Connect the advertised number to the intended regional business route
Google Ads call-asset audits for enterprise: distinguish presence, campaign routing and primary call outcomes
Understand the current audit's conversion trigger and coverage boundary, then prepare exact approved phone-asset associations without claiming complete call measurement or operational readiness.
Define the phone-route decision
An enterprise call-asset review should begin with the question the team actually needs answered. Is there any applied call asset, does one selected campaign use the intended regional number, or how many primary call conversions occurred recently? These questions concern different resources and measures. Establish the selected account, campaign scope and intended outcome before treating a simple asset-presence finding as a complete phone-acquisition analysis.
A read request does not authorise a live change. The assistant can identify applicable associations or explain the current audit without modifying the account. A request to add the confirmed business number to a campaign does require a supported proposal and confirmation. Keep that transition explicit, preserving the known target and number from the conversation while obtaining only the material missing information.
The enterprise setting makes routing particularly important. Several regions can share a central switchboard, while another service needs a separate appointment line. A technically valid phone asset can still lead customers to the wrong team. The organisation should define which route belongs to the campaign's offer and territory rather than let a familiar existing number become the default for every new association.
The wider enterprise audit workflow provides context for account review. This guide focuses on a narrower chain of decisions: interpret the audit signal, verify the intended resource and prepare the requested supported effect. It should help the team complete a useful configuration task while leaving measurement, staffing and customer outcomes tied to their own evidence.
Establish the business process behind the number
The advertised number belongs to a business process, not merely an asset field. Confirm who receives the calls, which questions they can handle and whether the advertised offer matches that team's responsibility. A central sales line can be appropriate for several campaigns, while a service-support line may be unsuitable for new acquisition. The assistant should use confirmed business facts rather than infer operational readiness from a stored phone number.
Consider the next step after a call reaches the business. The caller may need an appointment, a quotation or technical guidance. That purpose can inform creative and reporting choices, but the application does not inspect the enterprise's staffing or call-handling system simply by reading Google Ads. Keep supplied operational facts distinct from observed provider configuration so the final explanation does not imply access to systems the app never queried.
An operational owner should confirm the number's continuing suitability. A team can change hours, move a service or retire a route while the Ads association remains enabled. Assign that responsibility in the handover when the user provides it. Do not claim that the bot continuously tests the number, listens to calls or manages the enterprise's telephone infrastructure through the asset-creation action.
The review can legitimately end with no configuration change. If the current route is appropriate and the question concerns performance, read the relevant measure. If the route is uncertain, clarify it before preparing an asset. Activity is not the objective; the objective is a correct decision backed by the necessary evidence. Avoid manufacturing a new phone asset merely because the audit conversation began with a concern.
Understand the audit's conversion-category trigger
The current audit shows the call-extension finding when its enabled conversion-action inventory contains an action in the phone-call-lead category. That is a category-based relevance trigger. It does not inspect every possible call conversion type to decide whether the card appears, and it does not require that the matching enabled action be primary. Explain the actual trigger before drawing a broader conclusion from the card's presence or absence.
A secondary enabled phone-call-category action can therefore make the asset check relevant, while the application's ordinary primary-call performance measure excludes secondary actions by default. Those behaviours answer different questions and are not contradictory. One asks whether the account has a configured signal that makes a call asset worth checking; the other asks which selected outcomes belong in a primary-only report.
The absence of the card is not proof that the business receives no calls. A call-related action can use another category or type, and the audit trigger is deliberately narrower than a full phone-operation inventory. If the user asks to list call conversions, use the applicable conversational conversion catalogue rather than infer the answer from the audit card's visibility.
Similarly, an enabled conversion action does not prove that it recorded activity in the requested period. Configuration metadata and performance observations come from different reads. Preserve zero-activity actions in a requested setup list where the catalogue supports them, and obtain the selected period's metrics for an outcome question. A useful answer should identify whether it is describing configured measurement or recorded results.
Interpret the actual asset-presence check
The current call-asset finding checks whether the extension data contains a call type at account level or on any campaign in its declared campaign-coverage population. It is an account-level presence signal. A pass means the inspected data contains an applicable call association; it does not mean every campaign has its own call asset or every campaign uses the same correct number.
Read the actual unit in the summary. Present is different from complete campaign coverage. If the enterprise asks whether a named regional campaign has the intended number, resolve and inspect that campaign's applicable relationships rather than reuse a broad account pass. A single call association elsewhere in the account cannot establish the configuration of the user's selected campaign.
