Manual Review
Manual review is a step in deduplication where uncertain duplicate sets wait in a review queue until a person approves and merges them, or rejects them as different records. It holds the suspect duplicates that a matching rule found but that aren't certain enough to merge without a closer look.
Manual duplicate review sits between matching and merging. Clear matches, such as records that share an exact email address and name, can be merged in bulk or on a schedule. Sets matched on a single weak field, sets with conflicting values, very large sets and sets tied to high-value records go to the queue instead. A person compares the records, decides whether they describe the same person, company or deal, and acts on the set. The queue keeps uncertain sets out of automated merges without losing track of them.
See how Dedupely holds uncertain duplicates for review in your CRM
What Is Manual Review?
Manual review, also called duplicate review or a duplicate review queue, is the part of a deduplication process where a person decides what happens to a duplicate set. Matching finds records that look alike; manual review is where someone confirms whether they really are the same customer before anything is merged. It applies to every object a team cleans up: contacts and companies in HubSpot, accounts and leads in Salesforce, people and organizations in Pipedrive, leads in Close, and contacts in Nimble and Google Contacts.
Take two contacts matched on a similar name and the same company: "J. Smith, jsmith@acme.com, Sales Director" and "John Smith, john.smith@acme.com, CFO." They could be one person who changed roles, or two people at Acme who share a surname. A rule can't tell from those fields alone. A reviewer can check the phone numbers, the owners and the activity history, then merge the set or mark it as two different people.
Sets reach the queue in two common ways. In the first, a person spots a set that needs a closer look and sends it there by hand. Its strength is control: only the sets someone chose wait for review. In the second, rules route sets automatically, for example when a match falls below a confidence level or a field holds a value in the wrong format. Its strength is consistency: uncertain sets are caught even when nobody is looking. Many teams use both.
How Manual Review Works
Manual review often follows a few steps:
- Sort matches by certainty. After a duplicate search runs, a person or a routing rule sorts its results into sets that are safe to merge and sets that need a person. The clear sets move on to bulk or scheduled merging.
- Send the uncertain sets to the queue. Good candidates are sets matched on one weak field, such as a name alone; sets tied to high-value records, such as customers with open deals or contracts; sets where the records disagree on key values, such as two different phone numbers or owners; and large sets with many records.
- Hold them out of automated merges. Sets in the queue wait there until someone acts on them, and bulk and scheduled merging can skip them so they aren't merged by mistake.
- Review each set. The reviewer compares the fields, the activity and the linked records, and may check with the record owner or a team lead.
- Decide. A real duplicate is merged, often with the surviving record and winning values chosen by hand. A set that isn't a duplicate is rejected, so it stays out of future results. A set that needs more information can wait.
- Adjust the rules. If one search keeps sending false matches to the queue, a stricter field or filter on that search keeps them out next time, and the queue stays short.
Why Manual Review Matters for CRM Teams
A wrong merge is often harder to fix than the duplicate it removed. When two different customers are merged into one record, their emails, phone numbers, owners and activity end up together, and the record looks clean, so the error is hard to spot. Reps then contact the wrong person, reports count one customer where there are two, and in some CRMs the original records can't simply be brought back. Manual duplicate review puts a person in front of the sets where that risk is highest.
It also lets the rest of the cleanup move faster. Without a queue, a team has two choices for a mixed list of matches: check every set by hand, or merge them all and accept some mistakes. With one, the clear sets can be merged in bulk or on a schedule, and only the uncertain ones wait for a person. Automation then covers the volume, and people spend their time on the few sets that need them.
Some records carry more weight than others. A customer tied to billing, a signed contract or another system can affect invoices and integrations if it's merged wrongly, so a short review there protects more than the CRM.
A queue has its own cost. Every set in it is a duplicate that stays unmerged until someone reviews it, so it works best when it holds only the sets that need a person.
Getting Started With Manual Review
Start by deciding which sets need a person. A simple test is to ask what a wrong merge would cost. Sets matched on several strong fields can be merged in bulk, while sets matched on one weak field, sets with conflicting values, large sets and sets tied to important customers are the ones worth reviewing.
Send those sets to a review queue as you go through each search's results, and keep merging the clear ones. Set a regular time to work through the queue, and bring in the record owner or team lead when a set needs their knowledge of the customer.
When you review a set, decide on it fully: merge it, choosing the surviving record and values with care, or reject it so it doesn't come back. If one search keeps sending false matches to the queue, tighten that search instead of reviewing the same pattern each time it runs.
In Dedupely, Manual Review carries out this approach: you can send a duplicate set from any Duplicate Search to Manual Review, where it stays until it's merged or removed, and Auto Merge skips those sets and sends its own low-confidence matches there instead of merging them. Each field also has a "Send incorrect formats to manual review" checkbox, off by default, that sends invalid and filler values to Manual Review automatically.
