Fuzzy Matching

What is the Fuzzy Matching?

Fuzzy logic is a method of reasoning that allows systems to handle approximate or “inexact” matches, rather than requiring exact equality. In data matching, it helps detect similar or related values even when there are variations in spelling, formatting, or naming conventions.

Eg: Records using exact values (e.g., "Abcd Inc" only matches "Abcd Inc"). With fuzzy matching, Syncari can identify that "Abcd Inc", "Abcd", and "ACME Incorporated" likely refer to the same company.

Preparing Fields for Fuzzy Matching

Before using fuzzy matching in filters, the fields on which you need fuzzy match need to be enabled for fuzzy match in Schema Studio.

1. Open Schema Studio
2. Select the entity you want to configure (e.g., Contact, Account)
3. Click on a text field (e.g., Name, Email, Company)
4. Check the "Supports Fuzzy" option
5. Save your changes and Publish

Only text fields on Syncari entities can be enabled for fuzzy matching. Number, date, and other field types are not supported.

Fuzzy Matching Operators

Syncari provides six fuzzy matching operators. Each one is designed for a different type of matching scenario.

Resembles

Finds values that look similar to the search term by comparing character patterns.

• When to use: General-purpose matching for typos, abbreviations, and minor spelling differences
• Threshold: 0–100% (higher = stricter). Default: 30%
• Examples:
  • "Syncari" matches "Syncary"
  • "john.doe@email.com" matches "jon.doe@email.com"

Contains Words Like

Finds values where part of the text matches the search term.

• When to use: When the value you're looking for may be embedded in a longer string
• Threshold: 0–100% (higher = stricter). Default: 50%
• Examples:
  • "John" matches within "John David Smith"
  • "Abcd" matches within "Abcd Corporation Ltd"

Sounds Like

Finds values that sound like the search term when spoken aloud (in English).

• When to use: English names with different spellings but the same pronunciation
• Threshold: None — it's a yes/no match
• Examples:
  • "Smith" matches "Smyth"
  • "Stephen" matches "Steven"

Nearly Matches

Finds values within a set number of character changes (typos) from the search term.

• When to use: When you want precise control over how many typos are allowed
• Threshold: 1–5 maximum character changes. Default: 3
• Examples (with max 1 typo):
  • "acme" matches "acne" (one letter different)
  • "test" matches "tast" (one letter different)

Loosely Matches

Finds values that sound similar, with a confidence score indicating how close the match is.

• When to use: Name variations where you want to control how loose the match can be
• Threshold: 0–4 confidence scale (4 = sounds identical). Default: 3
• Examples:
  • "Robert" matches "Rupert" (confidence 3)
  • "Catherine" matches "Katherine" (confidence 4)

Is a Variant Of

Finds values that are phonetic variants, with strong support for international and non-English names.

• When to use: Matching names across languages and cultural spelling variations
• Threshold: None — it's a yes/no match
• Examples:
  • "Schwarzenegger" matches "Shvartseneger"
  • "Czerny" matches "Cherny"

Choosing the Right Operator

Tip: You can combine multiple fuzzy operators in a single rule. For example, use "Resembles" on email and "Sounds Like" on last name together in a dedupe rule.

Where to Use Fuzzy Matching

1. Merge studio Dedupe Rules

2. Lookup Syncari Records function

3. Data Studio Filters

Understanding Match Scores

When you use fuzzy filters in Data Studio, a score column appears alongside your results. Scores range from 0 to 100:

Note: Sounds Like and Is a Variant Of are yes/no operators — they don't produce score columns.