Too Many False Positives

False positives are records that MatchLogic has grouped together as duplicates but which are actually distinct records. A high rate of false positives means your matching criteria are too loose. The fixes involve tightening your match definitions without eliminating genuine matches.

1. Raise the Similarity Threshold

If you are using fuzzy matching, the similarity threshold determines how close two values must be to count as a match. A lower threshold (e.g., Low or Medium) will produce more matches but also more false positives.

Fix: Increase the similarity level to High. This requires a higher degree of character similarity before two values are considered a match, reducing spurious matches at the cost of potentially missing some genuine duplicates with greater variation.

2. Add More Criteria Fields

Matching on a single field — such as first name only — will almost always produce false positives because many people share the same first name. Combining multiple fields dramatically improves precision.

Fix: Add additional criteria fields such as:

• Last name alongside first name
• Date of birth or age
• ZIP code or city
• Phone number or email address

The more discriminating fields you include, the less likely two genuinely different records will score high enough to be considered a match.

3. Reduce Weights on Weak Fields

Fields like city, department, or job title are useful for breaking ties but can cause false positives if weighted too heavily. If two records share a city but differ in every other field, a high city weight might still push them over the match threshold.

Fix: Lower the weight on fields with low discriminating power and increase the weight on unique or near-unique fields such as email or national ID.

4. Use Exact Matching for Structured Identifiers

Account numbers, customer IDs, national insurance numbers, and similar identifiers should always be matched exactly. Applying fuzzy matching to these fields can cause false positives when numeric values are slightly different (e.g., "10234" matching "10244").

Fix: Set the match type to Exact for any identifier field.

5. Review the Score Distribution

In Match Results, check the match quality report. If a large proportion of matched pairs score below 70, your criteria may be generating many low-confidence matches that are actually false positives.

Fix: Apply a score filter to view only matches above a threshold (e.g., 75 or 80) and check whether the remaining matches are more accurate. Consider adjusting criteria so the natural score for genuine duplicates is consistently above that threshold.

6. Use "Not Duplicate" to Exclude Known False Positives

In Match Results, individual pairs that are confirmed false positives can be marked as Not Duplicate. This excludes them from the merged output without requiring you to retune the entire configuration.