Google Ads client reporting for agencies: explain performance without losing its meaning

Start with the client's decision

A client report should help somebody understand a business question and decide what to do next. Producing a table is only part of that work. The agency needs to explain which evidence answers the question, what the evidence does not establish and whether a decision is actually required.

Start with the purpose of the review. The client may want to assess purchases, understand spending against an agreed amount or investigate a particular campaign. Those are different questions. A familiar set of columns should not determine the purpose simply because it is easy to retrieve.

Ask what the client needs to decide at this point. A budget discussion may need observed cost and a period target. A lead discussion may need a selected conversion action and supplied business context. A launch review may concern current configuration rather than historical performance.

Keep the answer bounded. A clear result about one defined question is more useful than a long account summary that mixes unrelated observations. Additional evidence can follow when the client asks another question or the first result identifies a genuine need.

The best client communication makes its definitions easy to recover. A colleague should be able to tell which account, dates, outcome and population support the answer without reconstructing the entire conversation. This is the difference between a readable report and a persuasive-looking collection of unexplained numbers.

Keep the agency handoff consistent

An agency often has one person retrieving evidence and another presenting it. The reporting handoff needs to preserve meaning across that boundary. A specialist's precise metric can become a vague revenue or leads label when the account manager summarises it.

Imagine a fictional agency, Maple Quay Media, serving a retailer and a service company. Both clients want a monthly review, but their outcomes differ. A reusable format can preserve dates, sources and decisions without pretending the same conversion definition serves both businesses.

Assign responsibility for the report definition and for the client discussion. The analyst can establish the successful read and its scope; the account manager can obtain the client's commercial interpretation. Neither should represent the other's contribution as independently verified by the provider.

Keep account context explicit when names repeat. A campaign called Main Search can exist for several clients. A quoted performance result or proposed operation needs its connected account and actual target, not a remembered label from another channel.

A consistent handoff records the question, evidence definition, key result, open interpretation and next decision. It standardises discipline rather than recommendations. This lets a colleague continue the review without applying one client's priorities or assumptions to another account.

Agree on the outcome definition

Choose the measured outcome before writing the conclusion. ppctor's conversational framework distinguishes aggregate primary conversions, primary purchases, primary leads and named conversion actions. A client asking how conversions performed may mean any of these.

Aggregate primary conversions includes the selected account's primary actions. Secondary actions are excluded from that default. The measure does not deduplicate separate actions that record overlapping business activity; it reports the configured aggregate rather than repairing the advertiser's tracking design.

Primary purchases uses the purchase-category selection. Primary leads groups the lead categories together rather than treating their labels as an inferred quality hierarchy. A named action answers a more specific question and should remain selected throughout relevant follow-ups.

Ask for clarification when the distinction matters. The assistant should not select the measure that produces the most favourable-looking conclusion. It can explain the available choices and let the user confirm the intended business question.

Put the selected definition near the result. The client does not need every internal field, but it should understand whether the answer concerns purchases, all primary actions or one named conversion. This small statement prevents several later reporting disagreements.

Explain revenue without a vague label

Google Ads primary conversion value can include value from primary actions outside purchase conversions. Primary purchase conversion value is a narrower selection. A report should not call both simply sales revenue and assume the client will infer the difference.

When the client wants all primary value, name that definition. When it asks about purchases, use the purchase selection and retain the source. Multiple configured primary actions contribute to the aggregate; the report does not independently establish a deduplicated accounting total.

Analytics purchase revenue also needs a traffic scope. All-traffic purchase revenue is different from a Google Ads-related population. Select the intended scope before comparing sources or calculating a return measure.

Keep the denominator of a return metric explicit. All-traffic Analytics purchase revenue divided by an agreed advertising-cost scope is a blended ROAS or MER view. It is not automatically the same question as purchase value attributed in Ads divided by Ads cost.

These distinctions do not prevent useful reporting. They make the result interpretable. The agency can compare related revenue measures in a compatible visual while explaining their source and scope, rather than force equality or combine them into an unsupported sales total.

