How an Agency Resolves a Who Changed the Campaign Dispute With a Client
Alessandro Conti
シニア・パフォーマンスマーケター
The message every agency dreads arrived at 9 a.m.: "Our CPA doubled overnight — what did your team do?" The client was certain the agency had broken something. The account lead was certain it had not. The difference between those two certainties, and whether the relationship survived the next hour, came down to one thing: whether the agency had a real agency campaign change accountability workflow, or just a defensive theory. This guide walks through that workflow end to end, using a single dispute as the worked example.
Quick answer: To resolve a "who changed the campaign?" dispute, you need detection and proof. Change alerts surface meaningful edits as they happen; an attributed action history lets you reconstruct which change ran, who made it, and when. Filter to the client account, narrow the window, read the entry — and reply with a timestamped record instead of a guess.
This is a composite scenario, but the failure mode is real for any agency running multiple buyers across multiple client accounts. The accountability workflow below is the fix, broken into the steps you can adopt before your own 9 a.m. message arrives.
Why these disputes are unwinnable without a workflow
Trace the dispute through a typical agency that has no accountability workflow. The client claims the CPA spiked overnight. The lead pings the team: did anyone touch the account yesterday? Two buyers say no, one is unsure, one is out. Someone opens the native change history and finds edits — but they are all stamped with the same shared owner login, so there is no actor. By the time anyone replies, it is the afternoon, the answer is a hedge, and the client has decided the agency does not know what is happening on its own account. Worse, the harmful change — whatever it was — has been running into the client's budget for another full day.
That is the structural trap. Without a workflow, the agency cannot even establish whether it caused the spike, let alone which buyer did what. Two failures stack: no detection, so the change ran unnoticed; and no attribution, so the change cannot be traced. The accountability workflow fixes both, in that order.
The stakes are not abstract. Forrester reported in 2024 that B2B buyers cite a lack of transparency and trust as a leading reason for switching providers, and HubSpot found in 2024 that acquiring a new customer costs five to seven times more than retaining an existing one. An unanswered "who changed this?" dispute pushes directly on both numbers — it reads to the client as opacity, and opacity is what loses the renewal.
A change dispute is not won by being right — it is won by being able to prove it. Without detection you do not know a harmful change ran; without attribution you cannot say who ran it. An agency that answers "which change, by whom, at what time" in two minutes does not just resolve the dispute — it demonstrates the control that keeps the account.
The prerequisite: named seats, not shared logins
Before any alert or log is useful, the agency needs attribution, and attribution is impossible on shared logins. This is the part teams skip and then wonder why their audit trail is worthless.
When four buyers share one login on a client account, every change the platform records is stamped with that single identity. The native history is technically complete and operationally useless — it knows a bid changed, not who changed it. So the first step of the workflow is not a tool, it is a policy: every buyer gets a named seat with a scoped role, and the shared logins are retired. We make the full case for this in why shared logins are killing your ad agency — but the short version is that the entire accountability workflow rests on this foundation.
Named seats are not a nicety; they are the load-bearing wall of accountability. Every alert you wire and every log entry you read is only as useful as the attribution underneath it. On a shared login, your history shows a change happened but never who made it — exactly the question a dispute hinges on. Fix the seats first.
With named seats in place, every change made through the operating layer carries a buyer's name and a timestamp, across Meta, Google, and TikTok. Now the workflow has something to detect and something to prove.
Step one: wire change alerts so you see edits as they happen
Detection comes first because the cheapest dispute to resolve is the one you catch before the client does. The agency wired change alerts to a Telegram channel the leads watched, so meaningful edits surfaced near-prompt instead of in next week's report.
The alerts were not noise. The team scoped them to the changes that actually move CPA: budget jumps above a threshold, bid-strategy switches, campaign pauses and resumes, and large creative swaps. A junior raising a daily budget 5x, or switching a campaign from cost-cap to lowest-cost, fired an alert the lead saw within minutes of the next sync. This is the same alert discipline we lay out for agencies managing alerts across many client accounts: alert on the changes that cost money, not on every routine edit.
