An assistant notices that spend is up while conversions are down. A message saying “campaign performance looks bad” creates alarm but gives the account owner no reliable basis for deciding whether to wait, investigate, or intervene. In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
The operational question is not simply whether the assistant can complete the task. It is how the team can separate observation from diagnosis and place the anomaly in a usable comparison frame, preserve the manager’s decision rights, and leave evidence another person can review. A useful handoff defines the object, the permitted action, the stopping condition, and the proof of completion before routine work begins. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Start with the customer decision
Write one sentence explaining what the business is trying to accomplish and what must remain under owner control. That sentence prevents a tool setting or spreadsheet column from becoming the goal by accident. In this workflow, the central decision is to separate observation from diagnosis and place the anomaly in a usable comparison frame. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
List the systems and records involved: account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline. Give each object a plain-language purpose. If two people use the same label for different things, resolve that difference now. Ambiguous definitions become inconsistent actions when the work is repeated across dozens or hundreds of records. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Use a concrete example from the team’s normal workload. Walk it from intake through completion, including the handoff back to the manager. Then choose a second example that should be rejected or escalated. Positive and negative examples expose missing rules much faster than a broad instruction such as “use your best judgment.” In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Define authority before access
Access answers what the platform technically permits. Authority answers what the business has approved. They are not the same. A platform role may include actions outside the assistant’s assignment, while a narrow task can sometimes be completed through a manager-owned approval step. Document both boundaries. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
For each action, label the assistant as observe, prepare, execute, or escalate. “Observe” means gathering facts without changing state. “Prepare” means drafting a change for review. “Execute” means the action is pre-authorized within stated limits. “Escalate” means the owner must decide. Add an expiry or review date so yesterday’s permission is not treated as permanent. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
The assistant should not pause campaigns or change budgets unless the operating agreement grants that authority for the stated condition. Suspected tracking failure, policy action, billing trouble, or account compromise takes the escalation path. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Build the working record
A practical record should be small enough to maintain during real work. Capture the request, relevant object identifiers, current state, intended action, owner, due time, and evidence link. Add a notes field for exceptions, but do not hide required decisions in free text. Use controlled values for status and outcome wherever the same choices recur. In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
The completion evidence for this process is screenshots or exports with timestamps, change-history references, a reproducible comparison, and a clear statement of what has and has not been checked. Store it with the work item rather than in a private message. The goal is not surveillance; it is continuity. If the usual manager is unavailable, another authorized person should be able to understand what happened without reconstructing the task from memory. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Avoid collecting data merely because the platform exposes it. Record only what supports the decision, quality check, or audit trail. Sensitive information should stay in its approved system, with the handoff linking to the record instead of copying it into an uncontrolled sheet. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Run the workflow in four passes
- Validate the metric and date range. Record the result where the work owner and reviewer can inspect it. A completed checkbox without the underlying object, date, and reviewer is not sufficient evidence.
- Localize the change by campaign, device, geography, or creative. Record the result where the work owner and reviewer can inspect it. A completed checkbox without the underlying object, date, and reviewer is not sufficient evidence.
- Review tracking and recent account changes. Record the result where the work owner and reviewer can inspect it. A completed checkbox without the underlying object, date, and reviewer is not sufficient evidence.
- Offer bounded next actions with owners and deadlines. Record the result where the work owner and reviewer can inspect it. A completed checkbox without the underlying object, date, and reviewer is not sufficient evidence. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
The first pass establishes scope. The second protects data and permissions. The third proves the action on a limited example. The fourth closes the loop. Keeping those passes separate makes failures easier to diagnose. A scope problem needs a different owner from a platform outage or a simple data-entry error. In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
For a new process, run a small pilot before increasing volume. Select examples that reflect common work and known edge cases. Record not only whether the assistant finished but also how often the rule required clarification, how long review took, and which evidence was missing. Revise the instructions from those observed gaps. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Review quality without redoing every task
Quality assurance should focus on the highest-risk decisions and a representative sample of routine work. Review every irreversible, customer-facing, permission-changing, or compliance-sensitive action. For lower-risk repetitive work, sample across different days, sources, and exception types rather than checking only the easiest records. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Use failure categories that lead to action: wrong object, wrong rule, incomplete evidence, unauthorized action, missed escalation, or platform error. A single “incorrect” label does not tell the owner whether to clarify the brief, change access, improve training, or repair a system. Track rework separately from new work so volume does not disguise recurring defects. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
When the same exception appears several times, update the operating rule. Do not rely on the assistant remembering a conversation from last month. Put the decision beside the workflow, add an example, and identify who approved the change. This is how a small team turns daily corrections into a dependable operating system. In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Measure the handoff, not vanity activity
Useful measures include first-pass acceptance, exception rate, time to decision, overdue escalations, rework by cause, and the percentage of completed items with inspectable evidence. Raw activity counts can show workload, but they do not show whether the work was correct or helped a decision. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Set a review cadence that matches risk. A new or changing workflow may need weekly review; a stable routine may move to monthly sampling. Review access whenever the scope changes and remove it promptly when the assignment ends. Keep a short change log so performance shifts can be interpreted in the context of new rules or tools. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
A healthy workflow makes uncertainty visible early. The assistant should be rewarded for stopping at the boundary, not encouraged to conceal ambiguity to maintain throughput. That protects the customer, the account, and the usefulness of the underlying data. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Turn the process into a reusable SOP
The final SOP needs five parts: purpose, scope, ordered steps, escalation rules, and completion evidence. Attach one accepted example and one rejected example. Name the process owner and the next review date. Test the SOP by asking a second person to follow it without oral coaching. Their questions reveal what the document still assumes. In this specific handoff, use “validate the metric and date range” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
A paid ads operations assistant can maintain the routine, collect evidence, and surface exceptions once those decision rights are clear. Pair this procedure with the paid ads anomaly review so the task fits the site’s broader growth-operations system rather than becoming an isolated checklist. In this specific handoff, use “localize the change by campaign, device, geography, or creative” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
The handoff is ready when the assistant can explain what to do, what not to do, when to stop, and what proof closes the task. That is a stronger standard than “access granted” or “training complete,” and it gives the manager a practical basis for delegation. In this specific handoff, use “review tracking and recent account changes” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
Source note
Google Ads change history records many account changes and identifies who made them. It is useful evidence for investigation, but a ticket should still distinguish a correlated change from a proven cause. In this specific handoff, use “offer bounded next actions with owners and deadlines” as the review prompt and keep the result tied to account and campaign, metric definition, current value, baseline, date and timezone, change point, affected segments, recent edits, tracking health, budget status, suspected causes, and decision deadline.
- Google Ads Help, About change history, official guidance, checked October 5, 2026.
