How to set up duplicate rules
Duplicate rules encode how your CRM decides two records are the same. Get them right and duplicates never form; too loose and you get false merges, too strict and duplicates slip through.
Set rules per object, because what makes two contacts duplicates differs from what makes two accounts duplicates.
Short answer
Set up duplicate rules by choosing match keys per object (email for contacts, domain plus name for accounts), deciding whether each rule blocks or just warns, tuning fuzzy thresholds, and defining merge logic for which value survives. Start strict on blocking the obvious duplicates, then widen fuzzy matching as you gain confidence.
Step by step
Choose match keys per object
Pick the fields that identify each object: email for contacts, web domain plus company name for accounts, email or name plus company for leads.
Decide block versus warn
For high-confidence keys like email, block creation of a duplicate. For fuzzy matches, warn and let the rep decide, to avoid false blocks.
Tune fuzzy thresholds
Set the similarity threshold so it catches typos and variants without flagging genuinely different records. Start conservative and loosen as you validate.
Define merge logic
Specify which value survives for each field when a merge happens, so merges are consistent and do not need a manual decision every time.
Block the obvious, warn on the fuzzy
The safe pattern is to hard-block exact high-confidence duplicates (same email) and only warn on fuzzy matches, so you never wrongly block a legitimate record. As your fuzzy matching proves accurate, you can promote some fuzzy rules to blocking.
How Ardovo helps
Ardovo ships sensible duplicate rules per object out of the box - exact blocking on strong keys, confidence-scored fuzzy matching with review. Rook tunes and applies them automatically, so you get duplicate prevention without hand-building and babysitting rule sets.
Frequently asked questions
What match keys should duplicate rules use?
Per object: email for contacts, web domain plus company name for accounts, and email or name-plus-company for leads. Strong unique identifiers like email make the best blocking keys; softer signals like name are better for fuzzy warn rules.
Should duplicate rules block or warn?
Block on high-confidence exact matches like a shared email, where a duplicate is almost certain. Warn on fuzzy matches so reps can judge borderline cases, avoiding false blocks that frustrate users and hide legitimate new records.
How do you avoid false duplicate matches?
Use strict thresholds for blocking rules and looser ones only for warn rules, score matches by confidence, and require review before merging borderline candidates. Starting conservative and loosening as accuracy proves out avoids wrong merges.