The check also does not verify call reporting, conversion thresholds or the real telephone journey. These are separate layers. A number can be configured while measurement is incomplete or the business route needs review. The assistant should report what the presence check actually observes and offer the appropriate next evidence when the user's question reaches beyond that simple configuration signal.
A failed presence finding is a starting point for investigation, not permission to create a guessed asset. Establish the intended number, campaign and supported association scope. The enterprise may already use a lower-level route outside the card's population or may choose a different acquisition approach. The action should match the business instruction, with the finding providing context rather than dictating an automatic universal remedy.
Keep campaign and relationship scope explicit
The shared extension reader obtains enabled account and campaign associations. Its campaign coverage list is based on applicable enabled campaigns that served during the declared reporting window. A status-only campaign inventory and a served-period coverage list can differ. State which population the call check uses before comparing it with a manual account view or a launch-preparation question.
An unserved campaign can still exist and need an appropriate phone route before launch. Its absence from a served-coverage list does not prove it has no assets. Inspect that selected target when the user asks about readiness. The current account-level card should not be stretched into a complete prelaunch assessment of every campaign or every relationship path in a large account.
Lower-level associations can matter to the actual route. Google explains that more specific call-asset levels can take precedence over broader ones. Its call-asset guidance also distinguishes configured assets from actual appearance. Use those facts as a reason to inspect the relevant branch, not as a claim that this particular audit already retrieves and certifies every lower-level association.
Account-level and campaign-level counts should remain labelled when both are shown. One stored number, several explicit associations and several potentially covered campaigns are different units. Do not multiply them into a made-up total of phone assets. A generic report should preserve the declared unit across Slack, tables and downloadable output so the enterprise can compare like with like.
Separate unavailable evidence from absence
A provider read that fails cannot support a missing-asset conclusion. If conversion metadata is unavailable, the trigger itself may be unresolved. If relationship data is unavailable, presence cannot be assessed. The audit should preserve that evidence state instead of treating an error as an empty list. Otherwise a transient outage can become a confident recommendation to create duplicate configuration.
A successful empty result needs its declared path and status. No enabled campaign association on a selected target does not rule out a broader account association or a different lower-level relationship. Combine the applicable evidence for the question and explain what remains uninspected. Technical absence is only useful when the population and scope make the business interpretation clear.
Keep earlier observations dated when a current refresh fails. A previously observed number can identify the intended resource, but it cannot prove the current association still exists. The assistant should state the current check's limitation and perform a focused refresh when appropriate. It should not erase the conversation's useful context or fabricate a clean current result to maintain a confident tone.
Read failures and ambiguous writes need different recovery. A failed read can be repeated appropriately to obtain the missing fact. A write that may have succeeded should be reconciled against the specific asset and association before retrying. Do not turn both into a generic send-it-again instruction. The enterprise needs to know whether configuration may already have changed.
Resolve the selected asset and campaign
An existing call asset and its campaign association are separate resources. Resolve the intended phone asset and target campaign before preparing an attach or detach operation. The approval should show a recognisable business number and campaign name, not a raw resource reference. Internal ownership and type checks establish precise targeting while ordinary labels make the proposed effect understandable to the approver.
Several assets can contain similar names or routes. A central line and a regional variant should not be selected by a partial label match alone. Use the returned content and known conversation context, asking a focused clarification when ambiguity remains. The assistant should not choose the first returned asset just because its name sounds plausible for the business unit mentioned in the request.
Campaign names can repeat too. Preserve the selected account and established hierarchy when the user says the campaign we just discussed. Context reduces repetitive discovery, but it does not override actual ambiguity. If two current campaign targets remain possible, identify the unresolved distinction rather than ask the user to repeat the entire original request or silently switch to an unrelated account.
Do not append raw account IDs or asset IDs in ordinary output, and do not show undefined parent labels. A call asset attached at campaign level is not an ad group's parent. The visible explanation should represent the actual relationship: existing or new call asset, named campaign, intended association effect. This clarity matters more than exposing the technical string used internally to send the API request.
Confirm the intended business number
Use the business-approved number the user supplies or explicitly selects from verified existing content. Do not infer a regional number from a company name, an account billing location or an unrelated website snippet. The route should belong to the advertised offer and responsible team. A missing number is a legitimate material clarification for creation, even when the campaign target is already known.
