Google Ads change history: what changed before the drop?
Find Google Ads change history under Campaigns, then check the dates, campaign scope and filters before treating a missing row as missing work. To investigate a performance drop, build a timeline of the relevant changes and test alternative explanations; an edit before a drop does not prove it caused the drop.
On this page8 sections
Where is Google Ads change history now?
Open Change history within the Campaigns menu, then select the period and scope you actually need. These labels were checked in Google Help on 19 Oct 2026.
Google’s account-history instructions describe the date selection at the upper right and the Views bar for choosing all campaigns, an individual campaign or a campaign type. On a smaller screen, the panel may need expanding with its arrow.
- Start broad enough to find the event. Record the intended advertiser account and dates. Our suggestion is to include time before the reported symptom, rather than searching only the day the client noticed it.
- Set the scope deliberately. Use All campaigns when establishing the scope, then narrow to the affected campaign or ad group.
- Read the filters. Google documents Campaign, Ad group, Change type, User, Tool, Item changed and Campaign experiment filters. A narrow filter answers a narrow question.
- Expand a relevant entry. Google’s instructions use the arrow next to a change for its details; record what changed instead of copying only the row count.
Google also documents selecting a campaign or ad group on its list page and choosing Change History on the blue menu bar. That opens a panel for the selected date range. Do not mistake this selected-object view for evidence covering every campaign.
Why is change history missing, empty or incomplete?
Check the view’s coverage before concluding there was no change. A wrong date range, selected campaign or restrictive filter can leave the event outside the question you asked.
| What to establish | What an empty result cannot prove | |
|---|---|---|
| Account and scope | You are looking at the intended advertiser and affected campaign or ad group. | That an event in a different scope never happened. |
| Dates and filters | The event falls within the selected dates and retained history. | That a narrowly filtered list is a full account log. |
| Display issue | Google suggests cache/cookie checks or a private browsing window when specific changes seem missing. | That refreshing can recreate an event Google does not list. |
| Coverage limit | Google says that sometimes not all account-level settings changes or changes by representatives during consultations are listed. | That no work occurred outside the returned entries. |
Those display and coverage qualifications come from Google’s history overview. Our reading: distinguish “not visible in this view” from “not recorded here”, and leave that distinction in the incident note.
The two Google pages are not identical on this point. The account-history instructions list account-settings changes, including billing information, payment methods and notifications, as a change type. The overview says not all account-level settings changes are always listed; the instructions add that password changes are not tracked.
Our reading: treat the type list as what can appear, not as a guarantee that every such event will.
Two further Google notes explain apparent gaps. The instructions say change columns will no longer be visible on the Campaigns and Ad groups pages; historical information is on Change history.
The overview says details shown for each entry are limited. To see all details, Google recommends smaller changes, ideally affecting fewer than 2,000 entities. That is a recommendation about change size, not a claim that every larger change disappears.
If the visible list still seems inconsistent, capture the account, dates, filters and a specific expected change for the responsible operator. A useful question is “Why is this known campaign edit absent from this view?”, rather than “Why is Google hiding our history?”
How far back can you look, read through a tool or undo?
The interface history, a programmatic change-event read and the interface’s Undo option have different boundaries. Keep the surface named beside each number.
| Documented scope | Boundary to retain | |
|---|---|---|
| Interface history | The past 2 years, according to Google Help. | Do not promise older account history from that interface. |
| Change-event API | The query date range must be within the past 30 days, according to Google’s API guide. | This is not a two-year archive, and returned events need not match every interface row. |
| Interface Undo | Google’s instructions say most types of changes from the last 30 days can be undone. | Most is not all. If a row contains several changes, Undo reverts them together; removed items and already-undone changes can prevent reversal. |
not all changes to your account will have a corresponding change event entry.
The API guide also limits an individual query to 10,000 rows and describes further queries for additional results. That is not a universal cap on the account’s total history. A change can take up to three minutes to appear in change-event results.
Coverage differs as well. The interface instructions say history also shows changes made through Google Ads Editor. The API guide says fetching Editor changes with the change-event report is not supported
. Google’s client-type reference calls the Editor value a placeholder
that the API does not return.
For those edits the guide points to a different resource, change status, which this article does not cover. These are different surfaces, not interchangeable promises.
Our reading: ask a tool which window, filters and coverage it actually returned. A request that succeeded technically can still provide an incomplete answer to the business question. No missing history should be reconstructed from guesses.
What changed before the drop, and what would prove it matters?
Build a small timeline that separates observed changes from possible causes. Google’s history overview links changes with performance data; our proposed investigation adds the question each observation still leaves open.
