TL;DR
- Automatic merging works best where matches are clear and a wrong merge is cheap to fix.
- Contacts and leads with an exact email match are the usual first candidates.
- Companies can follow once a domain-based search has been reviewed.
- Deals, tickets and custom objects carry money, commitments or process, so review them by hand first.
- In Dedupely, Auto Merge works on every synced object, so the choice is yours, search by search.
Once a cleanup is done, the next question is how to keep the CRM clean without merging every new duplicate by hand. Automatic merging is the answer for some records. For others, a person should still look before anything merges.
The difference usually comes down to the object: how clear its matches are, and what a wrong merge would cost. This post goes through the common objects and where each one usually lands.
Two questions for each object
How clear are the matches? An exact email or external ID match is almost always the same person or company. A name match isn't.
What does a wrong merge cost? Merging two contacts by mistake is annoying. Merging two deals by mistake can change pipeline numbers, commission and forecasts.
Automate where both answers are good: clear matches and a low cost of mistakes.
Object by object
| Object | Usually | Why |
|---|---|---|
| Contacts | Automate exact email matches | Clear matches; mistakes are easy to spot |
| Leads | Automate exact email matches | New leads repeat often, and speed matters for routing |
| Companies | Automate domain matches after review | Clear matches, but branches and subsidiaries need checking first |
| Deals | Review by hand | Amounts, stages and owners affect pipeline and commission |
| Tickets | Review by hand | Two tickets from one person may be two separate issues |
| Custom objects | Review first, then decide | Depends on what the object holds and who relies on it |
Start narrow, then widen
Within each object, automate the safest search first:
- Exact email for contacts and leads.
- Exact external ID where another system assigns one.
- Website domain for companies, once you've checked a sample for branches.
- Looser matches, like name plus phone, stay in manual review.
Keep a person in the loop
Even on automated objects:
- Hold back risky matches, like very large sets or placeholder values, for review.
- Send uncertain matches to manual review instead of merging them.
- Check what merged after the first few runs, and then regularly.
How Dedupely handles automatic merging
In Dedupely, Auto Merge works on every synced object, so you choose which searches to automate. You turn it on per Duplicate Search, with a schedule: as soon as possible, every set number of minutes, hours or days, on chosen days, or between hours you choose. Before it starts, you confirm the search, matching, filters and schedule. Auto Merge skips sets in Manual Review, holds back matches that trip your Safety Settings and sends low-confidence matches to Manual Review instead of merging them. Merges from Dedupely Intelligence stay manual: you review and merge its matches yourself.
Automate the clear cases, review the rest
Automatic merging fits objects with clear matches and cheap mistakes, like contacts and leads matched on exact email. Companies can follow after a review. Deals, tickets and custom objects usually stay with a person, at least until you trust the search.
Frequently asked questions
Which CRM objects should I merge automatically?
Start with contacts and leads matched on exact email. Add companies matched on domain after reviewing a sample. Keep deals and tickets in manual review.
Is it safe to auto merge deals?
Usually not at first. Deals carry amounts, stages and owners that affect pipeline and commission, so review them by hand.
Can automatic merging work on custom objects?
In Dedupely, yes: Auto Merge works on every synced object. Whether you should depends on what the object holds, so review a sample first.
What should stay in manual review?
Loose matches like name plus phone, very large sets, records with placeholder values and anything where a wrong merge would be costly.
How do I know a search is ready to automate?
Merge a sample of 20 sets by hand. If all 20 were clear, the search is a good candidate.




Pipedrive Marketplace