Name Matching Across Documents: Accents, Maiden Names

Name matching means comparing the name written on one document with the name on another document or register and deciding whether they refer to the same person. Exact string equality is the wrong test. "José García-Lopez" on a passport, "JOSE GARCIA LOPEZ" on a bank statement and "Garcia, Jose" on a payslip are one person, while "Jon Smith" and "John Smith" with different dates of birth may not be. A usable rule normalises the name, compares the parts that matter, and sends the rest to a person with a defined next step.

This article is for informational purposes only and does not constitute legal, financial, or regulatory advice. Regulatory references are accurate as of October 2026 and cover the United Kingdom unless stated otherwise. Consult a qualified professional for your situation.

What does name matching compare, and why not use exact equality?

Name matching compares the surname, forename(s) and a second anchor field (usually date of birth) between two documents, after removing differences that carry no identity information. Exact equality fails on harmless differences and still passes a wrong person who happens to share a string, so it is both too strict and too loose.

Under regulation 28 of the Money Laundering Regulations 2017, regulated firms must verify identity on the basis of documents or information from a reliable source independent of the customer, which means the name has to be reconciled between sources, not just read once (MLR 2017, reg. 28). Lettings agents, recruiters and lenders outside AML scope face the same practical problem: the ID, the payslip and the bank statement were each typed by a different party.

This article sits inside our wider document verification guide. For the other comparisons in a file (address, income, bank details), see cross-document verification checks.

Which name differences are common, and what does the reviewer do?

Seven kinds of difference explain most name mismatches in applicant files. Each has a source field, a comparison field, a verdict and a next step.

The point of the table is not the examples but the three verdict classes. Cosmetic differences are accepted automatically with the normalisation rule logged. Explainable differences need one more document. Unexplained differences stop the file.

How should accents and diacritics be handled?

Strip accents and fold case before comparing, but keep the original spelling for the record. A name written "Müller" can legitimately appear as "Mueller" or "Muller", and a name written "Åsa" as "Asa" or "Aasa", depending on who typed it and which system stored it.

Passports add a second layer. The machine-readable zone at the bottom of a passport is limited to the letters A to Z and the filler character "<", so accented letters cannot appear there and each issuing state converts them by its own convention under ICAO Doc 9303. The visual zone printed above can show the accent; the machine line cannot. A mismatch between the printed name and the machine line is therefore expected for accented names and is not a sign of tampering.

Practical rules:

1. Compare on a folded form (lower case, no accents, no punctuation) but display both forms to the reviewer.
2. Treat "ü / ue / u" and "å / aa / a" as equivalent in the comparison.
3. When the printed name and the machine line disagree on more than diacritics, flag it.

How do maiden names and name changes show up?

A maiden-name mismatch looks like a different surname with the same forename and date of birth. In the UK a person may change name through marriage, civil partnership, divorce or deed poll, and the government describes the deed poll route on GOV.UK. Nothing forces a person to update every document at once, so a file can legitimately contain an older bank statement in one surname and a new ID in another.

Handle it in four steps:

1. Confirm the forename and date of birth match across the two documents.
2. Ask for the document that links the names: marriage or civil partnership certificate, decree document or deed poll.
3. Check that the linking document names both surnames and that its date falls between the two documents' issue dates.
4. Record the link, so the next reviewer does not ask again.

A name-change document proves the link between two names; it does not by itself prove that the person presenting it is the person named. Keep identity verification and name linking as separate steps.

What about compound names and several surnames?

Compound forenames and multiple surnames are compared as sets of parts, not as one string. "Mary-Ann", "Mary Ann" and "Maryann" should resolve to the same forename; "Smith-Jones", "Smith Jones" and "Jones" (when a person uses only one half) need a rule.

The safe rule is: every part present on the narrower document must appear on the fuller one, and the surname part must match. "Jones" against "Smith-Jones" passes the first test; "Smith" against "Smith-Jones" passes too. Both are accepted only when date of birth also matches, and both are logged as partial matches. Prefix particles ("van", "de", "O'", "Mc") are a frequent trap: "McDonald" and "Mc Donald" are the same, "Van der Berg" may be filed under V or B in a payroll system.

How do you handle transliteration from other alphabets?

Transliteration differences are matched against the passport machine-readable line, which acts as the agreed Latin spelling for that person. Names written in Arabic, Cyrillic, Chinese or other scripts have no single Latin form, so "Mohammed", "Muhammad" and "Mohamad" can all be correct renderings.

The reviewer's question is not "which spelling is right?" but "does this spelling connect to the identity document?". If the passport machine line reads MUHAMMAD and the payslip reads Mohammed, with the same date of birth, accept and log. If the bank account is held under a third spelling, ask the applicant for a bank letter or have the bank statement re-issued with the passport spelling. Only fuzzy matching (edit-distance scoring) should be used to propose a match, never to approve one alone.

What do users ask about name mismatches?

People who handle onboarding files and people who are being asked to prove their name tend to raise the same questions. These two come up most.

"My payslip has my married name but my passport has my maiden name. Will the check fail?" It will raise a flag, not fail by itself. Provide the marriage certificate, or update the passport through GOV.UK passport name changes. Reviewers then record the link and move on.

"The bank spells my name without the accent. Is that a problem?" No. Accents are removed by many systems, and a good comparison folds them. The date of birth and surname parts are what count.

Another frequent question is how much fuzziness is acceptable. Our answer: none for date of birth, some for forenames, and a documented rule for surnames.

Where can names be checked against a register?

A register check confirms a name against an authoritative record, which is different from comparing two documents the applicant supplied. For companies, Companies House lists officers by name and shows company status. For individuals there is usually no open public register, so the passport and a government-issued linking document are the anchors.

A register lookup shows that a person or company with that name exists on that date; it does not show that the document in your hands is genuine. Use it as an additional source, as regulation 28 expects, and state in the file which source answered which question.

Setting thresholds your team can apply

Write the rule down. A short policy removes the situation where two reviewers decide the same pair differently:

• Always required: date of birth equal, surname part equal after normalisation.
• Allowed with logging: missing middle names, hyphens, accents, order swaps, one transliteration variant.
• Needs a linking document: any surname that differs, any forename that differs by more than a transliteration variant.
• Must escalate: no link possible, or the account holder is a different person.

Apply data minimisation: compare only the name fields you need, in line with the ICO's data minimisation principle.

Automating the comparison

A validation tool should extract the name fields from each document, normalise them, and show the result field by field with the reason for each difference. CheckFile compares fields, dates and register data across a file and flags differences for review; it does not certify that any document is authentic. To see how this fits a bank or KYC onboarding file, look at our KYC and banking solution, check pricing or read how we handle data in security.

Frequently Asked Questions

What is name matching on documents?

It is the comparison of a person's name across several documents or registers, after normalising accents, case, punctuation and order. The result is a verdict (cosmetic, explainable or unexplained) with a defined next step.

Should "John" and "Jon" be treated as the same name?

Only when the surname and date of birth match and your policy allows a one-letter forename variant. Log the rule applied so the decision can be reviewed later.

What document proves a change from maiden to married name?

A marriage or civil partnership certificate is the usual link, or a deed poll for a legal name change. Divorce documents can link a name reverted to the earlier surname.

Is fuzzy matching enough to approve a file?

No. Fuzzy scoring is useful to suggest that two names may match, but approval should rest on stated rules and, where a surname differs, on a linking document.

Does a name match prove the document is genuine?

No. It shows that two documents agree on a name. A forged document that copies the same name would match, so register or issuer checks remain a separate control.