Google Ads client reporting for consultants: deliver an understandable account of results and next decisions

Start with the client decision

A client report should help somebody decide what happens next. A consultant may be asked whether a new offer warrants more budget, whether a lead programme is producing usable enquiries or whether an account change deserves more observation. These are different questions. A familiar set of charts becomes useful only after the report establishes which question it addresses.

Write that question before selecting metrics. For an appointment business, the immediate issue might be the number of enquiries the team can handle, with cost per primary conversion as advertising context. For a store, it might be whether advertising expenditure and recorded purchase value have moved together. Neither question can be answered fully by a generic statement that clicks increased.

Keep the consulting remit visible. A diagnostic engagement can document an account's measurement and recommend a follow-up investigation. A management engagement can also propose a supported live change for owner approval. The report should not imply that recommendations were implemented simply because the consultant has explained them convincingly.

The report's opening can be short: the selected period, the principal observed result, the important limitation and the decision now available. Detail then supports those statements. This gives the client a clear entry point while retaining enough evidence for the person responsible for advertising to inspect the reasoning.

Serve the people receiving the report

A finance lead, an operations manager and the person editing campaigns can need different explanations of the same result. The finance lead may need the spend and recorded value relationship. Operations may need the scale and timing of incoming enquiries. The account practitioner may need campaign contributions and conversion-action definitions. Start with the common decision, then provide the relevant supporting detail.

Use names the audience recognises. A campaign name can identify a business initiative without exposing an account identifier. If a finding belongs to an ad group, include its parent campaign. Explain an unfamiliar measurement once, close to its first use, rather than assuming that every stakeholder understands the Ads and Analytics reporting vocabularies.

Agree who receives the recurring report. A channel containing the account owner and the consulting contact can support a focused discussion, but a larger channel may include people who cannot approve a live change. Receiving an observation is distinct from having the authority to confirm a proposal. Keep that distinction out of the performance interpretation and explicit at the action handoff.

Do not make every audience read every breakdown. A concise observation with one appropriate requested chart may be enough for a meeting. The consultant can answer a later question with a selected table or a deeper breakdown. The supporting evidence must remain coherent across those views; simplification should remove clutter rather than silently change the measurement.

Define the measurement agreement

Before reporting begins, define the account, source, dates, included population and result measurement. Those choices form the practical agreement behind the numbers. A report on all primary Ads conversions has a different meaning from a report on one named contact action. A report on total Analytics purchase revenue has a different scope from revenue attributed to the linked advertising account.

Record which business interpretation belongs to each metric. A conversion count may include several actions that the client has chosen to treat as important. Purchase-only measurements select purchase-category actions. A lead report groups lead types, while a named action request selects that action rather than the whole lead population. These definitions should survive a change from a table to a chart.

Agree how exceptions will be described. If access fails, say that the required evidence was unavailable. If an input has no usable denominator, preserve an undefined ratio. If the account has no matching rows in a complete supported read, distinguish that result from an incomplete query. A client should never have to infer these states from a dash in a table.

Review the agreement when the business changes its measurement. Adding a primary action, changing the promoted offer or selecting a different Analytics event can affect comparability. Explain the change beside the result. Keeping an old metric label while changing the population makes a tidy report difficult to trust.

Name the reporting dates

Put the actual start and end dates in the report. A request for the last thirty days should use the intended rolling period, rather than silently becoming the previous calendar month. A request for last month should preserve that calendar boundary. The relative phrase is convenient in conversation; the resolved dates establish what the returned figures cover.

For configured completed-period reports, the current month runs through yesterday rather than including today's unfinished activity. On the first day of a new month, that selection can have no completed day. An empty current-month period is not evidence of lost traffic. Explain the boundary and choose another period only when the client asks for a different question.

The delivery schedule is a separate setting. A report sent on Monday does not necessarily cover the preceding calendar week, and a monthly delivery does not necessarily select last month. Choose the data range and comparison independently from the time and frequency at which the message should arrive.

