Start with email domain
The single highest-yield pass takes minutes. Pull every contact, reduce each email address to its domain, and group accounts by that domain. Any domain attached to more than one account record is a candidate. This catches the duplicates that name matching misses, because Acme Corp and ACME Corporation share nothing as strings and share everything as a company.
Filter out the noise before reading the output. Free mail providers produce false groupings. So do parent companies whose subsidiaries legitimately share a domain, and those need a decision about account hierarchy rather than a merge.
Then match on normalized names
| Step | What it does |
|---|---|
| Lowercase and strip punctuation | Makes "Acme, Inc." and "acme inc" comparable |
| Remove legal suffixes | Drops Inc, LLC, Ltd, GmbH, Corp before comparison |
| Collapse whitespace and articles | Handles "The Acme Group" against "Acme Group" |
| Apply fuzzy distance scoring | Surfaces typos and transpositions the exact match misses |
Work the queue in the right order
Sequence by what the current quarter depends on. Accounts with open opportunities come first, because those records feed pipeline coverage and the roll-up that leadership reads. Accounts with closed-won history come next, since split history distorts renewal and expansion planning. Dormant leads come last, and honestly some of them never need to be touched.
Before merging, re-parent open opportunities onto the surviving record and confirm the surviving record keeps the earlier create date. Losing a create date breaks cohort reporting quietly, and nobody notices until a retention analysis comes back with an impossible curve.
Confirm the ambiguous pairs through activity rather than names. Two records for the same company usually show alternating activity, where one goes quiet exactly when the other becomes active. ORM counts meaningful activity on an opportunity as a change in stage, close date, or amount, and that same signal works here. When one record's field changes stop on the week another record's begin, you are looking at one deal recorded twice.
Close the entry points
Detection is the easy half. Duplicates return unless form fills match against existing accounts before creating a lead, imports carry a required matching key, and reps search before they create. Run detection quarterly, keep entry controls permanent, and the duplicate backlog stops being a recurring project. That is what keeps sales forecasting rolling up to the right accounts instead of quietly double-counting them.
Frequently Asked Questions
What is the fastest way to find duplicate accounts?
Group accounts by the email domain of their contacts. One domain mapping to several account records surfaces the majority of real duplicates in a few minutes, and it works even when company names are spelled inconsistently. Exclude free mail domains and known holding structures before you read the results.
Should you use the CRM's built-in duplicate rules or an external tool?
Native rules catch exact and near-exact matches at the point of entry, which is where you want the cheapest catch. They are weak on the historical backlog and on cross-object matching. Export a snapshot and run the backlog analysis outside the CRM, then use native rules to hold the line going forward.
How do you review matches without spending weeks on it?
Tier the candidate pairs by confidence. Exact matches on a strong key merge automatically. Middle-confidence pairs go to a review queue owned by one person rather than distributed to reps. Low-confidence pairs get left alone until something else confirms them, because a bad merge costs more than a surviving duplicate.
Which duplicates should you fix first?
Accounts carrying open pipeline. They affect the current quarter's numbers, they are few enough to review by hand, and merging them protects the revenue reporting that leadership is already looking at. Dormant lead duplicates can wait, since nothing downstream reads them this quarter.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like how do you find duplicate records in a crm? into prescriptive action for your team.
Schedule a Demo