Key Takeaways: Manual Review
Manual review is the step where uncertain duplicate sets wait for a person to merge or reject them, so clear matches can be merged in bulk or on a schedule without taking the uncertain ones along. It suits sets matched on one weak field, sets with conflicting values, large sets and sets tied to high-value records. In Dedupely, you can send a duplicate set from any search or Match Option to Manual Review, and it stays there until it's merged or removed. Auto Merge and Bulk Merge both leave out sets in Manual Review, Auto Merge sends low-confidence matches there instead of merging them, and a per-field checkbox can send values in the wrong format there too. Rejected Matches keeps false matches out of current and future searches, and Safety Settings hold risky matches in a separate Excluded tab.
Frequently Asked Questions About Manual Review
When should a duplicate set go to manual review instead of being merged automatically?
Send a set to manual review when a wrong merge would be costly or the match isn't clear from the data. Four situations come up often. The set was matched on one weak field, such as a name alone. The records disagree on key values, such as two different phone numbers, owners or job titles. The set holds many records, which can point to a shared placeholder value rather than one customer. Or the records are high-value, such as customers with open deals, contracts or billing links. Sets matched on several strong fields, such as an exact email plus name, are often safe to merge in bulk or on a schedule. In Dedupely, you can send a duplicate set from any search or Match Option to Manual Review, and it stays there until it's merged or removed. Auto Merge also sends its low-confidence matches there instead of merging them, and it skips sets that are already in Manual Review.
Can you merge duplicates in a CRM such as Salesforce or HubSpot without manual duplicate review?
Yes, for duplicates matched by rules you trust, but keep manual review for the uncertain ones. Duplicate leads and contacts that share an exact email address and name, for example, are often safe to merge automatically in Salesforce, HubSpot, Pipedrive or another CRM. Run a new search by hand first and check its results over a few rounds before automating it. Sets matched on looser criteria still benefit from a person's check, because a wrong merge is hard to reverse. In Dedupely, Auto Merge merges new duplicates on the schedule you choose, using a Duplicate Search's criteria. Before turning it on, you confirm the search, matching, filters and schedule, and acknowledge that it merges all current and future duplicates matching the search and that merges can't be undone. Auto Merge sends low-confidence matches to Manual Review instead of merging them, skips matches excluded by Safety Settings, and skips sets already in Manual Review.
What's the difference between rejecting a match and sending it to manual review?
Rejecting a match says the records aren't duplicates; sending it to manual review says someone still needs to decide. A rejected set is removed from the results so it doesn't come back each time the search runs. A set in review stays open: it can still be merged, rejected or left for later, once the reviewer has checked it or asked the record owner. Reject only the sets you've confirmed are different records, such as two people at the same company who share a surname, and send the rest to review. In Dedupely, Rejected Matches removes a duplicate set from current and future Duplicate Searches, and a rejected set can be restored with "Remove from Rejected." Manual Review holds a set until it's merged or removed, so the decision isn't lost. Sets in Manual Review are also left out of Auto Merge and Bulk Merge, which keeps them out of scheduled and bulk merges while they wait.
How is manual review different from matches held back by merge safety rules?
Manual review holds sets that were sent for a person's decision, while merge safety rules hold back sets that tripped a risk rule. Safety rules check every match against limits set in advance, such as too many records in one set or placeholder values like John Doe, and keep those matches out of bulk and automated merging. Manual review is where a set goes when someone, or a routing rule, decides a person should look at it. In Dedupely, the two are separate. Safety Settings, set per object, automatically exclude risky matches from bulk selection and Auto Merge, and matches that trip a rule show in the Excluded tab on the next scan. The rules cover records per match (20 by default), poor or example data, and very short values. Excluded matches can still be merged by hand, rejected or sent to Manual Review. Manual Review holds sets that you send there, plus the two automatic routes: Auto Merge's low-confidence matches and the per-field incorrect-format checkbox.
How do you review suspect duplicates without missing a wrong match?
Compare more than the fields that made the match, then choose the merged values yourself. Start with the strongest identifiers, such as email, phone and company, and look at where the records disagree: a different owner, job title or lifecycle stage can mean two different people. Check the activity history and linked records, and ask the record owner when the data can't settle it. For a set that is a real duplicate, pick the surviving record and the value each field keeps, and preview the result before merging. In Dedupely, Custom merge lets you open a duplicate set, pick the winning value field by field, preview the result and then merge. Merge Rules can also set the Primary Record and field rules, and a preview shows the merged result before anything changes in the CRM. Merges can't be undone, and Merge History keeps each merge's original records and values and can be downloaded as a CSV.
