The Duplicate That Wasn't
Your database has 120,000 leads. A data quality audit finds 9,000 suspected duplicates by email address. You plan to merge them using Marketo's merge leads feature. Before merging, you notice that about 1,500 of the duplicate pairs have different values in Lead Score, Lead Owner, and Original Source. In 400 of those pairs, the 'newer' record (created 6 months later) has the higher score and the correct current owner, but the older record has the accurate Original Source. You also have a Salesforce sync running. A colleague suggests just deleting the older records to 'clean up.' Another suggests merging all 9,000 in one batch overnight.
1. Evaluate the two colleague suggestions and explain the risks of each.
Read the worked answer
Suggestion 1 - delete the older records: Dangerous. Deleting a lead in Marketo can propagate a delete to Salesforce (depending on sync settings and whether the record is synced), destroying CRM history and any related opportunities. It also discards the accurate Original Source on the older record and loses activity history attached to it. Deletion is irreversible in most cases and is the wrong tool for deduplication.
Suggestion 2 - merge all 9,000 in one overnight batch: Risky for three reasons:
- Merge is not reversible. Marketo merge keeps the 'winning' record and folds the loser's activity into it, but field values on the winner are not automatically the 'best' values - you must control which record wins.
- Bulk merge at scale can hit API/processing limits and create sync churn with Salesforce, generating thousands of updates and possible lock contention.
- No field-level reconciliation means you may keep the wrong owner or lose the accurate source.
Better path: merge in controlled batches with a defined 'winning record' rule and pre-merge field consolidation.
2. Design the merge approach for the 1,500 conflict pairs, specifying how to handle score, owner, and original source.
Read the worked answer
Approach: consolidate fields first, then merge with a deterministic winner rule.
Decide the winner rule. For each pair, the record with the higher score and current owner should win (the newer record in the 400 flagged pairs). Winner = record with most recent meaningful activity, or explicitly the one with correct owner.
Pre-merge field consolidation. Before merging, copy the accurate Original Source from the older record onto the winning record using a batch smart campaign or list import. Do this in a staging field so you can verify. Do the same for any other fields where the loser holds the truth.
Score handling. Merging sums or keeps? Marketo merge behavior: the surviving record's field values are retained; activity is combined. If you want the higher score preserved, ensure the winner already holds it (it does in the 400 pairs). For pairs where the loser has the higher score, set the winner's score to the max before merging.
Owner handling. Confirm the correct owner is on the winner. If ownership should follow the account/territory, verify against Salesforce before merge.
Batch size. Merge in batches (e.g., 500–1,000 at a time) during low-sync hours, and monitor the sync log for lock errors.
Post-merge validation. Re-run the duplicate report to confirm reduction, and spot-check 20 merged records for correct source, owner, and score.