How to Mass Update Salesforce Records Safely

A safe procedure for mass updating Salesforce records: scope with SOQL, snapshot before, batch the import, and verify after - with the automation traps to check first.

Updated 2026-09-09

A mass update is the one admin task where a small mistake scales instantly. The safeguards are procedural, not technical, and they take about ten extra minutes.

This is the checklist worth following every time, whichever tool you use.

1. Scope it with a query first

Never build the target list by hand. Write the SOQL, count it, then export exactly that set - the same filter that defines the update should define the snapshot.

Count before you touch
SELECT COUNT()
FROM Account
WHERE Industry = NULL AND CreatedDate = LAST_N_DAYS:90

2. Snapshot the current values

  • Export Id plus every field you are about to change - that file is your rollback.
  • Keep it until the change has survived a full business cycle, not just until the import succeeds.

3. Check what the update will trigger

  • Validation rules that will reject rows mid-batch.
  • Flows, triggers, and workflow rules that fire on update - a 50,000-row update can cascade well beyond the object you targeted.
  • Roll-up summaries and sharing recalculations on parent records.
  • Email alerts - disable them for the run if the change is administrative.

4. Batch it, starting small

  • Run 10 records first and inspect them in the UI.
  • Then a few hundred, then the rest.
  • Use the Bulk API for large volumes, and keep the success and error files from every batch.

5. Verify after

Re-run the original count query. It should return zero (or the expected remainder). Spot-check records the automation touched indirectly, not just the ones you updated.

Frequently asked questions

How do I undo a bad mass update in Salesforce?
There is no undo. You re-import the pre-change snapshot keyed on record Id, which is why exporting it first is non-negotiable.
Should I disable automation before a mass update?
Sometimes - and only deliberately. Deactivating a validation rule or flow for a run is legitimate; forgetting to reactivate it is the failure mode. Note it in the change ticket.
What is the safest batch size?
Start at 10 for verification, then 200 for API-mode work. Large Bulk API batches are efficient but make failures harder to isolate when triggers are involved.

More from the Admin How-To

Related reading