Skip to main content

How Fuzzy Matching Works in Entity Search and Duplicate Search

Issue

Users report that search "misses" records that definitely exist — or finds a record on one spelling but not on a very similar one.

The usual complaints:

  • Shortening a first name works, but shortening a surname finds nothing.

  • Adding more search criteria returns fewer results, not more.

  • A record is found with a hyphen included, but not without it.

  • One wrong digit in a date of birth returns nothing at all.

This is expected. Fenergo search tolerates typos, but it is not a SQL-style wildcard search.

Key Points

  • Search allows up to 2 wrong characters per word. Never more, however long the word.

  • Punctuation splits words apart. Ashworth-Bell is stored as two words: Ashworth and Bell.

  • First names are forgiving. Surnames, entity names, dates and dropdowns are not.

  • Every field you fill in must match. More criteria = fewer results.

Search is not case sensitive.

Resolution

The four rules

1. Up to 2 wrong characters per word.
A wrong character means one missing, extra, replaced, or swapped letter. Ashwrth finds Ashworth (1 missing). Ashwrt does not reach it (2 missing is the limit, and this needs more).

2. Punctuation and spaces split words apart.
Each piece is stored and searched separately, and the pieces are never rejoined. So Ashworth-Bell and AshworthBell behave completely differently.

3. First names get an extra shortcut. Nothing else does.
A first name also matches on just its first three letters — so Kat finds Katherine. Surnames and entity names have no such shortcut, so Ash will never find Ashworth.

4. Some fields must be exact.
Dates (including date of birth) must match to the day. Dropdown fields like Country and Entity Type must match exactly. Any difference removes the record from your results completely.

Example

An entity exists in the system:

First Name: Katherine Surname: Ashworth-Bell Date of Birth: 12 March 1975

You search for

Result

Why

Ashworth-Bell

Exact match

Both words match perfectly

Ashworth

Found

Matches the first stored word

Bell

Found

Matches the second stored word

Ashwrth-Bell

Found

One missing letter — within the limit

AshworthBell

Not found

No hyphen makes it one long word, too far from Ashworth

Ash

Not found

Far too short, and surnames get no shortcut

Kat + Ashworth-Bell

Found

First names match on their first three letters

Katherine + Ash

Not found

Same shortening, but surnames don't allow it

Katherine Ashworth-Bell + DOB 12 March 1976

Not found

One wrong digit in a date removes the record

The same idea applies to company names. If an entity is stored as Northwind Capital Partners Ltd, searching Northwind or Northwind Capital finds it. But searching all four words against an entity stored only as Northwind will not — you typed more words than the record contains.

How to search well

  1. Type less, not more. Start with the most distinctive word and add criteria only if you get too many results.

  2. Never abbreviate a surname or entity name. Type it in full, or type just its first word.

  3. Match the punctuation as it is stored or leave that part out.

  4. Only add a date or dropdown if you are certain. These are the criteria most likely to silently hide the record.

  5. Skip generic words like Ltd, Limited or Holdings. Thousands of records contain them.

  6. If a record won't appear, remove criteria one at a time to find which field is excluding it.

Verification

  • Search on one thing only — the entity's most distinctive name word. If the record appears, another criterion was excluding it.

  • Copy the name exactly as shown on the Entity Profile page and search that. It should always return an exact match.

  • If it still returns nothing, this is not fuzzy matching — raise a support ticket with the entity name and the exact text you searched.

Did this answer your question?