Google Ads reporting for freelancers: make the numbers useful to a client

Define the report's client decision

A freelance account manager needs reporting that helps a client understand the work and decide what comes next. Sending a large collection of metrics is not the same as making the account's recent activity understandable.

Begin with the report's purpose. A weekly review may ask whether spending and recorded outcomes moved in the intended direction. A promotion review may ask how a selected campaign performed over its actual dates. A measurement review may need named conversions rather than an account aggregate.

The purpose determines the scope and measures. It also determines what the report cannot conclude. A cost and conversion summary can describe recorded performance without establishing margin, lead quality or the client's final financial return.

For a freelancer, repeatability matters because the same person often collects evidence, interprets it and writes the client explanation. A settled definition avoids rebuilding the question every time and makes later comparisons easier to defend.

Keep the report useful when no live change is needed. It can confirm a stable observation, document a business question or establish that more evidence is required. A reporting service should not manufacture an edit to make the delivery appear productive.

Agree the reporting service

Define what the client will receive and how it will be reviewed. A scheduled account summary and a consultant's interpretation are different parts of the service. The client should know whether the delivery is an automated observation or a reviewed recommendation.

Agree the account, intended outcomes, recurring period and receiving destination. If the client wants an explanation of qualified sales rather than form submissions, establish how that outcome is available before designing the report around it.

Clarify which business facts remain the client's responsibility. Stock changes, offer amendments and sales follow-up may not appear in the reporting data. The consultant needs those facts to interpret movements accurately.

A recurring report does not imply an assurance that every statistic is final at delivery time. Provider processing and later adjustments can affect recent periods. Use wording that describes available reported evidence rather than settled accounting.

Agree how decisions will be approved. A report can recommend a budget change while the live action remains pending. The reporting arrangement does not silently grant authority for every future mutation.

Keep the agreement concise enough to revisit. A named purpose, defined measures and review process are more useful than a long list of every possible field. Update those choices when the client's objective changes.

Keep account and campaign scope clear

An account report describes its selected account population. A campaign report describes the chosen campaign. They should have visibly different scope even when the same metrics and dates are used.

The current scheduled Ads reporting workflow supports an account or campaign target. Select the appropriate target when configuring it. Do not use a campaign's results as the client's account total simply because that campaign currently supplies most of the activity.

For ad hoc investigation, ppctor's reporting framework can work with more specific supported dimensions and conditions. That flexibility does not mean every such conversational selection is a saved scheduled-report option.

Keep parent context when narrowing to assets. A keyword belongs to an ad group and campaign; a Performance Max product selection belongs to an asset group. Similar names in different parents can represent different targets.

When a client asks for a change in scope, say what changed. An account overview followed by a brand-campaign view is a useful progression. It should not be presented as two conflicting answers to the same question.

Use the channel's connected account deliberately. A freelancer managing several clients needs confident account separation, not a response assembled from whichever account was discussed most recently elsewhere.

Choose completed dates deliberately

The current scheduled ranges are yesterday, the last seven days, the last fourteen days, the last thirty days, this month through yesterday and last month. Each resolves a defined completed period rather than a vague reference to recent activity.

Rolling ranges end yesterday in the relevant reporting-day context. Last thirty days is not a synonym for last calendar month. On the same run date, those selections can have different starting and ending dates.

This month through yesterday excludes the current partial day. On the month's first day there is no completed day inside the new month. An empty current-month period should not be filled with last month's results to make the report look complete.

Inspect the printed dates rather than rely only on the range label. They help the client distinguish a rolling review from a calendar review and make it possible to reproduce the observation later.

For Ads, use the account's reporting time-zone context. For an Analytics report, the data period uses the GA4 property's reporting day. A consultant's local clock should not silently replace either source's boundaries.

If the client requests a custom period conversationally, preserve that explicit selection. Do not force it into a saved range whose dates answer another question. A custom investigation and a recurring configuration can coexist.

Separate delivery schedule from data period

Frequency tells the system when a report is due. Range tells it what data to include. A weekly delivery can report the last thirty completed days; a daily delivery can report yesterday. Those settings should be chosen separately.

The saved reporting workflow offers daily, weekly and monthly frequency. Configure the relevant time and day selections for the client routine. The cadence should fit the intended review rather than become an arbitrary default.