The dedicated creation path takes a country code and phone number and checks their declared shapes during preparation. Translate the user's ordinary country information into the applicable parameter where it can be resolved confidently. Client-facing dialogue should not require technical field syntax unnecessarily. The approval should still show the chosen number clearly enough for the business owner to notice an incorrect digit or region.
Passing the local shape check is not full provider eligibility or business verification. Country-specific requirements and the relevant phone-asset policy remain separate. Do not claim that punctuation accepted during preparation proves the provider will accept the complete asset. Applicable validation and provider preflight evaluate the request, while the owner establishes the business route and any evidence required outside the mutation.
Retain the selected number exactly through the proposal and verification. If a normalisation is needed, show the resulting intended value before execution rather than silently changing the route. A readable business number is necessary review information for this task, but it does not authorise exposing private account emails or unrelated technical identifiers. Keep the visible content limited to the actual decision.
Review regional relevance and actual routing
A campaign's region and a phone number's country context are related but distinct. A central route can legitimately serve several territories, while a local campaign may need a specialised line. The enterprise should establish which business process is intended. Do not assume the account's location targeting or billing information automatically determines the correct number for every service advertised within that account.
The number's apparent format also does not establish who answers. A business may route calls through a central system or regional service team. Use confirmed operational information when explaining the proposed association. If the organisation has not established that routing, record the missing ownership question instead of certifying the journey from the asset's content alone.
Inspect relevant hierarchy where the actual serving route is the question. A broader association can coexist with a more specific asset, and a location-related route can be relevant in appropriate contexts. The audit's presence result is not a substitute for that scoped investigation. The assistant should identify what it read and what the enterprise still needs to establish for the intended customer journey.
Changing one campaign association should not be described as a portfolio-wide routing migration. Other campaigns and broader relationships can remain unchanged. If the enterprise wants a staged regional rollout, keep each requested effect reviewable and sequence the appropriate resources. A central team can then verify the configuration milestone while the regional owner confirms the actual business route.
Assign hours and ongoing ownership
A configured call asset does not establish that the business can answer at every hour. The operational owner should decide how availability matches the campaign's offer and intended customer expectations. Record the relevant hours and follow-up responsibility where the user supplies them. The assistant should not claim that an association action tests staffing or automatically adjusts the enterprise's telephone opening hours.
The current dedicated call-asset creation parameters are limited to their declared number content and campaign attachment. They are not a complete hours, recording or call-reporting editor. If the user requests another setting, inspect the actual supported field-level operation and eligibility before proposing it. An hours note in the conversation does not silently create a provider schedule.
Temporary arrangements need a review condition. A campaign for a seasonal service may use a different business route until that offer ends. Assign the owner who knows when the association should be revisited. This is useful operational context even when no scheduling tool is used. It should not be presented as configured automation unless the application actually prepares and verifies that separate supported automation.
Ongoing ownership also helps reconcile later questions. If the team changes the service line outside the app, the next review should inspect the current selected asset and relevant associations rather than trust the old proposal forever. Retained context identifies the intended business purpose; a current read establishes mutable provider configuration. Both are useful, but they answer different parts of the question.
Use conversation for operational and creative planning
The assistant can help clarify a phone-acquisition brief without calling Google Ads for every sentence. Suggestions for a headline inviting an appointment, a description of the confirmed service or a concise internal routing checklist can be written from supplied context. Such creative and planning work should remain conversational until the user needs observed account data or requests a supported live change.
Do not invent business promises in call-oriented copy. Confirm availability, service scope and any necessary qualifications from the brief. A proposed contact line should not guarantee an immediate answer or a particular commercial outcome without support. Generated wording is a suggestion, not a provider recommendation or measured performance result. Keep that distinction visible when offering alternatives.
If the user selects an option and asks to apply it, retain the selected content and resolved target. A call asset does not itself contain the RSA headline being discussed, so the assistant must select the appropriate creative action for that separate resource. Generic dependency and action definitions should preserve this distinction rather than mix number fields and ad text into one malformed request.
Focused questions should concern material missing information. A conversation that already established the campaign and number should not repeatedly ask for them. Conversely, a request to attach this number is not permission to choose an unprovided destination for a separate ad. Preserve context intelligently and keep each action matched to the resource and fields it actually changes.
