Falaah Falaah AI

Duplicates, Merging and Data Quality

Find duplicate people and organizations, merge them field by field, and watch your directory's completeness score so reports and mailings stay accurate.

The same person often reaches you twice, once through a donation form and once through an event sign-up, and ends up as two records. This guide covers how Muin finds those pairs, how you merge them without losing history, and how to see whether your directory is in good shape.

When you’d use this

  • A donor appears twice in a report and the totals look wrong.
  • You imported a spreadsheet and want to clean up the overlap before a mailing.
  • Two organization records point at the same company, each holding half the invoices.
  • You want to know how many people in your directory have no email or phone before you plan a campaign.

Review duplicate people

  1. Go to Directory → Duplicates.
  2. Pick a Minimum confidence: All (20+), Medium (50+) or High only (80+). Start at High only and work down.
  3. Read each pair. Every row shows why Muin thinks they match.
  4. Choose Review & merge to open the merge screen, or Not a duplicate to dismiss the pair.

Muin scores a pair from the identifiers the two records share:

Signal Points
The same email address 50
The same email name on different domains (for example a.khan@ at two providers) 40
The same phone number 30
Exactly the same first and last name 20

A pair matched on name alone is never offered, it has to share at least one contact identifier too. The list returns up to 50 pairs at a time.

Duplicates is an owner and admin screen.

Merge two people

The merge screen opens with the record you came from as the one to keep.

  1. Check the two headers. The record badged Keep survives; the one badged Archive is retired. Use the ⇄ button to swap them if you picked the wrong way round.
  2. Work down Field comparison. The kept record’s value is used by default; click the other value to take it instead. Empty fields are filled in from the duplicate automatically. You can choose per field for: Email, Phone, Secondary Phone, Address Line 1 and 2, City, State, Postal Code, Country, Country Code, Preferred Language, Preferred Channel and Timezone.
  3. Read Affected records. This is the exact list of what moves across, gifts, pledges, cases, form submissions, registrations, conversations, invoices, receipts, documents, volunteer records and the rest.
  4. Read Unique constraint conflicts if it appears. Those duplicate rows are retired so the two records can become one.
  5. Click Merge persons, type MERGE to confirm, and confirm.

Everything attached to the duplicate is reassigned to the kept record, and the duplicate is archived so it stops appearing in searches, lists and reports. An audit entry records who merged what, which values were chosen, and a snapshot of both records as they were.

There is no un-merge. Separating two people again means recreating the record and re-attaching every row by hand. Swap the direction before you confirm rather than after.

When the merge finishes you land on a confirmation screen showing the surviving record’s id and the archived one’s, with links to review more duplicates or open the merged person.

Merge two organizations

  1. Go to Directory → Merge Organizations.
  2. Under Select Organizations, pick Primary (keep) and Duplicate (merge & archive).
  3. Merge Preview shows how many records are affected and breaks the count down by type.
  4. Under Field Conflicts, choose which value to keep for each field where the two disagree. Muin compares name, legal name, email, phone, website, tax id, industry, address line 1, city, state, postal code and country.
  5. Click Execute Merge, type MERGE, and confirm.

The duplicate’s contacts, conversations, documents, invoices and entity mentions move to the record you kept, and the duplicate is archived. As with people, there is no un-merge. After it completes you can choose Merge Another to do the next pair.

Check your directory’s data quality

Directory → Data Quality is a read-only report, not a duplicate list. It answers “is our directory good enough to run a campaign off?”.

Card What it tells you
Overall Quality Score The average of person completeness, organization completeness and entity-resolution coverage, out of 100, badged Good (80 and up), Needs Improvement (50 and up) or Critical
People How many have an email, a phone, an organization, plus a completeness percentage, and an Undeliverable email row when any address is bouncing
Organizations How many have an email and a phone, plus a completeness percentage
Entity Resolution How many document mentions have been matched to a real record, with a status breakdown and a link into Entity Resolution
Orphan Records People with no organization, and organizations with no people

Person completeness counts reachable email addresses only, so placeholder addresses do not inflate the score. A tenant with no people yet scores 100.

Data Quality is an owner and admin screen.

What it affects

  • Reports and giving totals. Two records split one donor’s history; merging restores a single lifetime total.
  • Audiences. A merged person is evaluated once, so a rule-based audience stops counting them twice.
  • Mailings. The archived record drops out of every recipient picker, which is usually the point.
  • The person’s timeline. Activities, messages, documents and submissions from both records appear on one timeline afterwards.
  • Entity resolution. Document mentions that pointed at the duplicate follow the surviving record.

When it goes wrong

What you see What to do
“Merge blocked” with a list of references A record type attached to this person has no merge rule yet, so Muin refuses rather than orphaning the rows. Nothing was changed. Send the blocked list to support; do not try to delete the records to get around it.
“Cannot merge a person with themselves” The two ids are the same. Go back to Duplicates and pick the pair again.
“The server was briefly unavailable and the merge didn’t go through. No changes were made” A transient failure, and nothing was written. Use Retry merge.
“Missing primary or duplicate person ID” You opened the merge screen directly instead of from a duplicate pair. Go back to Duplicates and choose Review & merge.
“Cannot merge an organization with itself” Pick a different record in one of the two organization pickers.
The same pair keeps reappearing in Duplicates Mark it Not a duplicate. Two people really can share a surname and a household phone.

Next step