The configured time is a delivery schedule, not a promise that every upstream statistic becomes final then. Queueing, evaluation and Slack delivery are separate stages. A run request must not be described as completed delivery before that result exists.

The manual Run Now path queues an enabled saved configuration. It can be useful for checking a report without waiting for the next scheduled occurrence. It does not mean a paused report silently becomes an active recurring one.

For Analytics reports, delivery cadence still uses the linked Ads account time-zone arrangement, while data dates use the property reporting day. Keep that distinction in the configuration record when the two differ.

Choose delivery times around your review process. A client may prefer a predictable morning summary, while you need time to investigate a follow-up. Scheduled output should support that relationship rather than imply constant personal attendance.

Choose a useful comparison

A comparison gives context only when the periods and measure definitions are appropriate. The saved choices are no comparison, previous period and the same period last year. Choose the one that helps the actual client question.

For rolling completed ranges, the previous period is the immediately preceding interval of the corresponding length. A calendar last-month report compares with a preceding interval resolved by the shared period logic; inspect actual dates before describing the result.

Current-month-through-yesterday comparisons use the corresponding portion of the previous month. If the previous month is shorter, the boundary can be clamped. State the resulting dates rather than imply equal-length periods when that is not true.

Year-over-year context can be useful, but it does not control for every business change. Different offers, stock, measurement definitions and campaign structures can affect the interpretation. Treat it as a comparison, not an explanation.

A percentage change needs a meaningful prior value. If the previous figure is zero, do not present an ordinary percentage as if the arithmetic were defined. An absolute difference can be more honest and easier to understand.

Keep the chosen conversion population stable across the comparison. Changing from all primary conversions to one purchase action can be a valid new report, but it should not be represented as continuation of an unchanged series.

Select measures the client can interpret

Choose a small set that answers the report's purpose. Cost, a defined outcome count and the corresponding value or efficiency measure often form a coherent view. Additional metrics should earn their place by supporting the decision.

Clicks and impressions describe advertising activity, not completed business outcomes. They can help explain changes in delivery. A client report should not imply that more of either automatically means a better commercial result.

Search impression share and its loss measures answer availability questions for the applicable Search scope. Their availability and meaning differ from a simple account total. Avoid stretching a Search-specific measure into a statement about every campaign type.

Check the saved catalogue's definitions before interpreting its labels. Scheduled reporting uses its declared provider measures. A conversational business report may apply an explicit conversion population that the generic saved metric does not.

Use clear units in display and explanation. Cost belongs in currency, rates in percentages and ROAS as a multiple or clearly expressed ratio. Values should not appear with needless decimal precision that obscures the comparison.

Remove a metric if you cannot explain its relevance to the client question. A report becomes more useful when every column supports interpretation. Technical breadth is not a substitute for a clear outcome definition.

Define the conversion population

Google's primary and secondary conversion guidance distinguishes the roles and explains that custom goals can affect reporting and bidding behaviour. Read the configured definitions rather than assume every conversion-labelled number has identical membership.

In ppctor's conversational business measurements, the default primary aggregate includes the selected primary actions and excludes secondary actions. Purchase and lead aggregates use their declared categories. A named-action request selects the specific action after its identity is resolved.

All lead categories are treated as leads for the primary lead aggregate rather than presented as an invented hierarchy. If the client needs a specific contact-form count, ask for that action by name instead of using the broader lead total.

Calls and offline conversions can also be available in supported account reporting. They may overlap with broader aggregates. Do not add overlapping totals together and claim a larger unique number of outcomes.

Primary actions are summed according to the chosen definition. The report does not repair a client's duplicated measurement implementation. If the client configured multiple primary actions recording the same underlying purchase, the aggregate reflects that configuration.

Keep the distinction between those explicit conversational measurements and a saved provider conversion metric. A more specific follow-up is valuable, but its different population must be visible in the explanation.

Keep value and revenue labels accurate

Value can be assigned to conversion actions that are not purchases. An all-primary conversion-value measure can therefore include calls, lead actions or another configured outcome with a value. It should not silently become product purchase revenue.

A purchase-value measure narrows to the selected purchase population. A named-action value narrows further. The client needs the definition beside the result, especially when several primary actions contribute to a broad account value.

