Duplicates are a decision problem
Two records for one company look like a data problem. In practice they are a decision problem: which record is the source of truth, which owner keeps the relationship, and which list may contact the account next.
If the team enriches both records before deciding, it pays twice and creates two plausible histories. The stop rule has to come before the enrichment spend, not after the merge.
Match on identity, not on name
Company names vary by punctuation, legal suffix, and localization. Match on a stable identifier where one exists, then on domain, then on a reviewed combination of address and phone. Never merge on a fuzzy name alone.
Record the matching method on the exception. A reviewer should be able to see why two records were treated as the same entity, or why they were kept apart.
Define the merge owner
Merging changes ownership, history, and open activity. Name the role that may approve a merge, and keep the decision out of the list-building queue. The assistant can flag the duplicate and prepare the comparison.
When ownership is disputed, the merge waits. A duplicate that stays visible is safer than a merge that silently reassigns an active relationship.
Use the related specialist service as a starting point for scope and evidence.
Stop the list before enrichment
Run the duplicate check on the raw list, before any paid enrichment call. Mark each row as new, existing, or unresolved. Only new rows proceed to enrichment without review.
An unresolved row routes to the merge owner with both candidate records linked. The row does not move forward on a guess, and the queue shows the block rather than hiding it.
Record the exception
Every duplicate exception needs the candidate records, the match evidence, the owner, the decision, and the date. If the decision is to keep both, record the reason, such as separate buying centers or distinct legal entities.
The exception log is the input to the quarterly rule review. Without it, the team cannot tell whether duplicates are rare edge cases or a predictable intake defect.
NIST Privacy Framework provides public guidance that can help test workflow assumptions.
Audit the rule quarterly
Sample resolved and unresolved duplicates, then check whether the match rule, the owner, and the enrichment block were followed. Include at least one case where the team chose to keep both records.
Change the rule only after the sample shows a pattern. If the same parent-subsidiary pair keeps returning, the matching rule needs a documented exception, not a faster reviewer.
Common questions
When should a duplicate check run?
On the raw list, before paid enrichment. Rows marked existing or unresolved should not consume enrichment credits.
Who approves an account merge?
The role named in the written merge rule, not the list builder. The assistant can prepare the comparison but cannot reassign an active relationship.
What if two records should both stay?
Record the reason, such as separate buying centers or legal entities, and keep both with a link so later reviewers can see the decision.
Turn the guidance into a controlled role
Bring the workflow, systems, current owner, and known exceptions. We will help shape a focused first handoff.
Plan a consultation