Skip to main content

Finding duplicates

Open Data Quality → Dedupe Console in the sidebar. Detection is configured in the right-hand Detection setup panel, in three numbered steps.

0. Pick a snapshot​

First, choose the backup snapshot to scan — the console asks for a backup and a run, and remembers your last choice. Detection runs on prepared snapshot data, so only prepared objects are selectable; the panel links you to Prepare this snapshot if nothing is ready yet. See Preparing snapshots.

1. Object​

Pick the object to scan (each shows its record count). System history/feed/share objects are never offered.

2. Matching fields​

Choose up to 8 fields that define "the same record". Two records are duplicates when every field you add matches. Each field row has:

  • a comparison type:
    • Exact — character-for-character.
    • Ignore case — "JOHN" matches "john".
    • Ignore case & spaces — "John Smith" matches "johnsmith".
  • advanced options per field:
    • Cross-field — match the value in any of several fields (e.g. Phone or MobilePhone).
    • Fuzzy (phonetic) — "Smith" ≈ "Smyth".
    • Transposed words — "John Smith" ≈ "Smith John".
    • Ignore punctuation — "A.C.M.E, Inc." ≈ "acmeinc".
    • Match blank values — treat two empty values as a match (off by default, so blanks don't cluster).

[SCREENSHOT REQUIRED: Detection setup panel with matching fields and advanced options open]

3. Record filter (optional)​

Narrow which records are scanned with a SQL condition over the snapshot — e.g. CreatedDate >= '2024-01-01' AND Country = 'France'.

Run it​

Click Find duplicates. The scan is read-only and costs zero Salesforce API calls. A first scan on a cold snapshot downloads and indexes the data and can take a few minutes; once prepared, scans usually return in seconds.

Reading the results​

Matches appear as duplicate group cards: each group's records in a table, a checkbox to include the group in a merge, and a crown marking the record that will be kept. Each row links to the record in Salesforce.

  • Use Columns to add fields to the table — fields not in the result are fetched with a quick re-scan while the table stays up.
  • Export downloads the scan's full result as CSV (the on-screen view shows a sampled preview; the CSV is the authoritative list, capped at 5,000 rows).
  • Found nothing? Loosen the comparison — try Ignore case & spaces or a fuzzy option — and run again.

[SCREENSHOT REQUIRED: Duplicate group cards with keep crowns and selection checkboxes]

Next: decide who wins each merge — see Winner rules and field rules.