Fuzzy Matching Algorithm & Free Online Calculator

Try the WinPureFuzzy™ calculator below to compare two values and see their similarity score, then learn how fuzzy matching works and when to use it.

What Is WinPureFuzzy™ Matching?

WinPureFuzzy™ is WinPure’s general-purpose fuzzy string matching algorithm.

It compares two strings and calculates how similar they are on a scale from 0 to 100%, where a higher score indicates greater similarity.

For example:

This makes fuzzy matching particularly useful when working with customer, supplier, product, address and other business data where small variations can prevent traditional exact matching from finding the same entity.

How Does WinPureFuzzy™ Work?

WinPureFuzzy™ analyses the characters within two values and determines how closely they correspond.

The algorithm considers several characteristics of the strings, including:

• characters shared between the two values;
• the positions in which matching characters occur;
• character transpositions;
• the lengths of the values; and
• similarities within the leading characters.

These factors are combined to calculate an overall similarity score.

For example:

Jonathan Smith

and

Jonathon Smith

are not identical strings.

An exact comparison would therefore return no match.

Caption 1: How WinPureFuzzy™ evaluates character similarity, position and transpositions to produce an overall similarity score.

Fuzzy Matching Examples

Fuzzy matching can identify similarities between values that would fail a traditional exact match. The examples below show some common variations that WinPureFuzzy™ can help detect.

Try these examples yourself: Enter any of the value pairs into the WinPureFuzzy™ calculator above to see their similarity score.

How Is the Similarity Score Calculated?

WinPureFuzzy™ compares two text values and returns a similarity score between 0% and 100%. A higher score means the values are more similar, while a lower score indicates greater differences.

Rather than simply counting the number of different characters, WinPureFuzzy™ considers several factors, including:

• Matching characters — how many characters the two values have in common.
• Character position — where matching characters appear within each value.
• Transpositions — characters that appear in a different order or position.
• Leading-character similarity — similarities within the first few character positions receive additional weighting.
• WinPure-specific weighting — these signals are combined using WinPureFuzzy™’s own scoring behaviour.

For example, Jonathan Smith and Jonathon Smith are not identical, but they contain many of the same characters in similar positions. WinPureFuzzy™ recognises this and produces a much higher similarity score than it would for two unrelated values.

What Does the Score Mean?

These ranges are useful as a guide, but a similarity score is not a probability that two records represent the same entity. The appropriate matching threshold depends on the data, the field being compared and the other conditions within the matching rule.

How Does WinPureFuzzy™ Work Within a Matching Rule?

A similarity score is only one part of a reliable matching decision.

In WinPure, each matching condition can also be given a minimum score and a level of importance within the rule. This allows stronger fields to have more influence than weaker ones.

Why This Matters

This helps avoid relying too heavily on a single field.

A customer name may be very similar, but if the address or email does not support the match, the overall rule can still reject it. Equally, highly reliable fields can be given more influence where appropriate.

That means WinPureFuzzy™ can be used as part of a broader matching strategy rather than as a standalone similarity score.

Caption 2: WinPureFuzzy™ works alongside other matching methods, field-level thresholds and weighting to build stronger, more reliable record matches.

WinPureFuzzy™ vs Exact Matching

Exact matching answers a simple question:

Are these two values identical?

WinPureFuzzy™ answers a different question:

How similar are these two values?

Consider these customer records:

An exact comparison of the name would fail because Robert Thompson and Robbert Thompson are different strings.

Fuzzy matching can recognise the strong similarity between them.

What Is a Fuzzy Matching Threshold?

A fuzzy matching threshold determines how similar two values must be before they are considered a match.

For example, a rule might require:

Name similarity ≥ 85%

Records scoring below 85% would fail that particular condition, while records scoring 85% or higher could continue to satisfy the matching rule.

Higher thresholds generally produce stricter matching.

Lower thresholds allow greater variation but can also increase the possibility of false matches.

There is therefore no single threshold that is appropriate for every dataset.

The best threshold depends on the type of data being matched, its quality and the other fields being used within the matching rule.

When Should You Use WinPureFuzzy™?

WinPureFuzzy™ is particularly useful when values are expected to contain relatively small textual variations.

It can therefore be useful for data deduplication, record linkage, CRM cleansing, customer matching, supplier matching and Master Data Management projects.

