Google Ads client reporting for small businesses: share understandable evidence and agree the next decision

Begin with the person making the decision

A small-business owner usually needs a report to decide what deserves attention, whether the current advertising plan remains sensible and which action should happen next. An adviser may need more detailed evidence to investigate the account. These needs can coexist without turning the owner's message into a technical export. Begin by identifying the intended reader and the decision they can make. A concise owner-facing report can link the discussion to the relevant evidence, while a follow-up provides the detail required for a particular investigation.

Here, client reporting means the reporting relationship between a small business and the person advising on its advertising. It does not mean reports about the business's own customers. The owner contributes trading context, funding intention and outcome quality; the adviser contributes account interpretation and supported recommendations. Neither perspective is sufficient alone. A clean table of advertising metrics cannot establish whether the business has stock or appointment capacity, while a recollection of a busy week cannot replace the account's actual spending record.

Agree who reads routine messages, who investigates a material change and who can approve a live account edit. A report arriving in Slack does not prove that any participant reviewed it. Likewise, an adviser explaining a result does not automatically have authority to change the connected account. Clear roles make a compact message actionable. They prevent the common situation in which everyone receives the numbers but nobody owns the next decision, or a general performance discussion is mistaken for permission to alter advertising.

Give the report a practical question

A useful report answers a bounded question. It might ask whether recorded spending fits the agreed allowance, whether purchases changed alongside cost, or which campaign merits a follow-up. The question determines the metrics and period. Without it, a report can accumulate every available number and still leave the owner unsure what to do. Choose a small set of measures that supports the decision, then add detail only where the evidence needs explanation. The aim is an understandable conversation rather than a comprehensive catalogue in every message.

Separate a routine review from an incident investigation. A weekly owner report can present the agreed cost and outcome measures. A sudden delivery problem can require current status, policy or destination evidence. Those are related account questions, but they should not be forced into the same data task simply because both involve Google Ads. The assistant should use the appropriate tools for the actual question. An unexplained conversion decline is not automatically a policy issue, and a reported disapproval is not itself a full performance analysis.

Write the decision in plain terms: continue the plan, investigate a named contributor, revise a business assumption or prepare a specific supported edit. A report need not force a change when the evidence supports continuing. An owner who sees stable outcomes may still need to preserve funding or capacity. An owner who sees a decline may choose further investigation before action. The report's job is to make those choices informed and explicit, not to produce a dramatic recommendation merely because a scheduled message is expected.

Name the account and coverage

The report should identify the account and whether it covers the whole account or a selected campaign. A company can use several accounts for markets, brands or operating arrangements. The connected Slack channel does not silently turn one account into a business-wide total. Keep that scope visible when sharing a message with an adviser. If the business plan concerns several accounts, obtain and reconcile the required evidence explicitly. Do not label a single account's cost as total company advertising unless that is actually the agreed coverage.

Filters change meaning. An enabled-campaign question identifies current candidates for action, while an unfiltered historical cost total includes recorded activity from campaigns that may now be paused. A selected Analytics traffic segment differs from all website traffic. A named conversion differs from a category aggregate. The owner need not read implementation details, but they need the scope that changes the interpretation. A short label or sentence can prevent a comparison from appearing contradictory when its two sides deliberately cover different populations.

Keep source access and audience access separate. The current saved reporting configuration belongs to the connected account and Slack destination. It should not be described as arbitrary cross-client broadcasting. Check who belongs to the channel before relying on it as the right place for an owner–adviser conversation. The app's account read and approval rules do not automatically establish that every channel participant should receive every business detail. Use an appropriate existing destination and preserve the account context throughout the reporting and later action handoff.

Agree what counts as a result

Choose an outcome that matches the business purpose. A shop may need purchase counts and purchase value. A service business may need leads, then the owner's assessment of whether those leads became suitable work. A booking business may need an identified booking action. Generic conversion totals can be useful, but only when their meaning is understood. A number called “conversions” may aggregate several primary actions, and a number called “revenue” may represent configured conversion value rather than purchase-only receipts. The report should not conceal those differences behind familiar labels.

