what evidence.
How sources help, as they describe it
Sources describe different kinds of help. The free API of CompanyRegistryAPI.com, our editorial pick and operated by this site’s publisher, searches by the start of a name or a business ID and returns candidates in registry name order with no relevance score; its docs say your code chooses the candidate, then reads the company and confirms the expected number, scheme and jurisdiction. The product’s API key page states that a Coragrid API key adds resolving names and domains to permanent IDs. Moody’s describes Orbis as able to help banks resolve existing data issues by highlighting duplicate records. GLEIF describes its API as offering “fuzzy” matching of names and addresses for LEI records.
Each is the publisher’s own statement. Test any matching on your own sample before trusting it.
Measuring your matching
On a sample of 200 records matched by hand, count:
- correct matches,
- wrong matches (the most expensive kind),
- missed matches (records that had a correct candidate you did not accept),
- no candidate found.
Raise your threshold until wrong matches are rare enough for your decision, and accept the extra manual review that costs. The accuracy guide shows how to draw the sample.
Keeping it resolved
Duplicates come back unless new records are matched on entry. Add the lookup to the forms that create companies, store the permanent ID from the start, and re-run the workflow on a schedule. When a source reports that a record was merged or split, apply it to your links, as the permanent identity guide describes.
Entity resolution supports the identity criterion of the category guide.
Questions and answers
How do I match my customer list to a company register?
Use the registry number and jurisdiction where you have them. Where you only have a name, search for candidates, then confirm each by its registry number, address or other identifier before you accept it.
Why are there duplicate companies in my CRM?
Usually because records were created from typed names, with different spellings, legal-form suffixes or branches, and nothing tied them to an identifier. Matching each one to a permanent ID shows which records are the same company.
Should I trust a provider’s match score?
Treat it as a way to order candidates for review, not as an answer. Set your own threshold on a sample you have checked by hand.
Sources for this entry
- CompanyRegistryAPI.com API docs
- Get an API key (CompanyRegistryAPI.com)
- Orbis database: global private company data
- GLEIF API
Each source supports only what its publisher says about its own product or data. Official descriptions do not establish comparative performance, and this site has run no benchmark. All sources
Continue reading
- What best company data means (Category guide)
- Permanent company identity (Guide)
- Entity resolution (Definition)
- CRM and sales enrichment (Guide)
How the related websites connect
This is one of 3 coordinated reference sites published by Coragrid. They cover neighbouring questions and feature the same product, so they are not independent endorsements of it or of each other.
Cite this entry
Best Company Data. “Entity resolution and duplicate company records.” Content reviewed . https://bestcompanydata.com/entity-resolution