The saved reporting catalogue labels its declared conversion-value metric as Revenue. When interpreting that report, explain the underlying provider definition and check whether non-purchase values contribute. For an explicitly purchase-only question, use the corresponding supported business measurement.

GA4 purchase revenue belongs to its own source and selected traffic scope. All-traffic purchase revenue is not the same population as Google Ads-attributed purchase value. They can be compared as separately labelled observations without being added together.

A change in reported value can reflect volume, assigned values, attribution or measurement configuration. Do not immediately translate it into a claim about the client's profit. Margin and final financial reconciliation need additional business evidence.

Ask the client about the meaning of assigned values when necessary. A lead valued for bidding is not automatically a realised sale. The report should preserve that distinction even when the arithmetic uses the configured values correctly.

Calculate period ratios from their inputs

A period ROAS calculation uses the selected aggregate value divided by the selected aggregate cost. It is not an average of daily ROAS values. Daily ratios with very different spend can produce a misleading average.

The same principle applies to cost per conversion: period cost divided by the period's selected conversion count. If the count is zero, the ratio is undefined. A displayed zero would misrepresent the measured efficiency.

For purchase ROAS, use purchase value with the agreed cost scope. For an all-primary value ratio, disclose the broader numerator. Similar labels should not hide different business questions.

Blended ROAS or MER uses the agreed all-traffic revenue and advertising-cost scope. It can be useful for a client considering the site's wider result, but it is not a replacement for an Ads-only performance question.

Average order value also needs a compatible purchase definition. Purchase value divided by the selected purchase count and GA4 average purchase revenue can be compared under a defined scope. Do not mix a broad conversion count into that purchase calculation.

When a chart is grouped by week or month, aggregate the inputs before deriving the ratio at that grain. The goal is a meaningful observation per period, not a more attractive line made from arithmetic that changed the question.

Use GA4 with explicit scope

A connected GA4 property can add a site-oriented view alongside Ads reporting. It needs the appropriate account connection and supported metric selection. The freelancer should establish that context before interpreting the resulting comparison.

Specify whether the GA4 observation concerns all traffic or a supported selected segment. All-traffic purchase revenue can include outcomes outside Google Ads. A Google Ads traffic selection answers a narrower question.

Session counts and outcome counts are different measures. Do not present more sessions as proof of more purchases or leads. Use the client-selected outcome alongside traffic when the report's purpose requires both.

For a lead comparison, confirm the site's corresponding GA4 event. The assistant can inspect available observed event names and suggest plausible candidates, but the client's event meaning must be settled before treating it as the lead counterpart.

Compatible Ads and GA4 measurements can share a chart when their units and intended business meaning fit. Preserve separate source labels and attribution context. Shared presentation does not imply the values should match exactly or be summed.

If the property is unavailable or the requested field is unsupported for the scope, say so. A missing observation should not become zero or a fabricated series. That limitation is part of the report's evidence.

Explain an invented reporting example

The following numbers are invented for illustration. A freelance consultant reviews two comparable completed periods for Cedar Supplies, a fictional retailer. The chosen outcomes are primary purchase conversions and their purchase value, with account cost as the denominator.

In the earlier period, cost is 1,000, purchases are 50 and purchase value is 4,000 in the relevant currency. In the later period, cost is 1,200, purchases are 60 and purchase value is 4,200. These are example aggregates, not customer results.

Cost and purchases both rise by twenty percent. Purchase value rises by five percent. Cost per purchase remains twenty, while purchase ROAS falls from four to three and a half because value grows more slowly than cost.

Average purchase value falls from eighty to seventy. That is an observation derived from the chosen inputs. It does not independently establish why baskets changed, whether margin improved or whether the campaign caused the change.

A useful client explanation says that more purchases were recorded at the same average acquisition cost, while lower average purchase value reduced the value-to-cost ratio. The next question is whether the client's product and promotional context explains the basket difference.

The consultant does not recommend a pause simply because ROAS fell. The client may have intentionally promoted lower-priced products. Review the commercial facts before choosing an action, and retain the defined purchase population for the later follow-up.

Keep the arithmetic and interpretation separate

The calculations are reproducible from the reported inputs. The commercial explanation is not established until the client supplies or confirms relevant facts. That distinction lets the consultant give a clear account of the results without inventing a diagnosis.