Use the relevant time zone consistently. If a client compares a report with another system, establish the dates and day boundaries in both before interpreting the difference. A late-night event can fall on different local dates. The consultant should resolve the period rather than attribute every mismatch to a measurement fault.

Choose a useful comparison

A comparison needs a reason. An immediately preceding period can show recent movement, while a calendar comparison can fit a monthly business review. A promotional week may need an explanation of the offer and trading conditions before any percentage comparison is useful. Choose the supported comparison that answers the agreed question, then name its dates.

Keep the selected population consistent. Comparing all campaigns in one period with currently enabled campaigns in another changes more than time. Historical cost from a campaign now paused can still be part of account activity. If the client asks for a current enabled subset, label it as that subset rather than presenting it as the complete historical account.

Check the reference value before stating a percentage change. An increase from a zero denominator cannot be described as an ordinary relative percentage. State the current and prior amounts and explain why a percentage is undefined. A new offer with no prior conversion value may still have a useful absolute result.

A comparison does not isolate the consultant's effect. Budget, demand, offer availability, measurement and the website can all change between periods. Describe the observed movement and the relevant known changes. A causal claim needs additional evidence; it should not appear automatically because the report has a before-and-after chart.

Separate a completed day from settled data

Yesterday is a completed calendar day, but its reported outcomes can still change. This matters when a client expects every newly delivered report to be the final accounting of that day. Name the observation period and avoid calling the most recent values permanently settled simply because the period ended.

Google's data-freshness guidance explains that availability differs by metric and that later adjustments can update reported statistics. A consultant should therefore distinguish a reporting cutoff from a guarantee that every associated outcome has arrived.

Use this distinction proportionately. A current operational review can still be useful for understanding recent spend and incoming results. A larger investment decision may need a follow-up after more outcome evidence arrives. State which decision the current observation supports instead of either treating fresh data as final or declaring it useless.

If a number changes in a later report, compare the same source, dates and definition first. An updated provider result differs from an accidental change of scope. The client needs to know whether the earlier report used a different measurement or whether the source revised its observation. Preserve that distinction in the explanation and any correction.

Explain what counts as a conversion

In the conversational business measurement, total Ads conversions include primary actions by default and exclude secondary actions unless explicitly requested. That is the intended population. If several primary actions record results, their reported counts contribute to the total. The tool does not reinterpret that selection as one universal type of customer outcome.

Ask for the action list when the client needs to understand the total. Showing the selected actions is more useful than guessing that every conversion is a sale. Call-related actions, lead categories and offline actions can have different business meanings even when the account classifies them as primary. Explain the selected group beside the count.

Lead reporting treats lead types as one lead group. When comparing that group with Analytics, the corresponding event must be selected. The AI can show observed event names and suggest a likely candidate, but a suggestion is not the client's confirmation. A site's enquiry event may have a specific name that cannot be inferred from a generic label.

A named action report answers a narrower question. For example, comparing the Ads action Contact Enquiry with the selected Analytics event enquiry_submit preserves each source's identity. Do not substitute all conversions because it is easier to retrieve. If the requested action is secondary, make the explicit inclusion part of the report definition.

Keep value and purchases clear

Total primary conversion value can include value from primary actions outside the purchase category. If a client assigns value to an enquiry or call, that value can contribute to the total. It is therefore inappropriate to describe every total conversion-value result as confirmed purchase revenue.

Purchase value selects primary purchase-category actions. Use that measurement when the question specifically concerns purchases, and preserve its count alongside value when average purchase value is needed. The selected action population remains important: multiple primary purchase actions contribute according to the data returned for those actions.

The saved Ads reporting catalogue labels its conversion-value measure Revenue. A consultant should clarify that catalogue definition when discussing it with the client. A familiar label does not change the underlying measure into a purchase-only number. If a different measurement is required, request that supported selection explicitly through the conversational reporting framework.

Keep profitability separate. Recorded conversion value divided by advertising cost gives a return ratio for that definition; it does not establish margin, fulfilment costs or net profit. The client may use those additional business inputs in a broader decision, but the advertising report should identify which figures it actually contains.

