Define the change you are trying to explain
Start with a concrete difference: a new keyword, altered budget, different location setting, or changed conversion goal. Record when it was first noticed and the campaign affected. A performance decline is a symptom, not yet an identified setting change. Keep the original report or screenshot privately so later adjustments do not erase the evidence. Avoid accusing a manager or automation before checking the account record.
Source checks: About change history · Manage auto-apply recommendations
Read the relevant history in context
Google's change history can be filtered by dates, change type, campaign, and ad group. Inspect the before-and-after values and the user or system information supplied. Automated and API entries may need explanation from the tool owner. A history label is a starting point for investigation, not proof of someone's intention. Check whether the entry represents a material live setting change or a system maintenance record.
Compare the record with the approved plan
Find the instruction, decision note, or experiment plan that was supposed to authorize the change. Confirm the campaign scope and timing. A broad edit may be intentional for one campaign and unintended for another. If no decision record exists, write the missing question and ask the responsible owner to resolve it. Do not circulate raw account emails or private identifiers in a public report; retain them only where the team needs the access evidence.
Inspect automatic subscriptions and connected tools
Review auto-apply recommendation subscriptions and their history, plus tools or agencies that can edit the account. Determine who owns each connection and which changes it is meant to make. Disabling a future subscription does not necessarily restore earlier settings. Document any necessary correction separately. Do not remove every integration merely because one edit was unexpected; identify the relevant source and protect the business's working measurement setup.
Correct the narrow unintended consequence
If the edit created an inaccurate offer, broken destination, or unaffordable spend exposure, address that specific issue promptly. Preserve a copy of the relevant settings before reversing them. Check whether other changes depend on the edit so rollback does not create a second problem. For performance interpretation, note that a date coincidence alone cannot prove causality: demand, auction conditions, and reporting changes may have occurred at the same time.
Make the handoff more accountable
Record the corrected setting, business reason, responsible person, and next outcome check. Establish which future edits need review and which routine tasks the manager can make within existing authority. Review qualified work and actual spend after the correction rather than treating a rollback as proof of recovery. The practical goal is a clear audit trail and manageable edit boundaries, so the owner can understand the next change without reconstructing the account from memory.
Worked example: illustrative figures
For example, an owner notices a budget rising from $20 to $30 on Monday and a new keyword appearing Tuesday. History identifies separate sources for the two edits. The budget change was authorized in a test note; the keyword offers a service the business does not provide. The owner keeps the planned budget test and removes the unsuitable keyword. The $10 difference is a 50% increase in the average daily budget setting, not a prediction that each day's billed cost increased by exactly $10.
Trace a change without a blanket rollback
Observed difference
Record campaign, setting, and first noticed date.
History evidence
Read old/new values and responsible source.
Approved plan
Separate intentional tests from unintended edits.
Narrow correction
Preserve useful settings and document the fix.
Illustrative budget and keyword changes. A setting change is not the same as settled billed spend.
Your practical checklist
- Name the specific setting or symptom being investigated.
- Filter history to the affected campaign and dates.
- Compare before-and-after values with the approved plan.
- Check relevant automation and tool ownership.
- Reverse only the unintended consequence after dependency review.
- Document the correction and future edit boundaries.
Your next action
Investigate one unexplained change and write a short before-and-after note identifying its scope, evidence, and next action.
Questions about this task
Does a system label prove automation damaged the campaign?
No. Read the actual changed values and ask the relevant tool owner when needed. Some records are maintenance or derived settings rather than harmful live edits.
Will disabling auto-apply undo previous changes?
Not necessarily. Review prior edits separately and correct any specific unintended setting through the supported account process.
Sources and editorial notes
Prepared with AI-assisted research. Official sources are linked below; worked examples and checklists are Getting Found Local recommendations, not measured client outcomes.
Sources checked October 13, 2026. Platform interfaces and local resources may change; follow the current instructions on the linked official page.
How we research, illustrate, and update these guides · Suggest a correction