Some duplicate are not detected, what can I do?

What to Do If Legitimate Duplicates Are Not Being Detected in Delpha

If some valid duplicates are not flagged by Delpha's duplicate detection process, there are several ways to improve match detection and scoring accuracy.

This article outlines the key configurations you can adjust to optimize the detection process and identify more true duplicate pairs.

1. Add Additional

What Are Screening Fields?

Screening fields are used during the initial filtering phase to decide whether two records should even be considered as a potential match. By default, this relies heavily on the Name field.

If two records differ significantly in name but match on other key fields (e.g., email, phone, domain), they may be missed.

How Screening Fields Help

• ✅ Adding relevant fields (e.g., Email, LinkedIn URL) increases the chance of catching real duplicates that don't match on name alone.

• ⚠️ Using too many screening fields can increase processing time and generate low-confidence pairs (false positives).

Example: Add Website and LinkedIn URL as screening fields for Company duplicate detection or add Email as screening for Contact duplicate detection

2. Review and Adjust Rules

• Mandatory fields reduce false positives by requiring data in both records for scoring.

• Optional fields allow records with missing data to still contribute to a pair’s score.

📌 Tip: Consider relaxing some fields from Mandatory → Optional to allow more flexible matching.

Add New Optional Fields

• Introducing new fields as optional comparison points (like Industry, Phone, Region) can help boost the match score and uncover missed duplicates.

3. Improve Your Underlying Data Quality

Duplicate detection accuracy directly depends on the quality and completeness of the data being evaluated.

Steps to Take:

• Run Data Quality assessments in the Data Steward View

• Ensure key fields used in deduplication (e.g., Name, Email, LinkedIn

Summary