Understand campaign asset creation and attachment
The dedicated create-campaign-extension action can create a call asset from the declared country code and phone number, then attach it to the selected campaign. It retains existing extensions. The approval should show creation and attachment in clear rows under the actual campaign. It does not create a campaign, replace RSA creative or enable the campaign as a side effect.
The prepared combined mutation uses an internal planned asset reference for the new association. That reference belongs in execution handling, not in the client-facing target. The user should see the business number and campaign name. Provider validation and subsequent readback establish the actual configuration result, while the organisation retains responsibility for the number's intended business purpose and operational route.
Creation parameters should stay specific to the selected asset family. Sitelink descriptions, structured-snippet values and unrelated fields do not belong in a call-asset payload. The builder rejects fields that do not apply. A generic action framework is valuable when it describes these permitted fields explicitly rather than let the model invent arbitrary mutation JSON for each request.
Creating and attaching a number is different from merely storing a draft or offering a suggestion. If the user wants to discuss a possible route, keep the response conversational. If the requested live effect is only attachment of an existing asset, use that operation instead of creating another copy. The selected tool should reflect the instruction's actual resource and outcome.
Attach an existing call asset precisely
Attaching an existing call asset starts with a verified resource from the selected account. The dedicated builder checks the asset type and obtains visible content for review before preparing the new campaign association. This prevents a generic asset reference from being treated as a phone asset blindly. The target campaign should already be resolved through the conversation and applicable scoped evidence.
The action changes the relationship, not the stored number. A proposal to use the existing appointment line should identify that line and its new campaign association without implying a rewrite. If the same enabled relationship is already present, no duplicate create operation is necessary. Report the observed state honestly instead of generating another mutation merely to produce a visible execution event.
Provider-valid content can still be unsuitable for the selected business purpose. Review an evident region or service mismatch with the owner before preparing the relationship. The assistant should not silently attach a support number to an acquisition campaign when the conversation identifies a different appointment route. Technical eligibility and confirmed operational relevance are separate checks that support the same precise decision.
Other associations remain retained unless separately requested. Attaching one existing asset to a new campaign does not detach it from another campaign or alter account-level configuration. State the exact new relationship and any material retained context in plain language. Broader discovery should be justified by the question rather than performed automatically for every simple attachment.
Detach the association while retaining the asset
Detaching a campaign call extension removes the selected campaign relationship. The phone asset itself remains stored, and other associations can remain intact. A result should therefore say that the association was detached rather than that the number was deleted from Google Ads. This distinction is essential when the enterprise shares a central line across several campaigns or business units.
Detachment also does not prove the number is ineligible through every other path. Broader account or more specific associations may still matter to the actual route. If the user's goal is to stop a number appearing for a particular branch, inspect the relevant applicable relationships and supported actions. Do not claim that one campaign-level removal automatically completes a universal exclusion.
The builder reads the selected enabled relationship set and removes the exact association where it exists. If it is already absent from that inspected set, the response should describe that state without claiming a new live mutation occurred. A truthful no-op can satisfy a configuration question, but it should not become a fake history entry announcing a change that was never sent.
A replacement sequence may include a new asset association and a separate old-association removal. Show both steps when both are requested. The creation of a new route does not imply the old one was removed, and detachment of the old route does not prove the new one exists. Verification should follow the actual declared effects rather than infer completion from one successful operation.
Keep the number separate from call measurement
Google's call measurement guidance distinguishes call assets and reporting configuration from the conversion action used to count outcomes. Those provider layers need their own appropriate setup and eligibility. The application can inspect configured actions and recorded metrics, but creating a phone asset alone does not implement complete call measurement or prove that the account records the intended business outcome.
The audit's phone-call category trigger is only metadata. It does not test the full measurement journey or certify call recording and downstream qualification. Avoid describing an enabled action as evidence that every incoming call is tracked correctly. The enterprise should establish the action's intended meaning and use actual period data for recorded results, with external operational evidence where the question requires it.
A click to a phone link and a measured call outcome can be different events. The report should identify the actual conversion definition selected from the account rather than use calls as an ambiguous label for every interaction. This is particularly important when comparing Ads with an analytics event on the website. The selected event's meaning belongs in the comparison explanation.
The bot should not claim access to private call recordings, transcripts or the enterprise's CRM merely because an Ads metric is available. A configured call action and its aggregate results are useful evidence, but they do not reveal the contents of individual customer conversations. Keep the analysis within the actual tool inputs and returned fields, asking for supplied business context when a wider interpretation is needed.