Define the population before discussing its efficiency. A cost-per-lead calculation needs a selected lead count. A purchase ROAS needs purchase-only value and an agreed advertising cost. An all-primary value-to-cost ratio answers a broader question. These measures can appear in the same investigation when clearly labelled, but should not be treated as interchangeable. An owner who compares a new report with an old one needs to know if the selected action set changed. Otherwise, a definition change can look like a real performance improvement.

The current saved metric catalogue and conversational measurement choices have different scope. A saved provider-metric report does not automatically adopt every semantic selection available in a conversational question. Check the chosen saved metrics and explain what they actually represent. If the owner needs a purchase-specific or named-action investigation, request that scope through the appropriate supported conversational path. The small-business reporting guide explains these definitions in more detail, helping the adviser keep recurring reports and focused follow-ups consistent.

Keep primary and secondary actions clear

Google's primary and secondary conversion guidance distinguishes their reporting and optimisation roles. For the app's ordinary conversational all-primary aggregate, the intended selection includes primary actions and excludes secondary actions unless the user asks otherwise. State the selected definition in the reporting discussion when it matters. Secondary actions should not silently enter a primary-only comparison because they happen to have recorded activity or assigned value.

When more than one primary action records outcomes, the selected total adds them as configured. Reporting is not an automatic repair of the owner's conversion design. If a primary purchase action and another primary action both contribute, the all-primary value total can include both. Purchase-only reporting selects the purchase category instead. Do not describe every primary value as purchase revenue. Keep an individual named-action view separate when the owner wants to inspect the specific action that drives a change in the aggregate.

Calls can be primary conversions and can carry value. Offline or imported actions can also be visible in the relevant account evidence. They may overlap with broader selected totals, so do not add a calls or offline subtotal onto an aggregate that already includes it. The same principle applies to leads and purchases. The report should make the selected population understandable without turning overlapping views into an inflated grand total. If the owner asks about a particular action, use its actual name and clarify ambiguity before preparing the comparison.

Label source definitions before comparing

Google Ads and GA4 can offer useful related evidence, but their similarly named measures need explicit definitions. Ads purchase conversion value and GA4 purchase revenue can be shown as a labelled comparison when supported, rather than automatically summed. They describe measurement from different systems. Cost comparisons also need the same intended advertising scope. A broad GA4 revenue figure may include traffic beyond Google Ads, while a selected advertising metric may concern a narrower population. Name the source and scope so the owner understands what the lines or table cells represent.

If the owner wants all-traffic GA4 purchase revenue divided by the agreed advertising cost, explain it as blended ROAS or MER and identify the spend source. That is different from a narrower campaign-attributed ROAS. A native GA4 ratio also has its own metric definition; it should not be substituted merely because its name looks familiar. Select the requested numerator and denominator explicitly. The purpose is to answer the business question, not to choose whichever available ratio returns a convenient number.

Lead comparisons require the corresponding GA4 event to be selected because websites use different event names and meanings. The assistant can suggest a likely candidate from actual available names, but the owner should confirm the intended event. Current delivery restrictions still apply to particular named comparison outputs. Do not promise that every possible pair is available in every format today. A report should be honest about the supported view and explain a valid next step, rather than guessing an event or presenting an unverified custom comparison as established functionality.

Distinguish report frequency from data period

The schedule answers when the message should arrive; the range answers which dates it contains. A daily message can cover yesterday or a rolling longer interval. A weekly message need not represent a calendar week. The current saved reporting workflow offers daily, weekly and monthly scheduling, with its available completed ranges selected separately. Keep those choices visible in the owner–adviser agreement. Calling a report “the weekly report” is not enough to establish whether it covers the last seven completed days or another selected window.

The available saved ranges include yesterday, the last seven, fourteen or thirty days, this month through yesterday and last month. Rolling completed ranges end at yesterday in the source calendar. Last month is the previous calendar month. They can differ substantially around a month boundary. A last-thirty-day question should not be replaced with last month just because a dashboard previously displayed that period. Exact date bounds help the owner recognise what changed and help the adviser reconcile the same evidence in another view.