If a separate all-primary value includes lead actions, it is a different numerator. Do not quietly substitute it to make the ratio look stronger. A changed definition requires a separately labelled observation.

Disclose unavailable or limited evidence

A report must distinguish a complete aggregate from a displayed selection. A table showing the largest campaign contributors is a preview unless it represents the full requested population. Do not count its visible rows as the account's total campaign count.

The conversational framework tracks result coverage and supported completion. An incomplete or capped result should not be used to manufacture a complete chart or exact total. Ask a narrower question or use the appropriate complete aggregate when available.

Missing values and zero values are different. An unavailable GA4 metric is not evidence of no purchases. A failed Ads read is not a low-spend observation. Preserve the reason the result could not be supplied.

Separate source-level failures when several observations are requested. One source may return usable data while another fails. The successful result should remain identifiable, and the failed source should not be quietly omitted from the explanation.

A current status view is also different from a historical outcome report. A paused campaign can have activity in the selected past period. Current status should not be used to erase that historical evidence.

Disclose limitations in language useful to the client. Say which part of the question remains unanswered and what would resolve it. Internal provider identifiers and debugging output rarely help a client make the next decision.

Write an explanation without inventing causes

A strong explanation begins with the observed change and the definition behind it. State what rose or fell, over which periods and for which scope. Then distinguish confirmed context from an investigative question.

Do not infer a promotion, tracking failure or change in customer intent merely because a curve moved. Those are possible investigative subjects, but they need evidence. The consultant can ask the client about relevant business events without asserting that they occurred.

Use the reported inputs to explain ratio movement. If cost rose faster than purchase value, that arithmetic explains the lower ratio. It does not establish why either input changed or whether the outcome meets the client's margin requirement.

Keep review language proportional to the data. A short period and a small outcome count may support a cautious observation, while a more substantial comparable interval can support a broader pattern description. Neither automatically establishes causation.

Separate reported outcomes from work completed. A verified settings change is an implementation result. A later performance movement is a measured observation. Presenting them side by side is useful, but their sequence alone does not prove the change caused the result.

End the explanation with the relevant decision or unresolved fact. A freelancer's value lies in making the next review useful, not in adding a confident-sounding cause that the account data did not establish.

Choose a useful table or chart

Choose the visual from the question. A table helps compare campaign contributions and inspect exact values. A line chart can show a sequence over time. A grouped bar chart can compare compatible measures across the same periods.

When requested conversationally, ppctor uses native Slack charts with up to three compatible series. Compatible source pairs retain distinct labels. Sharing one graphic does not justify combining unrelated units or summing different attribution populations.

A native chart supports at most twenty points per series. When daily detail was not explicitly requested, the framework can use a coarser weekly or monthly grain to preserve a readable single chart. The chosen breakdown should be stated.

When daily detail is explicitly required and exceeds one chart's capacity, the output explains the point limit and can split the series within the message's supported chart capacity. Longer requests may need a narrower period, coarser grain or table. They should not silently discard dates.

Use readable date labels, with the year retained when older-year context requires it. Weekly labels should identify the date range rather than expose an opaque year-week code. Displayed metrics should avoid more than two decimal places.

A missing series must not be filled with invented zeroes. If the returned evidence cannot support the graphic, a table or a clear limitation is preferable. The visual should make the data easier to understand while preserving its meaning.

Keep a chart request from becoming another data collection project

If the requested visual uses the same returned result, formatting can reuse it. A new period, dimension or outcome can require a new read. The consultant should not need repeated identical queries simply to improve presentation.

For a client who wants a simple answer, use the minimum useful display. A complete small table can be more readable than a dense chart with three series. The client's requested format should guide the choice.

Prepare a manageable solo review routine

A freelancer benefits from a reporting routine that can be repeated between other responsibilities. Keep a defined account question, selected measures and review sequence so each delivery starts from a settled basis.

First confirm the dates and scope. Then inspect the chosen outcome and its inputs. Identify the largest relevant contributions before writing the client explanation. This sequence keeps a striking percentage from outrunning the evidence.

Leave space for business context. Ask the client about known offer or availability changes when those facts could affect the decision. A report can be ready before the interpretation is complete.