Report primary calls with the selected definition
The conversational measurement catalogue includes a call group based on applicable call action types or the phone-call-lead category. Ordinary call reporting selects primary actions by default, excluding secondary ones from that measure. This is broader than the audit card's category-only relevance trigger. Use the report catalogue for outcome questions and explain the selected definition without deriving totals from the card's mere presence.
Several primary call actions contribute to the selected aggregate. The tool does not deduplicate separate conversion actions recording the same underlying business event. That is the account's chosen measurement structure, not a hidden cleanup task. The user can request a named action when a narrower definition is needed, and the report should preserve that selection across the summary and graph.
The catalogue can list configured actions even when they have no activity, with names, primary status and relevant metadata. Listing is different from reporting recent counts. A secondary action can be shown in that metadata list without being included in a default primary-only outcome total. Explicitly requested secondary analysis needs its own clear selection and labels rather than a silent expansion of an ordinary call report.
Do not impose a business-quality hierarchy from category names alone. The enterprise may use several lead stages, but an aggregate lead request groups them as leads according to the selected definition. For a specific call or contact outcome, the action's exact name can be the clearest scope. Ask a focused selection question when that meaning materially affects the user's requested comparison or live target.
Interpret value assigned to call conversions
Call conversions can carry assigned value. That value can contribute to an all-primary conversion-value measure when the action is primary. It should not automatically be labelled purchase revenue. The account's configured value meaning needs to be preserved, particularly when enterprise teams use monetary values to represent expected acquisition outcomes rather than a completed retail purchase.
Several primary value-bearing actions can contribute to the selected total, while secondary actions remain excluded by default. A report of call value should use the call action selection rather than assume that all account conversion value comes from calls. Conversely, a purchase-only measure needs its purchase scope. Accurate labels help users understand what the ratio or comparison actually represents.
Cost per call is spend divided by the selected call outcomes for the compatible population. A value-to-cost ratio uses the selected value and spend. Calculate from full-precision base metrics and round only displayed values. Missing denominators or unavailable data should be reported as undefined or unread, not converted into zero or an invented infinite performance claim.
Assigned value is not proof of realised sales, margin or cash received. The enterprise may use a useful planning value, but the report should describe it honestly. A later business-system comparison needs applicable supplied or connected evidence. Do not claim that creating a call association improved revenue merely because the account already records value on a call conversion action.
Identify offline actions without double-counting groups
The conversational conversion catalogue can identify its declared uploaded and imported action types as offline conversions. These configured actions remain available for listing and selected reporting within the tool's scope. A call-upload action can belong to both the call and offline groups, while another action can overlap leads or purchases. Group totals should therefore not be added together as if they were disjoint populations.
If the user asks which offline actions exist, provide their actual names and relevant metadata from the account. That answer does not require a GA4 counterpart. It also does not prove a complete upload pipeline is working or grant the bot access to the source CRM. Listing visible provider actions and diagnosing the business's external data flow are different tasks.
A named-action performance request can narrow the scope to the action the enterprise cares about. Preserve primary-only defaults unless secondary reporting is explicitly requested and supported. The actual action selection should appear in the definition so a graph, total and follow-up table all refer to the same population. Avoid fuzzy name matching that accidentally selects a similarly labelled conversion.
Configuration visibility should not be exaggerated into universal offline classification. Use the catalogue's declared types and returned metadata. If the enterprise uses an external route whose type or availability is not established, inspect the appropriate evidence rather than infer it from a suggestive action name alone. A truthful list with known scope is more useful than an unsupported claim to see every downstream system.
Confirm any GA4 comparison event
There is no universal GA4 event that automatically equals every Ads call conversion. A website may record a phone-link click, a contact request or another event, and those meanings can differ from the account's call outcome. If the user asks for a comparison, select the specific corresponding event for that site rather than force a generic call pairing.
The assistant can inspect available event names and suggest a plausible one, but the user should confirm the intended comparison when its meaning matters. A name such as contact_phone may be useful evidence for a suggestion, not proof of the implementation. Preserve the user's confirmed event in the report definition and label the two sources clearly so the enterprise can assess the comparison.
Aggregated Ads conversions can also be compared with a declared total of GA4 key events, while an individual Ads action can be compared with one named event. These are separate selection modes. A flexible reporting framework should express the populations explicitly rather than implement a dedicated one-off chart for a particular action name. The same definition should govern summary, graph and follow-up filters.