This month through yesterday has no completed current-month days on the first day of a month. That empty interval should not be described as measured zero performance. The owner can choose a different explicit question if current-day activity is needed. Preserve the distinction between a valid empty period, a successfully retrieved interval with no activity and a failed read. A routine report that handles these situations clearly will create less confusion than one that produces a tidy-looking number at the expense of its actual meaning.

Read the comparison's actual dates

The current saved workflow can compare with a previous period or the corresponding period last year, or show no comparison. For ordinary rolling ranges, the previous comparison is the immediately preceding equal-length interval. That is a clear basis, but it may include a different mix of trading events. Always read the date bounds. A percentage change should not be interpreted as a year-over-year seasonal comparison if the configuration actually selected a recent prior interval. The comparison label and dates are part of the report's evidence.

There is a particular detail for last month: the current previous-period calculation uses the preceding equal-length interval, rather than automatically selecting the whole previous calendar month. For example, an October 7 report for September 1–30 compares against August 2–31 under that option. This matters to an adviser trying to reconcile an August calendar-month report. Describe the actual interval without pretending it represents all of August. If a different comparison is needed, request it explicitly through a supported question rather than silently relabelling the result.

Month-to-date and prior-year calculations can require calendar clamping when the corresponding month or day does not exist. Preserve the reported shorter-calendar indication when applicable. A comparison with no recorded activity also needs careful interpretation, especially when its baseline is zero. Do not invent a finite percentage increase from a zero denominator. The owner can still compare the actual current and previous amounts. That straightforward presentation is more useful than a dramatic percentage whose arithmetic is undefined.

Keep source calendars and freshness visible

The Ads account's time zone defines its completed data dates, while an Analytics property can use a different calendar. The configured report-posting time is another setting. These should not be confused. A message received at the owner's local morning can concern yesterday in the source's clock. When comparing sources, check the intended date alignment instead of assuming the same date label always covers the same hours. A clearly scoped comparison remains useful even when the business operates across time zones, provided its basis is stated.

Google's Analytics freshness guidance explains that reporting data is processed at different intervals and recent values can change during processing. A completed date is a useful boundary, not a guarantee that every metric is immutable. Keep this distinction proportionate: a routine report need not recite provider processing details, but an investigation of recent differences should consider freshness before treating a small discrepancy as a permanent measurement error.

If another report disagrees, align the account or property, dates, time zone, traffic scope, selected metric and conversion population first. Then inspect remaining differences. A current-day dashboard, a completed-day Slack message and an invoice can each use different bases. Repeatedly refreshing all of them does not fix a definition mismatch. The adviser should give the owner a specific explanation of the difference that was verified, and obtain any further required evidence rather than using vague claims about “tracking issues” as a substitute for reconciliation.

Show enough inputs to interpret a ratio

A ratio is easier to assess when the owner can see its numerator and denominator. Pair cost per acquisition with cost and selected acquisitions. Pair ROAS with the selected value and cost. Pair average purchase value with purchase value and purchase count. This prevents an attractive ratio from hiding a very small sample or a declining absolute outcome. The adviser can keep the routine message compact while retaining the inputs needed to explain a material movement. The reader should not need to guess what was divided by what.

Calculate an overall ratio from the aggregated valid inputs rather than averaging daily ratios. A day with low cost can otherwise receive the same influence as a day with high cost. The chosen grain also matters for a chart: a weekly ratio should use the relevant week's sums. A missing denominator is not a favourable zero. If no selected acquisitions were recorded, state that the acquisition-cost ratio is unavailable on that basis while showing the actual cost and count where the read succeeded.

Use readable precision. Currency, percentages and efficiency values should not expose a string of unnecessary decimals in a Slack reply. A count can remain precise to the meaningful available value, especially where attribution produces fractional outcomes. Do not round early in a calculation simply to simplify its displayed inputs; use the valid evidence for arithmetic and format the result for presentation. A concise explanation of a changed ratio should connect its input movements to the business question without claiming that the arithmetic alone explains customer behaviour.

Separate no activity from unavailable evidence

A successful read that returns no recorded activity is different from a provider or connection failure. The first can support a no-activity statement for its actual scope. The second cannot establish zero cost, zero purchases or perfect efficiency. Preserve that distinction in the message and comparison. If current evidence is unavailable, explain the limitation and the relevant next check. If only the comparison is unavailable, retain the valid current result without presenting an invented previous value or a calculated change based on missing data.

