Duplicate detection

Configure Delpha's duplicate detection engine. Learn how to set the Detection Threshold, enable Continuous Assessment (12-hour batch), and customize Matching Fields for Salesforce records.

Overview

The Detection Settings page allows you to control the "search engine" Delpha uses to find duplicate records. Here, you define when Delpha looks for duplicates (automation), where it looks (scope), and how strict the matching logic should be (thresholds).

As shown in your configuration screen, the settings are divided into General Settings (core activation) and Optional Configuration (fine-tuning logic).

General Settings

These core controls manage the activation and scheduling of the duplicate detection engine.

• Activate Duplicate Detection: The master switch. Turn this on to enable the detection rules defined on this page.

• Continuous Assessment: Automates the search process. When enabled, Delpha runs a batch process every 12 hours to identify new potential duplicates created or updated since the last run.

• Run Evaluation: A manual trigger button. Use this to force an immediate evaluation of eligible records without waiting for the scheduled batch process.

• Detection Threshold: Defines the sensitivity of the AI matching. Delpha assigns a similarity score (0-100) to every pair of records. Only pairs with a score greater than or equal to (≥) this threshold are flagged as potential duplicates.

  • Tip: A higher threshold (e.g., 90) means stricter matching but might miss subtle duplicates. A lower threshold (e.g., 70) catches more variations but may increase false positives.

• Duplicates at Creation: Configures real-time handling. This setting determines how Delpha reacts when a user attempts to create a record that matches an existing one (e.g., blocking the creation or showing a warning) vs allowing the record to be created and the duplicate being detected asynchronously.

Fine-Tuning Detection Rules

For advanced control, use the specific configuration sections to refine your matching logic.

Record Filter (Optional)

Restricts the scope of duplicate detection. Use this to exclude specific records from being checked.

• Example: Ignore all "Inactive" Contacts or "Archived" Leads.

Matching Fields (Optional)

Defines the "ingredients" of the matching score.

• You choose which fields contribute to the duplicate calculation (e.g., Name, Email, Phone).

• You set the relative impact of each field. For example, you might give "Email" a higher weight than "City" because a matching email is a stronger signal of duplication.

Filtering Rules (Optional)

Configure strict logic to pre-process matches before they reach the review stage.

• Exact-Match: Define rules where fields must match character-for-character.

• Filter/Exclude: Automatically discard pairs that don't meet specific business criteria.

• Annotate: Add tags or comments to potential duplicates to aid the manual review process.

Advanced Settings (Optional)

Technical customization for the initial detection phase.

• Screening Fields: Choose the initial fields used to "candidate" records for comparison.

• Placeholder Values: Configure Delpha to discard generic values (like "[email protected]" or "555-555-5555") so they don't trigger false positives.