Do not equate event counts with recorded conversations without evidence. A comparison can be useful even when the measures have different coverage, provided that difference is stated. The assistant should ask the necessary focused question, then retrieve the compatible period and grain. It should not reject all cross-source analysis simply because attribution differs, nor should it pretend the selected measures are automatically identical.
Preserve dates, grain and exact conditions
Current phone assets and conversion metadata are configuration observations, while call results belong to the requested reporting period. Last thirty days means the appropriate rolling completed-day period, not the previous calendar month. Keep the account time zone and actual dates explicit. A current association cannot establish that the same route was configured throughout a historical period without applicable change evidence.
Use the requested grain for counts and ratios. Campaign outcomes can provide campaign context, but they are not automatically measured results for a particular shared call asset. Daily segmentation requires actual daily values, not distribution of a period total across dates. If the available field or source cannot support the requested grain, explain the specific evidence boundary and choose an appropriate supported next step.
Exact comparisons matter. Above excludes equality; at least includes it. Below, at most and equal are different operators. Preserve combined AND and OR restrictions and apply them at the declared entity level. Missing or failed data cannot satisfy an equality-to-zero condition, and unavailable cost per call should not become a zero ratio merely to populate a table.
For visuals, favour one native chart using an appropriate weekly or monthly grain for longer periods unless daily detail is explicit. Explain multiple panels when explicit daily detail exceeds the twenty-point native limit. Keep series labels, source and event selections readable, show no more than two decimals and reuse retrieved data for the visual. A chart request should not trigger redundant provider calls just to render the same values.
Follow an illustrative regional phone review
The following example is invented. Mina Ash manages advertising for Cypress Field Services, an enterprise with a central enquiry line and a separate regional appointment team. The audit's call-asset card passes because its inspected population contains a call association. Mina asks whether the North Installation campaign uses the appointment route. The assistant treats that as a campaign-specific question instead of presenting the account pass as its answer.
The relevant current evidence identifies the selected campaign and its existing associations. Mina confirms the appropriate appointment number and responsible team. She asks for a concise explanation of the intended customer journey; the assistant writes that from the supplied business context without running Keyword Planner or claiming it inspected the telephone system. The configured number and the operational confirmation remain distinct facts.
Mina requests creation of the confirmed call asset and attachment to North Installation. The proposal shows the number, country context and named campaign, then lists creation and attachment in separate readable rows. Existing extensions are retained. The action does not change call reporting, conversion roles, campaign status or the central line's other relationships. Mina confirms the exact requested effect.
The result verifies the asset's declared content and the enabled campaign association. Mina separately reviews primary call outcomes for the selected period and assigns the regional owner to confirm the real answering route. No revenue recovery or comprehensive measurement fix is claimed. The enterprise completes a precise supported configuration task while retaining the distinct operational and reporting checks.
Sequence resources and avoid unnecessary reads
A call association depends on a campaign and an asset. The local resource model can represent those relationships and required ordering without external calls to rediscover elementary structure. Existing target facts still need applicable scoped reads, while a new asset in a supported combined mutation uses a planned reference. Structural dependency knowledge and current account evidence are complementary, not interchangeable.
Reuse the account, campaign and selected number already resolved in the conversation. A follow-up asking to attach the same route elsewhere needs the new target, not another broad discovery of every conversion action. A request for recent call counts needs measurement evidence, not another asset creation plan. Selective retrieval improves responsiveness while keeping the answer grounded in the actual question.
Independent required reads can proceed concurrently where their dependencies allow it. A child operation that depends on a newly created parent must wait for the relevant reference. Shared authentication coordination and request-local metadata reuse can reduce duplicate work without mixing account scopes. The aim is a coherent evidence flow, not maximum parallelism regardless of resource dependencies.
When a request exceeds the supported safety limit for one proposal, ask the user to split it. Do not execute an arbitrary subset and imply the whole instruction completed. If a sequence partially succeeds, reconcile current resources before continuing so the next proposal does not duplicate an asset or association. Recovery should preserve both the user's intended larger task and the actual completed configuration.
Show each exact action before confirmation
The approval should name the selected campaign and show each logical action as a distinct step. Creation displays the confirmed country context and phone number. Existing attachment identifies the selected asset and new relationship. Detachment identifies the removed relationship and retained asset. These are meaningful review details; raw JSON, internal resource names and undefined parent fields are not.