Confirm the lead event correspondence

A lead report needs to identify what counts as a lead for this client's measurement. The Ads primary lead categories are grouped together in the conversational measure. Their category names do not establish a verified sales qualification stage.

For an Analytics comparison, choose the corresponding event. The assistant can inspect available event names and suggest a plausible candidate, but the user should confirm that it represents the intended activity. A name alone does not prove the event's implementation.

A comparison of a named Ads conversion with a specific Analytics event is useful when the client asks about a particular form or action. Resolve ambiguous names rather than selecting the first similar one. Preserve the selected pair in later tables and charts.

Keep call and offline context accurate. Primary call conversions can carry value, and imported or uploaded conversions can be visible in supported Ads reporting. Their presence does not imply access to the client's complete call-quality or CRM records.

The client can supply business context about the usefulness of leads. That context can guide a recommendation, but it should remain distinct from the provider's retrieved count. This keeps the report useful without pretending a conversion label independently measures lead quality.

State the population being reported

An account total, campaign result and keyword table represent different populations. State which one the client requested. A campaign result should not be presented as the whole account merely because it is the most important campaign commercially.

Keep status filters explicit. Enabled assets describe a current selection, while paused assets can still have historical results. An asset's own status and its parents' statuses can also differ. The report needs the relevant scope rather than an unexplained active label.

For segmented Analytics results, preserve the selected traffic population and dimension. A channel breakdown and a daily breakdown answer different questions. If both are required, the plan must support their combination rather than substitute fourteen-day channel totals for daily values.

Do not add overlapping levels. Campaign cost already includes lower-level activity in its scope. Summing campaign totals with their keyword rows would count the same spending again and make the client report misleading.

Use complete scoped totals and counts where supported. A bounded preview shows examples for readability; it does not define the population denominator. A report should indicate when evidence is incomplete instead of presenting partial rows as a comprehensive account statement.

Use actual dates and reporting basis

Name the actual start and end dates. Last month and the last thirty days are different periods. A current-period budget review is another question. The client should not need to infer the dates from an ambiguous title.

Keep complete-day boundaries aligned with the account or property context used by the report. A team member in another time zone can otherwise interpret yesterday differently from the selected source. Reporting dates should remain stable throughout the review.

The current conversational default for daily conversion outcomes uses conversion date. An explicitly requested interaction-date view has a different interpretation. Explain the chosen basis when it affects a comparison rather than treating shifted daily outcomes as missing results.

Google's data-freshness guidance notes that metrics can process at different speeds and receive later updates. Recent results should be read with that context. A complete calendar day does not guarantee that every conversion-related figure is final.

Preserve the period in presentation changes. Asking for weekly grouping changes the grain; asking for last month changes the question. The assistant should not shorten a requested window merely to fit a chart or match an old saved dashboard view.

Build a meaningful comparison

A comparison needs a defined reference period or population. Previous period and the corresponding period a year earlier can serve different business purposes. Select the one that answers the client's question and make its dates visible.

Keep the measure consistent across both sides. Comparing purchases in one period with all primary conversions in another does not show purchase growth. A changed definition can still be reported, but it needs an explicit explanation rather than a single percentage headline.

Check relevant commercial changes. The client may have changed its offer, budget or campaign structure. These are useful supplied context, not automatic explanations of every measured movement. The agency should distinguish an observed difference from a causal claim.

When the reference value is zero or unavailable, do not invent a normal percentage change. Report the defined values and explain why the ratio is not meaningful. A dramatic infinity-style conclusion rarely helps a client decide what to do.

Use comparisons to focus investigation. A decline can prompt a question about source, scope or recent changes; it does not automatically justify pausing the selected campaign. The report's next decision should follow the evidence rather than a generic response to a red number.

Compare sources without merging them

Ads and Analytics can provide related measurements with different scopes and attribution. A compatible comparison can be valuable without requiring the two values to match. The report should identify both measures rather than present them as interchangeable records.