Compare Ads and Analytics deliberately

A client may want Ads conversion value and Analytics purchase revenue on one requested chart. The two series can support a useful comparison when their meanings and populations are labelled. They are separate source observations. Do not add them together as though they were two independent pools of revenue.

Preserve the requested traffic scope. Total Analytics purchase revenue includes the selected all-traffic scope; it should not become Google-only revenue merely because the other series comes from Ads. If the client wants revenue attributed to the linked Ads account, select that scope explicitly. The answer should make the chosen population understandable.

For total conversions against Analytics key events, identify the corresponding aggregate definitions. For leads, calls or a named conversion, obtain the chosen Analytics event. A phone-click event is not proof of a completed call. Comparing the two can answer a tracking question, but the report must preserve what each source measured.

Distinguish a supported measurement definition from an available delivery capability. The current release controls still restrict named Analytics comparisons that lack the required verification evidence. If a requested path is unavailable, explain that exact capability boundary. Do not call a delivery restriction a universal inability of the provider to return event information.

Calculate ratios from their totals

Calculate a period return ratio from the period's selected value and cost. Averaging daily return ratios can distort the result because days have different spend. The same principle applies to cost per result: total selected cost divided by total selected results describes the period, while a simple average of daily ratios answers a different question.

When total Analytics purchase revenue is divided by advertising cost, describe the measure as blended ROAS or MER. Specify whether the denominator is Google Ads spend or the available advertising cost in Analytics. That choice matters when a business buys traffic through several channels; neither denominator should be inferred invisibly.

Undefined ratios need an explanation rather than an invented zero. Cost per result with no results cannot be treated as a free acquisition. Return on spend with no usable cost denominator cannot be represented as a normal return multiple. A table can show the available numerator and denominator while retaining the undefined ratio.

For average purchase value, keep the purchase selection consistent between value and count. Comparing the Ads purchase measurement with Analytics average purchase revenue is meaningful only when the reader understands the respective scopes. Round client-facing numbers to no more than two decimals without using rounded daily figures as the calculation inputs.

Preserve coverage and missing values

The report should identify what was successfully retrieved. A returned preview is not necessarily the full population, and the number of examples displayed is not the complete count. Count requests use the framework's complete-count path. A consultant should not count ten displayed campaigns and report that the account has exactly ten.

Keep unavailable observations separate from genuine zeros. If a required provider read fails, the missing figure cannot demonstrate no conversions, no revenue or no spend. A complete supported additive read can establish a zero where its coverage permits it. Those states need different wording and different follow-up actions.

For a combined report, explain which source is available when the other is not. Do not fabricate a second series from the successful source. If the client asked for a comparison, one source alone does not satisfy it; show the available evidence and identify the missing dependency directly.

Incomplete coverage also limits recommendations. A campaign ranking cannot be described as the worst account performer if important eligible campaigns were unread. Obtain the necessary coverage or narrow the conclusion to the successfully examined subset. A clear boundary preserves usefulness without overstating confidence.

Distinguish account results from selected groups

An account report and a campaign report can both be correct while showing different totals. Name the scope before interpreting the difference. An account-level observation includes the selected account activity; a campaign observation narrows that population. A consultant should not use the narrower result as evidence that the remaining account had no activity.

Separate historical reporting from current configuration. A campaign that is paused now may have spent during the reported dates. A currently enabled campaign may have been enabled only recently. When the client asks about performance over a historical period, preserve the intended historical population instead of quietly filtering by today's status.

Use hierarchical context for deeper findings. A keyword belongs to an ad group and campaign. A product listing occurrence belongs to its relevant group and campaign. The same product can appear in several groups, so a targeting-occurrence count does not equal the number of unique inventory products. The report should name the unit of analysis.

Keep commercial aggregation distinct from operational aggregation. Several finding rows can describe one underlying website dependency, while one campaign can contain many affected resources. Summing findings can help organise work but cannot establish lost sales. Use the performance dataset for performance conclusions and the relevant settings or findings dataset for configuration conclusions.