Change alerts move the dispute earlier in time, where it is cheap to resolve. Catching a junior's 5x budget jump the same morning means a thirty-second conversation and a quick revert. Discovering it in next week's client report means a doubled CPA, a week of wasted spend, and a client who found the problem first. The alert is a head start, not surveillance.
One honest caveat the team set expectations around internally: the sync runs about every 15 minutes through official APIs, not live. So an alert is near-prompt, not instantaneous. In practice that window is irrelevant to the outcome — the difference that matters is catching a harmful change in minutes-to-an-hour versus discovering it a week later, and a roughly fifteen-minute sync sits firmly on the right side of that line. The mechanics of routing alerts to the right channel are covered in our Telegram alert setup guide.
Step two: when the dispute lands, open the action history
Now the worked example. The 9 a.m. message arrives. With the workflow in place, the lead does not start a panicked group chat. She opens the action history, filters to that client's account, narrows to the past twenty-four hours, and sorts by time.
The list is short and ordered. There it is, attributed and timestamped: at 6:40 p.m. the previous evening, a junior buyer switched the campaign's bid strategy from cost-cap to lowest-cost-without-cap, chasing volume. The change ran overnight, the bid floor came off, and CPA climbed exactly as the client described. The investigation took under two minutes and followed the same steps as our how to investigate ad account changes guide: filter to the account, narrow the window, read the attributed entry, decide.
The action history is the proof half of the workflow. Detection tells you something changed; the history tells you precisely what, by whom, and when. The dispute that felt like an accusation becomes a lookup with a definite answer. Two minutes after the client's message, the lead knows more about the overnight account activity than the client does.
What the history showed that the client's report never could
The client's own dashboard showed the symptom — doubled CPA — but none of the cause. It could not show that the spike began at 6:40 p.m., that it coincided with a single bid-strategy edit, or that one named buyer made it. Those facts only existed in the attributed action history, and they were the difference between "we think something happened" and "here is exactly what happened."
This is the recurring lesson in the moments an audit log saves you: the report tells the client what the numbers did, while the log tells the agency what its team did to produce them. In a dispute, only the second one resolves anything.
Step three: decide and act — a human reverts, nothing is automatic
Knowing the cause is not the same as fixing it, and the workflow is careful about who does the fixing. The alerts notified and the history recorded, but neither one reverts anything. The decision to switch the bid strategy back belonged to the account lead, who reviewed the change, confirmed it was the cause, and approved the revert herself.
That distinction matters for client trust as much as for safety. An agency that says "our system auto-reverted the change" invites the obvious next question — what else is your system doing without a human looking? An agency that says "we identified the change, confirmed it caused the spike, and reverted it deliberately this morning" demonstrates judgment. The workflow gives the team detection and proof; the corrective action stays a human decision, approved before it runs.
Detection and proof are automatable; judgment is not. The workflow tells you a harmful change ran and who made it, but a person decides whether to revert it, when, and what to do instead. That human-in-the-loop step is not a limitation — it is what lets the agency stand behind every corrective action, because a person chose it.
Step four: close the dispute with a record, not a defense
The reply went out within the hour, and it was not a hedge. "Last night at 6:40 p.m. one of our buyers tested a new bid strategy on your top campaign to chase volume. It removed the cost cap and pushed CPA higher than we expected. We caught it this morning, confirmed it was the cause, and reverted it as of 9:20 a.m. — costs should normalize over the next 24 to 48 hours. Here is what we will do instead, and here is the change we have made to prevent an untested bid switch on a live campaign again."
That message did three things a defensive guess never could. It established the agency as more informed about the account than the client. It owned the specific mistake instead of denying a vague one. And it closed with a process change, drawn straight from the incident. The client did not churn; the responsiveness became a reason they later cited for renewing. The same trust dynamic plays out at scale in an agency that turned its audit trail into a retention tool.
The dispute is not closed by proving you were blameless — the junior did make a bad change. It is closed by proving you are in control: you detected it, attributed it, reverted it deliberately, and changed the process so it cannot recur. Clients do not expect a perfect agency — they expect one that knows exactly what happened and fixes it on purpose.