WinPureFuzzy™ vs Jaro-Winkler

WinPureFuzzy™ and Jaro-Winkler both calculate string similarity, but they should not be considered the same algorithm.

Jaro-Winkler is a well-known string similarity algorithm based on matching characters, transpositions and a prefix adjustment.

WinPureFuzzy™ uses its own matching behaviour and weighting designed for WinPure’s data matching engine.

Although the algorithms share some underlying concepts, their calculations can produce different similarity scores for the same pair of values.

This is why WinPureFuzzy™ and Jaro-Winkler are available as separate matching choices within AdaptiveMatch™.

Rather than assuming one algorithm is always better, users can select the comparison method most appropriate for the type of data being matched

WinPureFuzzy™ vs Levenshtein Distance

Levenshtein Distance measures the number of edits required to transform one string into another.

Those edits typically include:

• insertion;
• deletion; and
• substitution.

WinPureFuzzy™ takes a different approach. It evaluates the similarity and positioning of characters rather than simply counting the number of edits required.

This means the two algorithms can respond differently to spelling mistakes, reordered characters and variations in string length.

Neither algorithm is universally superior.

The most appropriate algorithm depends on the characteristics of the data.

WinPureFuzzy™ vs Phonetic Matching

WinPureFuzzy™ compares the characters within strings, while phonetic algorithms compare how values are likely to sound.

For example, phonetic algorithms such as Soundex, Double Metaphone, NYSIIS and Caverphone can be useful when names have different spellings but similar pronunciations.

WinPureFuzzy™ is generally more appropriate when differences are caused by typographical errors, missing or additional characters, or transpositions.

AdaptiveMatch™ brings these approaches together. It allows different matching algorithms to be applied to individual columns and, where appropriate, enables fuzzy and phonetic matching to be used together. Users can also choose from different phonetic algorithms depending on the characteristics of their data.

AdaptiveMatch™ can automatically recommend the appropriate matching type, algorithm and score for each column, while still giving users full control to select or adjust them manually.

Advantages of WinPureFuzzy™

One of the main advantages of WinPureFuzzy™ is that it provides a flexible general-purpose comparison for imperfect business data.

It can help organisations:

• identify duplicates that exact matching misses;
• tolerate common spelling and typing variations;
• detect transposed or misplaced characters;
• control matching through similarity thresholds;
• combine fuzzy similarity with other matching conditions; and
• reduce the amount of manual record comparison required.

However, fuzzy matching should not normally be treated as proof that two records represent the same entity.

Two different people, companies or products can have highly similar values.

Reliable entity matching therefore usually considers multiple attributes rather than relying on a single fuzzy score.

Using WinPureFuzzy™ for Data Deduplication

Fuzzy matching becomes particularly powerful when it forms part of a multi-field matching rule.

Instead of asking whether two names are similar, a matching rule might consider:

Name — WinPureFuzzy™ ≥ 85%
Address — WinPureFuzzy™ ≥ 80%
Postcode — Exact Match

This provides substantially more evidence than comparing the name alone.

Different fields can also use different algorithms.

For example:

Customer Name → WinPureFuzzy™
Address → Token Sort
Postcode → Exact
Telephone → Exact

This is the principle behind algorithm-level matching in WinPure AdaptiveMatch™.

WinPureFuzzy™ in AdaptiveMatch™

WinPureFuzzy™ is available as a matching algorithm within WinPure AdaptiveMatch™, alongside a growing library of string similarity and phonetic algorithms.

AdaptiveMatch™ takes the guesswork out of matching by automatically selecting the most appropriate algorithm, match type and score for each column, while still giving users full control to fine-tune the configuration when needed.

Available algorithms include methods such as:

• WinPureFuzzy™;
• Levenshtein;
• Damerau-Levenshtein;
• Jaro;
• Jaro-Winkler;
• Smith-Waterman-Gotoh;
• Token Sort; and
• phonetic matching algorithms.

This makes it possible to construct matching rules around the actual characteristics of the data.

A person’s name, company name, address and product description may all require different approaches to matching.

AdaptiveMatch provides the flexibility to use the right algorithm in the right place.

Frequently Asked Questions

Written by

Team WinPure

The WinPure Team shares official updates on our products, features, and company news. From new releases and enhancements to behind-the-scenes developments, this space keeps you informed on how WinPure continues to deliver secure, reliable, and innovative data quality solutions.