For purchase-related value, keep Ads purchase conversion value and Analytics purchase revenue separately labelled. For all primary conversions compared with Analytics key events, make the aggregate definitions clear. Named action and event comparisons require their selected correspondence.

Do not add two source values to manufacture a combined outcome. They may describe overlapping activity. A two-line chart compares the series; it does not establish that the sum is a larger amount of revenue or more unique conversions.

Choose compatible units for a shared visual axis. Revenue values can be compared under their supported definitions, while a money measure and a count measure need another appropriate presentation. Different units should not be forced into one unlabeled axis.

If a comparison cannot be constructed from the available compatible data, explain the actual missing requirement. Do not claim a universal provider limitation when the problem concerns the selected plan, event correspondence or current tool path.

Keep users, sessions and outcomes distinct

Users, sessions and conversion outcomes are different units. A session count is not a customer count, and a selected event count is not automatically a unique number of people. Client-friendly language should preserve these distinctions.

Google's reporting-identity guidance explains how Analytics uses identity spaces in user reporting. This reinforces the need to name the actual measure rather than translate every traffic metric into visitors or customers.

A daily breakdown of users also should not be assumed additive into a unique period total. The same person can appear across days. Use the appropriate supported aggregate for the selected measure rather than adding displayed rows indiscriminately.

Paid traffic needs an explicit population. A request can concern a defined paid channel grouping or a particular source. The assistant should preserve that filter in the daily plan instead of substituting all traffic or a single advertising source silently.

Explain the unit in practical language. The client can understand that the report concerns sessions from the selected population and a separately measured purchase value. Clear units prevent a persuasive narrative from making claims the dataset does not establish.

Calculate aggregate ratios correctly

Aggregate ratios need their combined inputs. Total selected value divided by total selected cost gives the corresponding combined ROAS. An equal average of row-level ratios can overemphasise small rows and misrepresent the client's spending distribution.

The same principle applies to CPA and rates. Use the selected total cost and acquisitions, or the appropriate combined numerator and denominator. Do not average daily percentages merely because a chart already shows them.

Keep undefined ratios separate from normal numeric values. No selected acquisitions does not produce an ordinary CPA that can be displayed with false precision. Missing required inputs also should not become zero.

Retain the revenue definition in ROAS. Primary-value ROAS, purchase-only ROAS and a blended ROAS or MER view can all be useful, but their numerators answer different questions. A label should make the selected relationship clear.

Use at most two decimals for client-facing metrics. Rounding is a readability rule, not permission to hide an undefined value. A clean table should show meaningful precision and preserve unavailable evidence with an understandable label.

Report missing evidence honestly

A successful zero and an unavailable value are different facts. If a requested metric could not be returned, the report has not established that the client produced no activity. The response should say what remains unavailable.

Incomplete coverage also needs a visible qualification. A capped result should not be presented as a comprehensive account total. Narrow the selection or choose another supported breakdown rather than silently discarding rows to produce a tidy chart.

Keep provider-read problems separate from business outcomes. Lost access, an unsupported field combination or incomplete metadata does not prove that the campaign stopped generating results. The next task is to investigate the evidence path.

When the client supplies another total, reconcile definitions first. It may use different dates, action selections or source scope. The agency should not dismiss the client's figure or declare its own correct without establishing what each measures.

An honest incomplete answer can still move the conversation forward. State the evidence successfully retrieved, the missing requirement and the next supported step. This is more useful than filling a gap with a plausible number or an unsupported explanation.

Write an evidence-backed narrative

Lead with the answer to the client's question, then explain the evidence definition. A concise narrative can state the selected period, outcome and key movement before giving the supporting comparison. It should be understandable without an opening paragraph of field names.

Distinguish observation from interpretation. The tool may establish that selected cost rose and selected purchases fell. A hypothesis about competition or landing-page behaviour requires further evidence. The narrative should not convert a plausible explanation into a confirmed cause.