A preview can also omit rows without indicating the account has no further data. Tell the owner when only a bounded set of examples is displayed. Do not use the preview's count as the complete population. Similarly, a successfully read sparse daily series can have meaningful no-activity dates, while incomplete coverage or unsettled data requires different handling. The assistant should not fill every gap with zero merely to make a graph look continuous. The resulting visual must preserve the evidence that supports each point.

An owner-facing explanation can remain polite and direct without technical clutter. Say which result could not be obtained and what needs to happen next. If the channel lacks the required connection, use the appropriate connection path rather than suggesting that the business performed badly. If a metric definition is incompatible with the selected view, identify the supported selection needed. Clear absence handling protects decisions: the owner should never authorise a spending increase because an unavailable read was presented as unused allowance or weak delivery.

Add business context without inventing causes

The owner can supply facts the advertising system does not know: a promotion, stock shortage, changed opening hours, a website incident or a capacity limit. Put these facts beside the report when they matter. They help explain why the same measured result can have different commercial significance. A higher lead count during full capacity may require a different decision from the same count when appointments are available. The assistant should preserve the origin of the context rather than claiming it discovered a fact that the owner supplied.

Context is not automatic causal proof. A promotion during a cost increase can be relevant without establishing that it caused the entire movement. Several account and trading changes can occur together. Explain the observed metrics and identify evidence that could support the next conclusion: a campaign breakdown, relevant change history or a selected outcome comparison. Avoid turning a chart into an account of customer psychology. A bounded investigation is more credible and useful than a confident narrative built around an unverified explanation.

Distinguish planned changes from completed ones. A note saying a campaign was intended to pause does not establish the pause occurred. A website manager's plan does not prove the page now works. The report can include the plan as context while assigning the relevant state or implementation check. This habit keeps the owner–adviser conversation grounded. It also prevents later performance from being interpreted against an action that was never applied, or an execution receipt from being mistaken for proof of a commercial improvement.

Follow an invented shop's reporting discussion

Consider an invented shop, Meadow Paper Goods, comparing two clearly defined completed periods. Suppose the earlier period recorded 800 of advertising cost, forty selected purchases and 3,200 of selected purchase value. The current period recorded 1,000 of cost, fifty purchases and 3,500 of value. These figures are an illustrative explanation, not a customer result or a recommended target. Cost and purchases each increased by 25%, while value increased by about 9.38%. The owner can understand that the account recorded more selected purchases while value grew more slowly than cost.

The selected cost per purchase remains 20 in both periods. Purchase ROAS changes from 4 to 3.5, and average purchase value changes from 80 to 70. Those ratios are calculated from their total inputs, not from averages of daily ratios. A useful owner summary says that purchase volume rose at the same cost per selected purchase, while average value fell and value-to-cost efficiency declined. It does not say that every new customer became less valuable or that the adviser caused the change. The evidence establishes the selected arithmetic, not an entire commercial explanation.

The owner adds that a lower-priced collection was promoted during the current period. That is relevant context, while a product or campaign breakdown can investigate its contribution. The adviser asks whether margins, stock and funding remain suitable before proposing any account change. The next decision might be to continue the plan, examine a named contributor or prepare a supported edit. A readable report enables that discussion. It should not turn the lower ROAS into an automatic campaign pause or claim that spending should increase merely because purchase counts rose.

Connect commerce reporting with real economics

For a shop, purchase reporting is useful evidence about measured advertising outcomes. It is not a complete profit statement. Margins, returns, fulfilment costs and available stock can change whether additional advertising is desirable. The owner or financial records may supply these facts. The assistant should not invent them from a conversion-value total. A report can flag a movement in purchase value or order count while clearly identifying the commercial context needed to judge the plan. This keeps the conversation grounded in what the available evidence actually supports.

Average purchase value can help explain a revenue movement when the selected purchase definition is consistent. In Google Ads, purchase-only value divided by purchase conversions can provide the intended account view; GA4 has its corresponding average purchase revenue metric for its measurement scope. Match the selected population before comparing. Do not use an all-primary conversion count as the denominator for a purchase-only average simply because it is already in the saved report. Definitions that mix populations can produce a plausible number with the wrong business meaning.