Read an invented consulting report

Consider an invented business, Alder Lane Learning. Its consultant compares two complete selected periods using the same primary conversion-value definition. The earlier period contains cost of 1,200 and primary value of 4,800. The later period contains cost of 1,500 and primary value of 5,250. These figures illustrate reporting choices; they are not customer results.

The later period has 25% more cost and about 9.38% more primary value. Its aggregate return ratio is 3.5, compared with 4 previously. A useful opening states that value increased while the selected value-to-cost ratio decreased. Saying only that revenue grew would hide a relevant part of the investment decision.

The client then asks whether the additional value represents course purchases. The consultant inspects the action definition rather than assuming it does. If primary enquiry value also contributes, the report explains that population and requests purchase-only evidence for the new question. The original total remains valid under its original definition.

A campaign breakdown shows which initiatives contributed to the change, but it does not prove that the consultant's edits caused it. The handoff identifies the observed contribution, the known promotion and the next review question. A proposed budget adjustment remains a separate owner-approved action, with the exact target and value shown before execution.

Write the explanation around the evidence

Start the narrative with what the data establishes. Name the period, scope and principal movement in ordinary language. Then explain the business relevance and the important limitation. A paragraph that says spend increased while primary value grew more slowly is easier to assess than a wall of percentages without a measurement definition.

Use a short distinction between observation and interpretation. The returned campaign contributions are observations. A suggestion that the offer may need review is an interpretation to investigate. A recommendation to pause a campaign is a decision proposal. Keeping these roles clear lets the client challenge the interpretation without losing the underlying evidence.

Do not turn every movement into a crisis or success. An increased cost per lead can coincide with a deliberate change in lead quality, a new market or delayed measurement. Those possibilities require actual business context. Ask the relevant follow-up rather than filling the narrative with unsupported causes.

Make the next step concrete. If the key uncertainty is the selected conversion population, inspect the action list. If it is a campaign contribution, retrieve that breakdown. If the website changed, assign a journey check. Each step should resolve an identified question rather than merely produce more reporting material.

Correct an earlier explanation explicitly when its definition was wrong. Preserve the valid figures, name the mistaken interpretation and state the revised conclusion. Replacing an earlier message with an unrelated chart can leave the client with both versions and no reason to prefer one.

Use requested visuals purposefully

A requested chart should answer the client's visual question. Use a line for change over time when requested, or a bar chart for the requested category comparison. The framework renders returned provider data through its data reference; the AI does not supply invented numerical arrays. The explanation and chart should describe the same selection.

Compatible series can share one native chart. Cost and value in the selected money unit can be compared with distinct labels. Counts and money need a different presentation unless a supported representation preserves their units. Up to three selected series keep the native output readable. A fourth request needs a narrower selection rather than a crowded chart.

Native chart point limits also affect the breakdown. When the client has not explicitly required daily detail, the framework can choose weekly or a coarser readable period. State that breakdown. When daily detail is specifically required, preserve it and explain any necessary split; do not silently change a daily request into a weekly chart.

The current delivery uses at most two chart panels and twenty points per series per panel. A longer explicitly detailed request can require a narrower interval, a different grain or a supported table. Keep the response native; an image is not the fallback for exceeding this limit.

Current-year date labels omit the year, while older dates retain it where needed. Weekly labels should convey the actual date interval. Client-facing numbers use no more than two decimal places. These choices make the chart readable without removing the report's actual dates and measurement definitions.

Provide detail without overwhelming the meeting

Decide which detail belongs in the initial explanation and which belongs in a follow-up. A campaign total and one supporting contribution may answer the current decision. A complete table can answer a later request for all selected campaigns. The short answer should identify its scope rather than imply that the displayed examples are the entire result.

Use a requested table when the client needs exact values across categories. Columns should have understandable labels and units. Internal resource identity is useful for joins and exact targeting behind the scenes, but it need not become a client-facing column. An action proposal can name the asset and its actual parents without displaying raw identifiers.