Use supplied business context explicitly. The client may confirm that a promotion ended or a service became unavailable. That fact can help interpret the result, but it came from the business rather than the performance API.

Keep recommendations proportional to the evidence. An unresolved measurement discrepancy can justify another scoped read or clarification. It does not automatically justify a campaign pause. A confirmed commercial mismatch can support a more precise targeting review.

End the narrative with the next decision or open question. The agency should explain what information would change its recommendation and who can provide it. This turns reporting into a useful conversation instead of a polished account of numbers with no practical purpose.

Work through an invented client report

The following names and figures are invented and do not describe customer results. Maple Quay Media reports on a retailer's selected primary purchase outcomes over two defined equal-length periods. The report uses the same account, purchase definition and reporting basis on both sides.

The defined cost and purchase count both rose by 20%, while selected purchase value rose by 14%. The corresponding purchase-only ROAS fell from 4.00 to 3.80. These calculations describe the invented inputs; they do not establish the cause of the change.

A clear client explanation would state that purchases increased under the same selected definition, while value grew more slowly than cost. It would ask about relevant offer or account changes before recommending a specific operation.

The agency should not call the selected value a deduplicated accounting total or claim that every extra purchase was profitable. The report does not contain margin, all business costs or a causal experiment. Its value is a defined observation that helps the client choose the next investigation.

Choose a readable chart or table

Select a visual only when it helps answer the question or the user requests one. A table can compare exact selected measures, while a line chart can show a trend. A pie requires additive nonnegative parts of one measure; ratios are not parts of a whole.

Conversational native charts support up to three compatible series, with bounded points per series. For an unqualified long trend, a coarser breakdown can keep one chart readable. If the user explicitly requests daily detail, preserve that requirement within the supported output and explain any necessary split.

Use compatible units on a shared axis. Two related revenue series can remain distinct and labelled. A revenue series and a session count should not be forced onto the same unlabeled axis merely because both were requested in one sentence.

Keep date labels readable. Current-year labels need not repeat the year, and weekly presentation should convey a meaningful date range. The selected underlying period should remain unchanged when the formatting becomes simpler.

Missing values need an appropriate presentation. A table can preserve unavailable observations where a native chart would imply continuity or substitute a false zero. Readability should reveal evidence limits rather than erase them.

Ask for the report in Slack

In the connected account conversation, state the dates, selected outcome and desired scope. A request for purchase-only ROAS over the last thirty days with a trend gives the assistant a clearer plan than show performance.

The generic tools can describe available fields, retrieve a validated compatible dataset and render requested visuals from returned data. The assistant should select evidence that answers the question rather than invent values or run unrelated reports.

If the client asks for a comparison with Analytics, resolve the linked property and intended traffic or event scope. A lead comparison may need user confirmation of the event. Missing correspondence should become a clear question rather than a guessed pair.

Use normal asset names and relevant parents in the response. A client report should not expose campaign or ad-group identifiers simply because the internal tool uses them for resolution. The human needs recognisable context.

For the production workflow, see agency reporting. This client-communication workflow adds interpretation and decision ownership; it does not imply that every conversational request creates a scheduled report or live account change.

Preserve context in follow-up questions

A follow-up should retain the established account, period, outcome and population unless the user changes them. Asking show that as a table should preserve the evidence definition. Asking for a different metric or grain may require another compatible read.

Use returned data for a new presentation when it is sufficient. A formatting request should not trigger a second identical provider query. Reuse makes the conversation quicker and reduces the chance that two presentations reflect different retrieval states.

Resolve ambiguous references. If several campaigns appeared in the previous result, pause it needs a target clarification. If only one unambiguous resolved target is in context, the assistant should preserve that target rather than discard the conversation.

Keep a client's correction meaningful. A user who says use total paid traffic is changing the population. The next plan should preserve that explicit instruction instead of repeating the earlier narrower selection.

Explain any real limitation specifically. An unsupported field combination, missing source or capped selection needs a concrete adjustment. A vague statement that metrics cannot be compared should not replace checking the framework's available compatible measures.

