Google Ads Change History: Find What Changed Before Results Shifted

An office-cleaning business raises its Google Ads budget. By Wednesday, Monday–Tuesday’s $500 of spend shows five quote-request conversion credits: $100 each. The owner wants to reverse the increase. This fictional incident has another change to investigate: a website release that stopped the mobile quote form working.
Google Ads change history helps us establish what happened inside the account. To choose the right response, we also need website records, working-form checks and a report that allows for late conversions. Here is how we separate the fault to fix now from the budget question that still needs evidence.

Pin down which result changed
Write an observation someone else can verify: “This campaign reports five quote-request credits for Monday–Tuesday, as exported at 08:30 on Wednesday.” Save the report with the campaign, conversion action, dates and account time zone. A later export may contain more conversions.
Then locate the gap. Fewer impressions point to a different investigation from a working ad that sends people to a broken form. Are quote requests absent from Google Ads, the business’s inbox, or both? Our guide to connecting Google Ads inquiries to your lead records explains how to reconcile those stages.
Use the same outcome definition throughout. Conversion credits, received inquiries and qualified opportunities are different measures; switching between them can create a decline that the underlying records do not support.
Read the old and new settings in change history
Open Campaigns > Change history in Google Ads. Set dates that include the period before the first symptom and the affected days. Start with the relevant campaign, then widen the scope if a shared setting may be involved.
Expand candidate entries and record the old value, new value, timestamp and source. User, tool, change-type and item filters help identify manual edits, automated rules and third-party tools. Google’s change-history instructions explain these controls and Undo.
For each candidate, ask whether its timing and scope fit the symptom. A location exclusion is worth examining when delivery changes in that area. An ad pause in one group needs a breakdown of that group, not just the account average. An edit preceding a decline makes it a candidate; it does not establish the cause.
Add the changes outside the account
Website releases, form changes, outages, offer changes and sales-system interruptions belong on the same timeline. Get dated records from the person responsible for each system. “We did not touch the ads” leaves those possibilities open.
Google’s description of change history says it covers the previous two years and notes that some account-level or representative-made changes may be absent. A website deployment needs its own record.
Convert timestamps to the account time zone, retaining the originals when systems use different zones. Two entries marked 09:00 may describe events hours apart. That distinction can determine whether a proposed cause happened before the first failure.
Follow the fictional incident in order
The company, amounts, checks and outcomes here are invented, not DIGITALTHROB client results or benchmarks. All times below use the Google Ads account time zone.
Monday, 09:10 — budget increased. Change history records the manager changing the average daily budget from $200 to $250. The old and new values are confirmed; the effect on inquiry quality remains unknown.
Tuesday, 11:20 — form released. A website release makes the property-type field required. The release record identifies the version and time; on mobile, the field has no usable selection.
Tuesday, 11:24 onward — validation errors recorded. Website logs show errors after the release. Desktop submissions still reach the inbox, narrowing the technical investigation to the mobile path.
Wednesday, 08:30 — fault reproduced. The manager exports the five-credit report. A fresh mobile test cannot submit; a desktop test succeeds. The tester labels both attempts so they are not mistaken for genuine inquiries.
We now have a reproducible fault with a matching release time and device pattern. That is stronger evidence for a form correction than the budget edit alone. It does not reveal how many genuine inquiries were lost: a failed attempt is not automatically a valid lead.
Read the same spending period twice
The manager’s first comparison also mixes reporting maturity. The older Monday–Tuesday period has had longer to collect conversions than the recent one. These three records show why that matters:
Earlier Monday–Tuesday, mature report: $400 spent and 10 quote-request credits. Cost per conversion: $400 ÷ 10 = $40.
Recent Monday–Tuesday, Wednesday’s export: $500 spent and five credits. Cost per conversion: $500 ÷ 5 = $100.
The same recent period, exported one week later: spend remains $500, but nine credits are now reported. Cost per conversion: $500 ÷ 9 = about $55.56.
The later export adds four credits to the same click-date period; it does not add another $500 of spending. Those credits are not a count of distinct customers.
Google’s conversion-lag guidance explains why recent acquisition costs can look higher while conversions are still arriving. Here, the later export reduces the apparent cost from $100 to $55.56. That remains above the older $40 figure, but neither comparison isolates the budget’s effect: the form fault, traffic mix and small outcome count still complicate the result.
Keep the measurement definition stable, too. A newly counted action changes the denominator. Our guide to primary and secondary Google Ads conversions covers that configuration question.
Correct the proven fault and hold the unresolved setting
In this fictional incident, the manager pauses the affected quote campaign at 08:35 while the website owner restores the working form configuration. At 09:15, mobile and desktop retests confirm submission, receipt in the inbox and the expected completion event. The technical fault is corrected. These checks do not establish ad attribution for every future inquiry.
After the repair passes, the manager resumes the campaign at the existing $250 average daily budget, within the remaining approved allowance. There is no further increase. The budget decision stays open until comparable post-repair results are available; reverting it during the repair would introduce another change without resolving that question.
If the timeline instead points to a bid-strategy change, inspect its status and conversion cycle. Google explains that automated strategies may need time to recalibrate. That can inform the performance review; it is not a reason to keep sending people to a confirmed broken destination.
Before using Undo, inspect the whole entry. Multiple changes in one row may be reversed together, and not every change is eligible. Save the current state, choose the specific correction and verify the resulting settings.
Close the repair without closing the budget question
The incident record now has three clear decisions:
Repair completed: the required-field fault is linked to Tuesday’s release. The website owner’s Wednesday repair has passed mobile and desktop retests.
Budget held: the manager has resumed at $250, with no additional increase. The initial $100 reported cost is not being treated as a final result.
Outcome review assigned: the campaign manager will assess a fixed post-repair period after allowing for the account’s observed conversion delay. The review will include inquiry quality; spend and renewed submission failures need monitoring immediately.
Carry these decisions into your monthly advertising review. Anyone reopening the issue can see what was corrected, which setting was held and what evidence is still needed.
Through our Google Ads management service, we can help investigate campaign changes and their effect on the customer journey. Bring your change history, reporting dates and unresolved performance questions to a strategy call.
Comments