Keep advertising-attributed and all-traffic outcomes distinct. The owner may want the broader relationship between total purchase revenue and an agreed advertising cost, especially when reviewing the business's overall funding plan. Label that blended ROAS or MER and identify the source and spend coverage. A campaign-specific efficiency discussion is another question. Neither view is automatically superior; they support different decisions. The adviser should explain which question the report answers so the owner does not treat an all-traffic measure as the return directly attributed to one campaign.

Connect service reporting with suitable work

For a service business, leads are often an early stage rather than the final commercial outcome. A form, call or selected booking event can indicate interest. The owner may later determine whether the enquiry was in the service area, involved suitable work and became a paying job. Keep those stages separate. A platform lead count should not be described as completed work unless the selected action actually represents that stage. A useful report invites the owner to add outcome quality rather than assuming every recorded enquiry has equal value.

If comparing Ads leads with GA4, identify the corresponding event and confirm its intended meaning. Websites can use different names for enquiries, and a phone-link click need not establish a completed call. The assistant can inspect available events and suggest candidates by name, but the owner's confirmation matters where the correspondence is ambiguous. Current named-comparison delivery restrictions should be honoured. Do not promise a universal finished comparison when the supported output path is not available, or substitute all key events for a selected lead event without agreement.

Capacity and seasonality can change the decision even when cost per lead remains stable. An owner whose schedule is full may prefer fewer enquiries, while another may have room for more suitable work. A higher lead count outside the service offer can be less useful than a lower count of appropriate enquiries. Supply that business context in the thread. If it warrants a targeting or status change, the adviser can request a supported exact proposal. The lead report itself does not authorise the assistant to choose and apply that change.

Write a useful owner-facing explanation

Lead with the observed result and its practical significance. State the period and selected outcome, explain the material input movements and identify the next decision. A short summary can be complete without repeating every cell. For example, distinguish rising volume at unchanged acquisition cost from rising cost with declining volume. Use the actual evidence and avoid formulaic claims that every increase is good or every decrease is bad. The owner should understand why the adviser wants a particular follow-up, rather than having to infer concern from a coloured percentage.

Separate observation, interpretation and recommendation in plain language. The observation is what the selected data shows. Interpretation connects it with known business context and may identify further evidence required. The recommendation names the next check or supported action. These distinctions can fit in connected prose; they need not become a technical checklist. If something important is unresolved, ask the specific question that resolves it. A vague statement that performance “may be affected by many factors” does not help the owner decide what happens next.

Keep the message polite and focused. Do not expose implementation details or raw asset identifiers that the reader does not need. Use understandable names and relevant parents when a target is involved. A report about results should not read like a mutation payload. The client-facing explanation also should not claim provider certainty that was not returned. If a result is unavailable or a definition differs, state the verified limitation and appropriate next step. The report's credibility comes from useful evidence and clear decisions, not exaggerated confidence.

Choose a readable table

A table helps when the owner needs to compare a small set of contributors or inspect the inputs behind a conclusion. Choose columns that serve that purpose. A campaign review might use campaign name, cost, selected outcome and efficiency. A period comparison might use current value, prior value and change. Avoid adding every available diagnostic simply because the data library contains it. The recurring owner report can remain compact, while a focused follow-up adds the fields that answer the next investigation.

Display the actual units and scope. A value column should say whether it is purchase value or another selected conversion value. A lead column should represent the agreed lead population. A campaign table restricted to enabled campaigns should state that restriction. If only a preview is shown, identify it as a preview. The number of displayed rows does not establish the complete account count. Sorting and formatting the same valid result can reuse the existing evidence instead of repeating the provider call.

Current conversational output restrictions still apply. Do not promise arbitrary multiple tables in a single delivered reply where that capability is not currently enabled. Split the task into a supported readable response or narrow the selection as appropriate. The reader need not see internal capability switches, but the behaviour should honour them. The same principle applies to unsupported custom comparisons: explain what can be shown and request the required selection rather than creating a misleading table whose headings conceal unavailable evidence.