Use recurring delivery for the stable question and conversation for the exception. If the weekly report exposes a movement in one campaign, a focused follow-up is more useful than permanently adding every possible dimension to the saved output.

Keep unresolved questions visible in your work record. A consultant switching between clients should be able to resume the investigation without rereading every chat message. Preserve the specific missing fact and the intended next check.

Review whether the routine saves useful time rather than only producing more output. A report that consistently leads to clearer decisions is more valuable than one that requires extensive explanation because its definitions remain ambiguous.

Configure scheduled reporting in Slack

The current command family is /ppctor reporting. Its management workflow lets you create and manage the connected account's saved reporting configurations. Use /ppctor reporting help for the current command options.

Choose the report source, target and supported metrics in the configuration view. Select the completed range, comparison and delivery frequency separately. Read the final selections as the question the report will answer.

A campaign target needs the intended named campaign. An Analytics target needs the connected property and supported segment context. Do not assume the modal can supply data for an unconnected property or a different client's account.

Use the management controls to edit, pause, resume or delete the saved configuration as appropriate. Pausing a report changes delivery behaviour; it does not pause campaigns. Deleting the schedule is not a live Google Ads edit.

Run Now queues an enabled configuration. Treat the acknowledgement as a queued request until evaluation and delivery complete. If the report is paused, decide whether to resume it rather than assume the command bypasses its enabled state.

For a broader introduction, see Google Ads reporting in Slack. Keep the saved report definition with the client agreement so later management changes have a clear reason.

Ask focused conversational follow-ups

A follow-up should preserve the report's context unless you deliberately change it. Name the dates, selected account and outcome when there could be ambiguity. This makes the answer reproducible and reduces accidental shifts in scope.

Ask for the additional evidence needed by the decision. “Which campaigns contributed most to this period's cost?” is a contribution question. “Compare primary purchase value with GA4 purchase revenue” is a cross-source measurement question that needs clear traffic scope.

The assistant can discover available fields and datasets before reading the account. The field library establishes supported choices, not the client's current values. Account questions still require appropriate returned evidence.

Conditions can narrow the investigation when supported. Express the comparison and its boundary clearly, including whether equality belongs in the filter. Boolean combinations should preserve the requested meaning rather than become a series of unrelated manual guesses.

If the conversation changes a named event or action, retain the rest of the settled context where appropriate. A correction of the numerator should not reset the period or selected campaign without instruction.

Ask for a list when you need to choose a named conversion or observed GA4 event. The assistant can present candidates, but the client-specific business counterpart still needs confirmation before it becomes a lead comparison.

Use creative help without unnecessary reads

Not every reporting conversation requires another API call. You can ask the assistant to make an explanation clearer, draft a client note or suggest a more useful heading from the evidence already returned.

Ordinary brainstorming also can remain conversational. Suggested headlines, descriptions or keyword ideas do not need live account reads unless you request actual research, existing settings or a measured account claim.

Keep creative suggestions labelled as suggestions. The assistant should not invent customer testimonials, business guarantees or product features to make the copy persuasive. Supply the factual constraints that the draft must respect.

When a rewrite introduces a new claim about performance, check whether the existing result supports it. Turning “cost rose” into “competition became more intense” adds a cause the original evidence may not contain.

Use a focused data request when a genuinely new fact is needed. A new date range or measure may require a read; a clearer sentence for the same observation does not. This distinction keeps the freelance workflow responsive.

For the general distinction between ideas, account evidence and live actions, see asking Google Ads questions in plain English. It helps set the right expectation for each kind of answer.

Keep the report separate from execution

A report can inform a proposed change, but it does not itself execute one. Establish the client instruction, supported operation and exact target before preparing a live-edit proposal.

The proposal should show the action and affected entity clearly, with parent names where needed. For several requested steps, review each logical operation rather than accept a compressed technical description whose effects are unclear.

The authorised connected owner must confirm in the proposal's context. A consultant's ability to ask reporting questions does not automatically grant authority to execute every account change. Arrange the appropriate handoff for the connection.

Current settings are checked again before execution. A proposal can become stale when another person changes the target. Resolve the new state instead of assuming the earlier report authorises overriding the intervening work.

