Job crm-duplicate-pairs-reviewed-before-any-merge · revised 12 October 2026
Find duplicate contacts and companies in your CRM, and check them before any merge
You get duplicate groups in three bands, each with the fields that matched and a pre-merge export. We merge nothing: your admin decides and merges.
You might be seeing
- Two reps contact the same person or company in the same week
- Contact and company counts in reports are higher than anyone believes
No passwords, keys, card details or admin invites needed to start.
What usually happened
Duplicate contacts and companies build up from imports, forms and reps. A merge cannot be undone in HubSpot, so merging on a guess destroys data. Nobody trusts a bulk merge, so the duplicates stay and the reports stay wrong.
Who it’s for: Marketing or sales-operations manager at a 10 to 200 person firm whose CRM holds the same buyer more than once.
Usually starts when: Reps ring the same buyer twice, reports double count, or a migration or campaign is coming.
The result: You receive duplicate groups in three bands (certain, likely, uncertain), each with the fields that matched, the rule you approved for which record to keep and a pre-merge export. Your admin merges the certain band and leaves the rest.
Check whether this job fits
Answer these without sending files or logins. Nothing is submitted unless you choose to contact us.
What you get
- The match-rule sheet
- The grouped list with evidence
- The pre-merge export
- The re-run report
Included
- One CRM and up to 20,000 contacts and companies, from an export or a read-only user
- A written match-rule sheet: what makes two records the same, which survives, how conflicts resolve
- Groups in three bands, each with the matching evidence
- A pre-merge export that reproduces the old values
- A re-run after your admin merges, to show no certain groups remain
Not included
- Merging, deleting or editing any record
- Records that hold special-category data
- Stopping new duplicates at the source: that is a separate job
- Third-party data matching or buying data
- A promise that the likely or uncertain bands are right
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
On a review sample of 50 groups drawn from the certain band after the rules are frozen, your named person finds 0 wrong pairs, which shows a wrong rate under about 6% with 95% confidence.
Evidence: The sample list with your marks.
In the likely band your person marks 30 sampled groups, and the wrong count is reported for information. Any wrong pair in the certain band fails the whole test.
Evidence: The marked sample and the wrong count.
Every pair in the uncertain band is left unmerged and listed with its evidence.
Evidence: The uncertain list and a count that equals the groups in that band.
After your admin merges the certain band, a re-run finds 0 certain groups, apart from any listed as blocked by the 250-merge limit, and the pre-merge export reproduces the old values on a test copy.
Evidence: The re-run report and the restore check.
Sign-off. You re-run the test yourself, check the list of severe errors and sign off in writing. Payment follows sign-off.
If it fails. If the agreed pass mark is not met, we say so, hand over the results and the failures, and do not call the output working. You do not pay for this scope; we agree whether to stop or re-quote.
When it fits, and when we stop
It fits when
- An admin on your side can run the merges and approves the rule sheet
- The match fields (email, domain, phone, name) are filled for most records
- Real records are read only after a written agreement that covers how your records are processed and a separate access step you control; until then we test on invented or redacted data.
We stop and tell you if
- You want us to merge records
- Another system re-creates the duplicates after each merge
- The records hold special-category data
What could go wrong
We change nothing in your CRM, so there is nothing to undo. The pre-merge export lets you restore old values by hand.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A wrong merge destroys data that cannot be restored | We merge nothing. The certain band is sampled before you act, and the pre-merge export keeps the old values. |
| The likely band is read as safe | Its wrong count is reported for information, not as a pass mark, and a person reviews the band. |
| Personal data in the export | Only agreed columns are shared, through the agreed route, after written terms. Our copies are deleted at the end. |
A separate automated reviewer checks the groups, the review sample and the pre-merge export before hand-over, and no person on our side reviews unless your agreement names one. Your named admin approves the rule sheet and decides every merge. At launch the work is largely automated, and we say so.
How we deliver
We arrange the work and independent review, then show you the result against the agreed checks. You keep authority over your systems.
- Agree the match rules, which record to keep and the pass marks in writing
- Receive the export by the agreed route, redacted where the job allows
- Group records into certain, likely and uncertain bands with the evidence
- Draw the review sample after the rules are frozen and record the results
- Independent automated review of the groups, the sample and the export
- Hand over the lists; your admin merges and we re-run
This is a one-off job, not a subscription or emergency cover. We confirm eligibility, price, a start window and a delivery date before you accept, and work starts only after agreed inputs and access are in place. Nothing is sent, merged, posted or changed in your live systems by us, and an enquiry creates no charge or booking.
Need to keep it working?
Discuss a monthly list of duplicate groups still outstanding.
Ongoing work is separately scoped and quoted: no monitoring, response-time guarantee or automatic subscription is included in this job.
Explore an ongoing managed lane, or mention the responsibility you need in your enquiry.
What you can check
This is a new service. We have not delivered this job for a client yet.
Other ways to get this done
- Your CRM may have a built-in duplicate manager. If it covers your match fields, use it and review its groups before merging.
- Admin-defined matching and merge rules are easier to audit than a rule nobody can read, whoever runs them.
Questions
Will you merge the records for us?
No. HubSpot cannot unmerge, so your admin merges the certain band after you have checked the sample.
Do you use AI to match?
Rules first. A language model may help only with ambiguous names, and it never decides a merge. Every group shows its evidence.
What about duplicates that keep coming back?
That is a source problem. The web-form job stops new ones from one path.
Send an enquiry
Send us
- The CRM name and plan, and roughly how many contacts and companies it holds
- Which fields you trust for matching, in a few sentences
- Who would approve the rule sheet and run the merges
Later, once you agree
- A read-only export with agreed columns, or a read-only user you can revoke
- The approved rule sheet
- A secure handoff route agreed before any record is shared
You keep the CRM and every credential. We work from an export or a read-only user you create and can revoke, and your admin does every merge. Which service reads your records is agreed in writing before any real record is shared; the entry test uses invented or redacted data, or your own provider account.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “crm-duplicate-pairs-reviewed-before-any-merge” as the subject.