Allow useful creative discussion

Reporting conversations can include brainstorming. The client may ask for possible explanations, a clearer headline or alternative wording for an offer. The assistant can respond from supplied context without making an API call for every creative suggestion.

Keep the suggestion distinct from measured evidence. A drafted headline is not a tested winner. A possible explanation is not a verified cause. A generated keyword idea is not a retrieved search-volume result unless the user requested and the tool returned that research.

Do not invent product facts or commercial promises. If the client has not supplied a guarantee, certification or feature, the assistant should not add it to a proposed response merely to make it sound persuasive.

Use discussion to clarify the decision. The agency can explore what evidence would support a hypothesis or how to explain a measurement difference. This can be valuable even when no additional provider read is necessary.

Move to execution only when the user requests a supported account action. Agreeing that a draft is useful is part of the conversation; the exact live operation still needs its reviewed proposal and confirmation boundary.

Use recurring reports for defined evidence

The saved Reporting workflow supports defined Ads or Analytics reports with a selected scope, period, metrics and cadence. Daily, weekly and monthly delivery can make routine evidence available without each client question rebuilding the same selection.

Current supported ranges include yesterday, rolling seven, fourteen and thirty days, the current month through yesterday and last month. The saved definition should match the intended client review rather than remain on an old period by habit.

Supported comparison choices include no comparison, previous period and year over year. Keep their actual dates and selected metric definitions visible. A recurring report is useful only if the client understands what its comparison represents.

A report delivered to Slack does not imply an account manager has reviewed it. Assign responsibility for interpretation and follow-up. A scheduled evidence message and a client recommendation are different stages.

Do not assume every conversational business measure is identically defined in every deterministic report. Check the saved metric selection and output labels. A named-action investigation should not be compared blindly with a broader provider total.

Separate recommendations from live edits

A report can recommend investigation or a supported change, but its existence does not authorise an edit. The user needs to request the action, and the assistant must prepare the exact target and operation.

Show asset names and relevant parents in the proposal. A keyword action should identify its ad group and campaign; a campaign action should identify the campaign. The client should not have to infer the target from a technical resource string.

Use distinct steps for a multi-action request. Adding keywords and enabling their group are different operations. The confirmation should make both visible and preserve any required order rather than hide the second action inside a long description.

Respect actual operation support. The ordinary campaign-budget edit supports non-shared daily budgets, not every provider budget type. An audit finding or reporting field does not automatically imply a conversational write exists for it.

The proposal remains unexecuted until Confirm and Execute is used and has a defined expiry. The client can review the exact change while no live mutation has yet occurred. See the confirmation workflow for that handoff.

Explain execution and business results separately

After confirmation, the edit path checks relevant authorisation and current settings, sends the supported request and verifies what it can establish about the result. A verified configuration is an account-operation fact.

It does not prove that the client will receive more revenue or lower acquisition cost. The business follow-up requires a defined later measurement. A success message should not imply a guaranteed commercial improvement.

For creative, saved content and provider approval are separate milestones. A verified replacement can still be awaiting review. The client should see what was completed and what remains dependent on Google's process.

If execution is uncertain or not fully verified, inspect the current account before retrying. A mutation may already have succeeded even when the readback was incomplete. Blind repetition can duplicate a creation or apply another unintended operation.

Retain the selected follow-up definition. Changing the outcome or date basis after the edit can make the apparent result misleading. Explain any intentional definition change instead of presenting the new figure as directly comparable.

Prepare a focused client review

A useful review meeting begins with the question and evidence definition. The client should know which dates, source, population and outcome support the headline. The agency can then discuss interpretation without spending the whole meeting reconciling labels.

Present one or two meaningful observations before expanding into detail. A bounded table or chart can support the conversation. A long list of technical rows usually makes it harder to identify the decision the client needs to make.

Separate confirmed context from open questions. If the client has not explained an offer change, ask rather than infer it. If a source discrepancy remains unresolved, identify the exact next read or event confirmation.