The current verified delivery supports one table rather than an unrestricted collection of simultaneous tables. If the client requests several separate tabular views, resolve which view answers the immediate question or prepare a supported combined selection. Do not claim that a capability exists merely because the model can describe an attractive layout.

Offer a file or another report surface only when the relevant tool supports that output. Some research datasets have supported export paths, but that does not establish a universal export for every advertising question. The consultant should choose a real available delivery method and preserve the same metric selection across it.

A meeting appendix should remain navigable. Group related measures, explain the conversion population and retain the dates. More rows are useful when they let the client inspect a decision; they are unhelpful when they bury a simple question under every field the source can return.

Handle client questions in Slack

A client can ask for a result in normal language: how did the selected primary value and spend change over the last thirty days? The reporting framework discovers the appropriate registered measurements, validates the requested plan and reads the required source data. The answer should state the resolved period and definition, not expose the internal tool vocabulary.

A follow-up such as show that weekly should preserve the prior scope and selected series while changing the requested breakdown. A follow-up such as only the training campaign changes the population. Maintaining that distinction prevents a readable chart from silently answering a different question than the preceding conversation.

Creative discussion does not require a performance call. If the client asks for wording to explain the result in a board meeting, the AI can work from the established conversation. If the client asks whether the current account value has changed, a current read is appropriate. The need for evidence follows the question rather than a rule that every sentence must trigger a provider request.

Ask a focused clarification when the metric is genuinely ambiguous. For an Analytics comparison of leads, ask which event corresponds to the intended enquiry outcome and offer observed names when available. For a blended return ratio, ask which spend source the client wants. Avoid repeating a broad request for context when the thread already contains the needed account and period.

Keep the response ordered for the reader. The explanation and requested assets belong together, with the conversational user mention before the feedback footer. Progress on an approved edit is a separate status update. Neither feedback controls nor execution messages should interrupt the explanation of the performance evidence.

Configure recurring reports explicitly

A recurring report requires a saved configuration. Select the source, account or campaign target, metrics, data range, comparison, time zone and delivery schedule. A conversation about this month's results does not by itself create a recurring obligation. The client should be able to inspect the saved choices before relying on the report.

Choose a delivery time that fits the decision process. A regular meeting can benefit from a report arriving beforehand, but the consultant should still explain recent-data limitations. The schedule does not make the underlying source refresh instantly, and it does not convert the chosen completed period into a live dashboard.

The data range and comparison remain independent settings. A weekly schedule can deliver a rolling period; a monthly schedule can deliver another supported range. When a report appears to cover the wrong dates, inspect these saved selections rather than assuming the delivery frequency determines the observation period.

Revisit the configuration when the engagement changes. A new campaign focus, recipient channel or measurement definition can make an old recurring report less useful. Explain what changed and why. A paused or disabled configuration should not be described as an active reporting commitment.

Distinguish queued work from delivery

A manual run request enters the reporting workflow; it is not proof that the final report has reached Slack. Keep accepted, queued, completed and delivered states distinct. The client needs an immediate acknowledgement when work starts, followed by the actual outcome when it is available.

Saved configuration revisions protect the selected report definition as work is dispatched. A stale task should not be interpreted as current work merely because it exists in a queue. The reporting process checks whether the configuration remains eligible, while the run record preserves the occurrence and selected observation dates.

After an interruption, the scheduler collapses an outage backlog to the newest missed occurrence and records older skipped occurrences. That behaviour avoids flooding the channel with every delayed report. It does not mean the skipped periods were individually delivered. Describe the actual completed occurrence when explaining a later report.

If delivery fails, preserve the available run evidence and state the failure. A provider-read failure and a Slack-send failure need different remedies. Repeating the data query is unnecessary when the original result is valid and the problem is solely delivery. Conversely, a successful send cannot repair missing provider evidence.

Turn recommendations into separate decisions

A report can recommend an investigation or a supported account change. These recommendations should be visually and logically distinct from the results. The client needs to know which facts were observed, which explanation is proposed and which live operation would follow approval.