Use one chart to answer one comparison

A chart should make a defined relationship easier to see. When the owner asks for two or three supported metrics over the same period, the app's conversational native chart can show them in one appropriately labelled graph rather than automatically creating separate charts for each. Keep source, unit and scope clear. Two revenue measures can be compared as distinct series without adding them together. A cost series and a count series need labels that prevent the owner from treating their values as the same unit.

The current native chart limit is three series and twenty points per series. When the requested interval would exceed that point count and daily detail is not explicit, a weekly or monthly grain can keep the result in one readable chart. State the breakdown. When daily detail is explicitly requested, explain a necessary split and the point limit instead of silently omitting dates or presenting two panels without context. For intervals beyond the supported panel capacity, a narrower period, coarser grain or table may be needed.

Use intelligible dates. Current-year labels can omit the year, while earlier-year dates need enough context to remain unambiguous. Weekly labels should use actual date ranges rather than raw year-week codes. Monetary and ratio values should have readable precision. A ratio at a coarser grain must use the correct aggregated inputs, and missing observations must retain their real meaning. The visual should answer the owner's question from available evidence, not manufacture a complete-looking line through failed reads or label a rolling period as a calendar month.

Configure the current saved reporting workflow

The current /ppctor reporting workflow manages saved reports for the connected account and channel. It selects Google Ads or Analytics as the report source, the supported target or segment, compatible metrics, timing, range and comparison. These settings define the recurring evidence and destination. A saved configuration uses one selected source; it should not be described as a universal arbitrary cross-source report builder. Conversational questions can perform supported focused investigations separately, while the saved report remains a clear recurring agreement between the owner and adviser.

The saved configuration accepts between one and ten compatible metrics. Compatibility depends on the source and target, so a list of metric names is not permission to combine every field. Saving queues the intended work rather than immediately proving that a report was delivered. A manual Run Now also operates on an enabled saved configuration in the current flow. Keep save acknowledgement, queued evaluation and delivered result distinct. The owner should know whether a configuration was accepted or the actual evidence has arrived.

Review the selected range and comparison when setting the schedule. A weekly posting time does not automatically set a weekly data range. Analytics data uses the connected property's reporting calendar, while scheduling remains part of the saved configuration. Optional notification choices should match who needs attention. The app does not need to mention every participant on every deterministic report unless the configured behaviour requests it. A sensible routine delivers useful evidence at an agreed time and gives the business an identifiable person responsible for reviewing it.

Use follow-ups to investigate the same evidence

A report can start an ordinary Slack conversation. The owner might ask why ROAS changed, which campaigns contributed to higher cost or whether a named action explains a conversion increase. The assistant should preserve the account, period and selected definitions from the discussion. It can explain the already retrieved arithmetic before deciding whether another read is needed. A request to reformat the same result as a chart or table should not automatically trigger an unrelated account-wide investigation that loses the original context.

Fresh evidence is needed when the request changes the relevant period, metric, population or live state. A campaign's current budget cannot be inferred from a historical performance table. A purchase-only result cannot be invented from an all-primary total without the necessary selection. A named event comparison can require clarification and supported output handling. Use the available metric and dependency framework to choose the data task, rather than adding a special case for every phrasing. The owner should get a coherent answer to the actual follow-up.

Creative conversation also belongs in the flow. The assistant can suggest an explanation outline, brainstorm truthful headlines or help the owner consider a trading plan using supplied context. Those tasks do not inherently require provider calls. If the user requests current account facts or an actual edit, the tool path then obtains the appropriate evidence and authority. This balance makes the conversation more useful than a rigid report menu without allowing creativity to invent measured results or bypass confirmation for a live account change.

Turn the discussion into an assigned decision

Close a reporting discussion with the next task, owner and intended review. The task might be to inspect a campaign, confirm lead quality, check a page or prepare a supported edit. If the decision is to continue the plan, record the reason and the next check. A report does not need an action for its own sake, but it should not leave a material concern indefinitely unexplained. Clear ownership helps a small business avoid repeatedly discussing the same percentage without resolving what evidence or decision is still missing.