Step five: turn the incident into a standing rule
The best agencies do not stop at resolving the dispute. They convert it into a guardrail so the same class of change never causes a surprise again. After this incident, the team added two standing rules to its workflow.
First, untested bid-strategy switches on live campaigns above a spend threshold now require a second buyer's sign-off, and the change alert for bid-strategy edits is wired to the lead specifically. Second, every meaningful change to a client account is reviewed against the action history weekly, so attributed changes become routine team hygiene rather than something only examined during a crisis. The junior who made the change was not punished — the point was never blame. The point was that the change had been invisible, and now it was not.
This is how the accountability workflow compounds. Each dispute it resolves also teaches the team which changes need a tighter gate, and the alerts and history make those gates enforceable. Over time the dramatic 9 a.m. messages get rarer, not because mistakes stop, but because the harmful ones get caught and corrected before the client ever sees them.
What the workflow does — and what it deliberately does not
To be precise about the boundaries, because they matter for how you sell this internally and to clients. The workflow's two tools — change alerts and the action history — detect and prove. They notify you when a meaningful change happens, and they let you reconstruct who made which change and when. That is their entire job, and they do it across the platforms a client runs on.
What they do not do is act. The alerts do not pause campaigns, the history does not revert edits, and neither one optimizes or relaunches anything on its own. Every corrective action is a human decision, approved before it runs. The sync is about every 15 minutes through official APIs, not live, so detection is near-prompt rather than instantaneous. And platform depth varies — Meta carries the fullest change coverage, with Google and TikTok covered for the core flows. None of those boundaries weaken the workflow, because the workflow's value is exactly in giving humans fast, attributed information to act on — not in acting for them.
The pattern generalizes beyond a single dispute. Any agency with multiple buyers on multiple client accounts faces the same risk surface: changes nobody noticed, made by people nobody can identify, discovered after the damage. The accountability workflow — named seats, scoped change alerts, an attributed action history, human-approved corrections, and standing rules distilled from incidents — is the standing answer, whether the trigger is a client dispute, an offboarding, or a quiet weekly review.
The takeaway for any agency that manages client accounts
The "who changed the campaign?" dispute is unwinnable on shared logins and a defensive theory, and it is routine with an accountability workflow. The difference is not luck or a better story — it is detection and proof wired in before the dispute arrives. Named seats so changes carry a name. Scoped alerts so harmful edits surface the same day. An attributed action history so any change can be reconstructed in two minutes. And a firm rule that a human, not a system, decides every correction.
To see how change alerts and an attributed action history resolve your next client dispute across Meta, Google, and TikTok — with named seats, sync about every 15 minutes through official APIs, and every corrective action left to your team — start a 14-day Wevion trial alongside the permanent free plan and wire the workflow before your next 9 a.m. message. If you are weighing rule-and-alert platforms in parallel, see how Wevion compares to Revealbot.
This guide is part of our agency tools hub — explore the full cluster for related playbooks.
よくあるご質問
The Ad Signal
推測を拒否するメディアバイヤーのための週刊インサイト。1通のメール。シグナルのみ。
関連記事
Who Changed the Campaign? Why Your Ad Accounts Need a Real Audit Log
A budget triples overnight. A winning campaign goes dark. Nobody on the team will admit to the change, and the native platforms only show a fraction of the story. Here is why a unified audit log across every ad account turns finger-pointing into a two-minute lookup.
How to Investigate an Unexplained Ad Account Change Using Action History
When a metric moves and nobody admits to the change, you do not need a meeting — you need a method. This is the exact step-by-step for tracing any unexplained ad account change using a unified action history, from filtering to fix to a weekly review habit.
How One Agency Turned an Audit Trail Into Its Best Client-Retention Tool
An eleven-client agency was bleeding accounts not to bad performance but to bad answers. When a client asked what changed, 'we think someone adjusted the bid' was not survivable. Here is how a unified audit trail across every client account turned the most dangerous question into the easiest one.