For a supported edit, resolve the exact target, its relevant parent context and the intended setting. Show the individual action steps plainly in the confirmation proposal. A suggestion to review a budget is not permission to change it, and a good performance explanation does not bypass the connected owner's confirmation.

The confirmation proposal states that no changes have been made yet and expires after fifteen minutes. When approved, relevant settings are checked again before execution. A change in current state can invalidate the preparation. Explain that result rather than applying an old proposal to a newly changed target.

Execution has its own evidence. A verified stored setting does not prove a later improvement in performance. An acknowledged operation without complete verification remains a different outcome, and an uncertain dispatched write must not be blindly retried. Keep the resulting account task and the next performance review separate.

Link the decision to budget tracking, wasted-spend review or an audit when those workflows answer the remaining question. They provide complementary evidence, not automatic proof that a particular change will improve the account.

Assign actions and review dates

Every agreed next step needs an owner. The consultant may own the account analysis, the client may own offer availability and a website specialist may own the customer journey. Naming the responsible person prevents the report from ending with a recommendation that nobody can act on.

State what completion means. Inspecting the action list, correcting a selected URL and verifying an enabled campaign are different tasks. A vague instruction to improve performance cannot establish whether the work happened. Use an observable outcome and retain any remaining dependency.

Choose a review date that fits the task. A configuration change can be checked immediately when the supported readback is available. Its commercial effect requires later performance evidence. A lead-quality question may also need the client's operational feedback. The same deadline is not appropriate for all three milestones.

Track decisions the client declines as well as those accepted. A deliberate choice to retain a campaign can be reasonable under the client's business context. Record the rationale without presenting it as the consultant's verified prediction. Later reporting can revisit the relevant evidence under the same agreed definition.

Reuse the right evidence

The reporting framework returns a data reference that a requested visual can reuse. A text answer, table selection and chart should not each trigger the same provider read merely because they are different presentations. Compatible results can retain their totals and definitions while changing their delivery format.

Field and relationship discovery can use the local catalogue. Learning which measurements and datasets are supported does not require scanning the client's full account. Actual action metadata or observed Analytics event names still need their relevant reads when the question depends on them.

Do not confuse reuse with current-state verification. If the client asks whether a newly changed setting is now present, an earlier performance dataset cannot answer it. If a new period or population is requested, retrieve the necessary selection. Reuse removes duplicate work while preserving the evidence demanded by the question.

Plan the breakdown before requesting data. A single coherent dataset can often support the total explanation and requested trend. Repeated speculative calls for unrelated measures add latency and can obscure the result. The consultant should narrow the question enough to obtain useful evidence in a deliberate sequence.

Leave a usable engagement handoff

At the end of an engagement, leave the report definition, latest actual result, unresolved questions and action owners together. A successor should be able to understand the measurement without reading months of Slack conversation. Names, source, dates and selected populations matter more than a long archive of screenshots.

Identify saved configurations that remain active. State their chosen schedule and data range, who receives them and who now owns follow-up. An automated report without an accountable reader can continue producing messages while the underlying business question has changed.

Preserve uncertainty honestly. An unread source, unconfirmed event selection or unresolved dispatched write needs a concrete reconciliation task. Do not close these items merely because the consulting contract ended. The handoff should make the next responsible action clear.

Use reporting measurement guidance for the technical definitions and monitoring for ongoing exception handling. Client reporting brings those observations to a business decision while keeping their original evidential boundaries intact.

Answer common client-reporting questions

Does a monthly delivery automatically mean last month?

No. Frequency and data range are separate saved choices. Name the actual selected dates in the report.

Does Revenue always mean purchases?

No. The saved Ads catalogue uses conversion value for that label. Purchase-only questions need the purchase selection; total primary value can include other primary actions.

Can a client request a combined native chart?

Yes, when the selected measurements and units support that presentation. Preserve separate source labels, the requested scope and the native series and point limits.

Does a queued report prove delivery?

No. It establishes accepted work. The completed result and delivery outcome are separate evidence.

Does a recommendation execute an account change?

No. Supported live edits require an exact proposal and connected-owner confirmation. Later performance remains a separate observation.