Separate business funding from delivery controls. Revising a monitoring allowance changes a planning comparison; changing a campaign budget changes a supported account setting. The budget-tracking guide explains recorded cost and the current trajectory. An owner can deliberately revise funding when the business context changes, but the report should not make that look like improved advertising efficiency. Likewise, a lower campaign budget can protect cash without repairing unsuitable demand or a measurement problem.

Where several responsibilities are involved, list the tasks separately. The website manager may own a destination repair, the adviser may prepare a creative revision, and the connected owner may confirm a live proposal. A successful account edit does not complete the website task. A website note does not prove that the ad now uses the intended page. A concise handoff gives each person the relevant evidence and next check, preventing the client report from becoming a single reassuring message that obscures an unfinished dependency.

Keep account edits behind exact approval

An owner who decides to act can ask for a specific supported proposal. It should show the understandable target, actual relevant parents and exact action, with separate visible steps where needed. Raw resource identifiers should remain behind the scenes. A keyword needs its group and campaign context; an ad belongs to its actual parents; a Performance Max asset group must not be described as an ad group. Accurate hierarchy helps the owner recognise what is being approved before the live edit occurs.

Confirm and Execute is the approval point for the authorised connected owner. Before that confirmation, no live changes have been made. The proposal expires after fifteen minutes, and relevant settings are checked again before execution. A performance report or adviser recommendation does not replace that authority. If current state changed or a requested operation is unsupported for the target, the flow should explain the condition rather than silently applying an alternative. Multi-step requests need correct dependency order without claiming universal atomic execution or automatic rollback.

The visible in-progress state should make a valid click clear, and the final receipt should distinguish verified execution from an accepted write whose complete result is uncertain. Inspect uncertain target state before another mutation; do not blindly replay a possible success. The approval guide describes that handoff. Later business effects need their own scoped review. A verified budget, status or creative change establishes the requested setting, not guaranteed purchases, suitable enquiries or a particular future ROAS.

Maintain the report as the business changes

Review report definitions when the business changes its offer, account structure, conversion setup or decision process. A metric that once represented the main outcome can become incomplete after a new booking flow or imported action is added. A campaign-specific report can become less useful if the offer moves to another campaign. Make these changes explicit. An improved number after a definition change is not necessarily improved performance. The owner and adviser need to know which comparison remains meaningful and which baseline must be interpreted differently.

Review scheduling and audience as well. A report can be valid but arrive when nobody responsible reads it. A former adviser may no longer need notifications. A channel connection can change. Do not assume that continuing message delivery proves the reporting arrangement remains appropriate. Maintain the enabled configuration and its selected destination deliberately. Pausing a saved report is different from pausing advertising. If the business no longer needs the report, manage that configuration separately rather than assuming silence means the campaigns stopped.

Retain a concise decision history rather than copying every metric into a growing document. Record the period, material observation, agreed context, chosen action and later review. This helps a new adviser understand the owner's intention and helps the owner see whether a recurring concern was resolved. The history should preserve uncertainty where evidence remained unavailable and should distinguish executed settings from commercial results. A well-maintained client report supports a sequence of informed choices, not just a set of attractive messages that cannot explain why the account changed.

Answer common client-reporting questions

Should the owner receive every available metric?

No. Select the measures that answer the agreed decision, with enough inputs to interpret important ratios. Add detail in a bounded follow-up. A compact message can be complete when its scope, period and meaning are clear.

Can a weekly report cover the last thirty days?

Yes. Posting frequency and data range are separate settings. Label the actual range and dates so the reader does not mistake weekly delivery for a seven-day data window. Last month and the last thirty completed days are different ranges.

Can the owner compare Ads and GA4?

Supported related measures can be compared with explicit definitions, traffic scope and dates. Do not add overlapping revenue values or guess a corresponding lead event. Current restrictions on particular named comparison outputs still apply, and the assistant should explain the supported next step honestly.

Does the report authorise live changes?

No. It provides evidence for a decision. A supported live edit needs its own exact proposal and authorised confirmation. After execution, verify the stated account change separately from the later business effect. A clear client-reporting routine keeps those stages understandable for both owner and adviser.