Make approval responsibilities visible. A recommendation to change an account setting should lead to a reviewed proposal, not an implied decision hidden in meeting notes. The human needs to assess the named target and exact action.

Close the discussion with the next owner, evidence requirement and review point. This is an agency operating method rather than a claim that the tool creates an external project-management workflow. Its purpose is to make the reporting conversation actionable.

Respond when the client challenges the headline

A challenge is a reason to inspect the definition, not repeat the same conclusion more confidently. If the client says its sales were different, identify what its figure includes. It may concern all traffic, another period or a business accounting total. The agency's report may concern attributed purchase value under a narrower selection. Neither statement should be declared wrong until their meanings are established.

Return to the smallest evidence question that can resolve the discrepancy. Confirm the dates and source first, then the selected action or event. If the client requests a named conversion, resolve that name rather than giving the broader primary aggregate again. If it requests all paid sessions, preserve that population in the next plan. A direct correction should change the analytical selection, not merely the wording around the old result.

Where the original output was incorrect, correct the finding and its affected conclusion. Do not leave a misleading recommendation in the meeting record while adding a technical caveat elsewhere. The next decision should use the revised evidence. This keeps the discussion accountable and prevents an old headline from driving an inappropriate edit.

When the difference remains real after definitions are aligned, explain the remaining scope without guessing a cause. A source comparison can be useful precisely because the measurements differ. The agency should state what it verified and what further evidence would resolve the open question.

Retain a concise decision record

Keep a short record of the client's question, selected evidence, key observation and decision. Include the account context and dates. Another colleague should be able to understand the result without reproducing the full Slack transcript.

Label supplied business facts separately from returned data. A client-confirmed promotion change can inform interpretation, but it is not a provider field. A tool-returned metric cannot independently verify the client's commercial agreement.

Record an action's stage precisely. Proposed, approved, executed and checked are different states. Any uncertain result needs an owner and investigation step rather than a generic done label.

Retain the planned follow-up measure and period. This prevents the next report from silently changing the denominator or outcome definition. If the business changes its objective, document the new definition and effective point.

Keep amendments visible across reporting cycles

A client can change which outcome matters between reviews. For example, an invented service business might move from all primary leads to a named contact-form action for a particular campaign discussion. The new choice should be recorded as an amendment, not applied retrospectively to an earlier headline without explanation.

Keep the earlier definition available so a colleague can tell why totals differ between reports. This does not require exposing the internal query or storing a long technical transcript in the client reply. A short description of the selected measure, scope and dates is enough to preserve the comparison boundary.

If a chart uses a coarser grain for readability, record that presentation choice separately from the evidence period. Weekly grouping does not mean the request changed to a different week range. A later request for daily detail should retain the same period and use the compatible supported data rather than choose convenient dates.

This amendment discipline also helps approval. A live operation recommended under one objective should not be carried forward automatically after the client changes the question. Reassess the evidence and exact target, then prepare a fresh supported proposal if an account action is still requested.

Use related guides for distinct tasks: agency budget tracking for agreed spending, agency audits for settings triage, and metric explanations for conversational definitions. Keep their evidence and authority boundaries clear.

Agency client-reporting questions

Should every client use the same conversion total?

No. Preserve the client's selected objective. All primary conversions, purchases, leads and a named action answer different questions. A shared report format should retain those differences.

Can Ads and Analytics revenue appear together?

A compatible comparison can show separately labelled series with their source and scope. Do not add them into an unsupported combined total or assume they must be equal.

Does the AI need research for every suggestion?

No. Creative ideas and explanations can use supplied context conversationally. Account measurements require returned evidence, while live operations require the supported confirmation path.

Does a successful edit prove better performance?

No. It verifies the account operation within its supported checks. Measure later business results separately under a defined period and outcome.

What should a client report retain?

Retain the question, evidence definition, observation, open interpretation and next decision. That concise record supports continuity across agency colleagues without unnecessary technical identifiers.