Consider an invented local repair campaign. The owner reports fewer suitable enquiries after several changes. This is a made-up investigation, with no client data or claimed performance result:
| Invented observation | Next evidence check | |
|---|---|---|
| Monday · budget edit | A returned entry shows a budget change on the affected campaign. | Confirm old/new values and the current setting. Check whether actual delivery changed, rather than assuming the edit alone reduced enquiries. |
| Tuesday · website release | The site owner reports a form change; this is an external note, not an account-history row. | Repeat the enquiry path and inspect tracking evidence. A missing form event can make reported conversions fall without proving fewer enquiries. |
| Wednesday · reported drop | The owner sees a weaker recent period in a campaign report. | Keep dates, outcome definition and reporting basis comparable. Separate a reporting gap from a business-outcome change. |
| Still unknown | The timing fits more than one explanation. | Ask for form-test evidence, the relevant campaign settings and comparable outcome data before choosing an intervention. |
The important finding is not “Monday’s budget edit caused the loss”. It is that a budget edit and a website release both need checking, and the available evidence does not yet separate them.
For a suspected website measurement failure, use the conversion-tracking investigation. For a delivery stop, use the zero-impression checks. A timeline locates the next question; it does not replace either diagnosis.
Who made the change is not the same as why
Google describes an email in the User column for interface changes, possible system labels for automated/internal activity, and possible API or tool names for API changes. Preserve the label actually returned rather than assigning an unnamed human behind it.
The API guide provides old/new values for updates, new values for creations and user information when it is visible in the interface history. A blank previous value on a newly created item is not itself proof of a broken history reader.
Our suggestion: add the decision owner’s reason and expected outcome to a separate note. The event records a setting change; it does not establish the owner’s intent, the quality of the decision or the amount of work someone performed.
What can Variafy read from the history?
Variafy offers read-only Google Ads access after your beta invitation. It reads recent change events over a chosen 1–30-day period in the account’s time zone.
The returned evidence includes available time, resource, operation, originating client, user email and changed fields with old/new values. It retains coverage limits, and displayed values or fields can be shortened. This is not a full historical archive, a guarantee of every interface row or a rollback function.
It does not read edits made in Google Ads Editor, which Google’s API does not return through this report.
Our suggested reading request is: “For this account and period, list the returned changes around the reported symptom. Preserve the author/client labels, old/new values and coverage limits. Build a timeline with alternative explanations and missing evidence; propose no changes.”
The person responsible for the account must supply any missing evidence and decide on an intervention. Variafy does not change Google Ads or recover inaccessible history.
What should you do with the timeline?
Turn the timeline into a specific check or an explicitly approved intervention. Our suggestion is to keep evidence, explanation and the next owner together so the client can understand what is known. For setting a re-check after a bid-strategy change, see whether to wait, check or repair a Learning bid strategy.
Our suggested incident note
“Symptom and reporting basis: […]. Relevant returned changes: [time, object, old/new value, reported user/client]. Other observed events: […]. Possible explanations: […]. Missing evidence: […]. Next check and owner: […]. Decision checkpoint: […]. No causal conclusion or reversal is approved by this note alone.”
Do not undo a whole row merely because it predates a weaker week. The interface instructions make clear that grouped changes reverse together, and not every change is reversible. Establish what should be restored and who can authorize it before an operator acts.
If the useful finding is that nothing in the returned scope explains the symptom, say so. That is a legitimate result. Broaden the investigation with the Google Ads audit or check the approved budget brief, without treating more edits as proof of better management.
How we checked
Checked on 19 Oct 2026 against 4 public Google documents: the Google Ads Help overview of change history; the Help instructions for finding, filtering and undoing account changes; the Google Ads API Change Event guide, last updated 14 Oct 2026; and its version 26 client-type reference. Interface, API and Undo boundaries are kept separate, including Editor coverage.
The incident timeline and decision note are our suggestions; all events in the repair-service example are invented. Variafy is our product.
Questions
Where do I find change history in Google Ads?
Open Change history within the Campaigns menu, then set dates and the intended scope. Google also documents selecting a campaign or ad group and opening Change History from the blue menu bar. Check filters before treating a missing row as missing activity.
How far back does Google Ads change history go?
Google’s interface Help describes the past 2 years. Its change-event API guide restricts queries to dates within the past 30 days. A tool’s history window and coverage must be stated separately from the interface’s retention.
Why is Google Ads change history not showing a change?
Check account, selected campaign, dates and filters, then investigate a possible display issue. Google also documents missing categories of changes and different API coverage. An empty result does not prove no edit or work occurred.
Can I undo any change in Google Ads?
No. Google describes Undo for most change types made in the last 30 days, with exceptions. Multiple changes in one row reverse together. Inspect the actual option and affected settings before an authorized operator acts.
Does a change before a performance drop prove the cause?
No. It identifies a hypothesis to test. Keep the reported author and old/new values, comparable outcome data, website events and other explanations in the same timeline before deciding what caused the drop or what to change.