A verified receipt establishes the checked implementation result. It does not establish a later performance improvement. Keep the chosen follow-up measure and period for the subsequent review.

If the result is uncertain, inspect the account before repeating a potentially duplicating action. The write may already have succeeded even when complete verification failed. The report and its recommendation should not trigger an automatic retry.

Retain the client decision

A useful report record includes the purpose, periods, chosen measures and the decision made from them. This is more informative than retaining only a screenshot of the largest number.

Keep confirmed business context beside the interpretation. If the client explains a promotion or stock change, distinguish that supplied fact from the account observation. A later reader should understand the basis for the conclusion.

Record when a live proposal remained pending or when no change was needed. Those are legitimate outcomes. Otherwise, a later review can wrongly assume a recommendation was implemented simply because it appeared in a report.

For executed work, retain the final status and the named target. Human-readable names and parents make the record understandable. Public or client-facing explanations should not expose unnecessary internal identifiers.

State the next review question when follow-up is required. “Review purchase value and cost over the next suitable completed interval” is more useful than “check performance later,” especially when another client interruption intervenes.

Keep changes in definitions visible over time. A report series that moves from all primary conversions to purchases needs a clear note. Without that context, historical comparisons can look continuous while measuring different things.

Make the report understandable during handover

A freelancer may need to hand work to another consultant during absence or at the end of an engagement. The report should be interpretable without access to every earlier conversation.

Provide the saved source, scope, period, measures and comparison choices. Include any business definition that was confirmed separately, such as the site's lead event. A list of metric labels alone may not explain the intended outcome.

Identify unresolved questions and pending proposals. The recipient should know what was investigated, what remains uncertain and which actions have actually been applied. A draft recommendation is not completed work.

Clarify reporting access and edit authority. A replacement reviewer can receive a summary without automatically becoming the connected actor authorised to execute a proposal. Do not imply a custom approval chain that the current workflow does not implement.

Review recurring delivery arrangements when the service changes. A scheduled report should not leave the client assuming that the former consultant continues a personal review commitment. Management of the saved configuration and the service agreement are separate tasks.

On return, inspect the current configuration and recent observations. Do not infer the whole reporting history from the latest message. Changes in source, range or conversion definition can affect its meaning.

Maintain the reporting configuration

Review the configuration when the client changes its goal, offer or account structure. A report can continue running successfully while answering a question that is no longer relevant.

Confirm that the selected target still represents the intended scope. A campaign rename or restructuring can make an old explanation confusing. Establish current identity before assuming a familiar name still means the same selection.

Review metric definitions when conversion configuration changes. Newly primary actions can affect broad aggregates. A specific purchase report may remain useful, but its scope should be checked rather than inferred from an older agreement.

Revisit the comparison after seasonal or commercial changes. A previous-period view can remain informative while requiring additional context. Do not replace the comparison simply to produce a more favourable percentage.

Adjust delivery timing deliberately when the review routine changes. A quieter schedule should have an agreed purpose, and a more frequent one should correspond to a useful decision. Frequency alone does not make the service more thorough.

Keep the report concise enough that the client can recognise its question. Add detail through focused follow-ups when needed. A stable, explicit definition is the foundation of useful recurring reporting.

Freelancer reporting questions

Is last thirty days the same as last month?

No. The rolling range ends yesterday and covers thirty completed days. Last month selects the previous calendar month. Check the printed boundaries so the client knows which period was used.

Does Run Now mean the report has already arrived?

No. It queues an enabled saved configuration. Evaluation and delivery follow separately. Treat the acknowledgement as a requested run until the result is available.

Does a Revenue label guarantee purchase-only value?

No. Check the declared metric. The saved conversion-value measure can include non-purchase value under its provider definition. Use the explicit purchase measurement for a purchase-only conversational question.

Can Ads and GA4 revenue appear together?

Compatible selected measures can share a separately labelled chart. Their traffic scope and attribution remain distinct. Do not add the two sources together or imply that different populations must give identical results.

Can the assistant rewrite the explanation without reading the account again?

Yes, when the existing evidence supports the rewrite. New account claims, periods or measures can need additional data. Creative phrasing must not invent a cause or business fact.

Does a report recommendation apply the change?

No. A supported live action uses a separate proposal and authorised confirmation. Its verified execution result is distinct from the later performance review.