Keep retained effects explicit where they matter. Creating a call extension retains existing extensions. Attaching an existing asset does not rewrite its number. Detaching one association does not delete the asset globally. If the user requested another campaign or conversion setting, prepare it as a separately named supported effect rather than hide it inside a vague improve-calls instruction.
The confirmation notice states: No changes have been made yet. This proposal expires in 15 minutes. The live edit occurs after Confirm and Execute. The enterprise approval guide explains the wider review workflow. Preparation and applicable validation can happen before the click, but the interface should not present them as completed live account changes.
Once clicked, show a prominent change-in-progress state and remove the available confirmation control while execution proceeds. Do not add a user mention to the interim message. The person should immediately understand that the click registered. A final status then reports the actual outcome, with separate rows for completed and unresolved steps rather than a delayed ambiguous wall of text.
Verify content and the requested association
The dedicated creation path reads back the written call-asset content and checks the enabled campaign association using the returned resource reference. That confirms the configuration milestone represented by the prepared payload. Validation before the write checks the request's applicable constraints; verification afterwards checks its result. The two stages are distinct, and neither alone proves the external phone journey.
For attachment and detachment, the relevant relationship set is checked against the expected outcome. The result should report exactly that association effect. It should not infer a number rewrite, global deletion or conversion-reporting change from the relationship check. Verification is credible when its scope matches the specific operation the user reviewed and approved.
If the requested state already exists, describe the no-op honestly. An already attached asset need not be attached again, and an absent enabled relationship need not produce a remove mutation. The configuration can satisfy the user's intended state without a new execution. Avoid creating misleading history that celebrates an account change when the system only observed an existing condition.
Ambiguous writes require reconciliation. A timeout can occur after the provider accepted the asset or association. Read the specific relevant state before replaying a creation request, and explain which result is known. Partial verification should remain partial in the output. The owner needs a precise next check, not an instruction to resend a whole sequence that may already have changed the account.
Keep policy and the real phone journey separate
Successful configuration readback does not override provider policy or prove eligibility in every context. A number can exist with the requested association while another condition needs review. Observe the applicable current status when the user asks about that milestone. Do not use an immediate Executed and verified result for the asset fields as a universal assurance that the number will appear or every policy requirement is satisfied.
The real phone journey needs its own responsible observation. The enterprise can confirm that the number reaches the intended team and that the team handles the advertised offer. That evidence can be supplied in the conversation without pretending the Ads mutation performed a telephone test. Record the completed setup milestone and the operational confirmation separately in the handover.
A later issue should be investigated at its actual layer. A wrong route calls for operational and asset review, a missing recorded outcome calls for measurement investigation, and a provider restriction calls for its specific policy evidence. Avoid treating every symptom as a reason to create another phone asset. Precise diagnosis prevents repetitive edits that add configuration without addressing the user's problem.
Commercial improvement also requires later evidence. A newly correct route can be an important necessary change, but the tool should not announce reduced cost per call or recovered sales immediately after attachment. Use a compatible follow-up period and the same primary or named-action definition to evaluate outcomes. Configuration success and observed business results remain separate milestones.
Deliver a readable result and accountable follow-up
A successful final message should prominently display Executed and verified and list the actual completed effects beneath it. Identify the call asset by its recognisable business content and the campaign by name, without raw IDs. If one requested effect remains unresolved, use an accurate partial status. Strong presentation should make the scope clearer rather than amplify a claim broader than the verification supports.
For ordinary AI replies, place the user mention before feedback text and controls so feedback remains last. Keep provider timestamps, internal syntax and private account details out of client-facing prose unless they are necessary in an appropriate restricted investigation. A phone number intentionally being approved is relevant task content; unrelated identity information is not a useful substitute for a readable target name.
Reports should preserve the chosen measure across summaries, charts and tables. Use current-year dates without unnecessary years, actual weekly ranges where grouped, and no more than two displayed decimals. Keep source and event labels explicit. Reusing the retrieved dataset for a requested visual helps both speed and consistency, while a changed scope requires the appropriate new evidence.
The handover should identify the phone-route owner, any hours review and the next measurement observation the enterprise needs. A call-asset audit is most useful when it distinguishes category trigger, asset presence, campaign routing, primary outcomes and operational proof. It can support a precise approved association change without claiming to complete every layer of the organisation's phone-acquisition process.