Make every additional destination relevant to its actual account branch
Google Ads sitelink audits for enterprise: review shared content, descriptions and exact campaign associations
Separate linked-asset completeness from campaign coverage and actual display, then prepare readable approved sitelink changes with clear destinations, ownership and retained relationships.
Define the enterprise sitelink decision
An enterprise sitelink review should begin with the decision the team needs to make. The question might concern missing description lines, a destination that no longer represents the offer, or a regional campaign that needs a relevant existing link. These are different tasks. Establish the account, campaign scope and intended effect before turning an audit finding into a broad instruction to replace every additional link.
Separate information requests from changes. Which applied links lack descriptions is a read. Suggest two clearer descriptions is a creative conversation. Add a new link to this campaign is a live configuration request. An assistant can move coherently between those stages, but it should not treat a suggestion as permission to mutate the account or treat a requested live action as an invitation to give only generic advice.
The enterprise context makes scope especially important. Central navigation can be shared across many campaigns, while service availability and offer details can differ by region or business unit. A link that is sensible for the main brand campaign may be inappropriate for another branch. The review should connect content to the actual destination and audience rather than assume that a familiar corporate label makes it universally relevant.
For the broader account review, see Google Ads audits for enterprise. This guide focuses on the additional link itself and its relationships. Its useful outcome is a clear content or association decision supported by actual evidence, followed by a precise approval where a live change is requested. It does not certify every surrounding campaign setting simply because the sitelink check passes.
Give the extra destination a clear customer purpose
A sitelink should help a person reach a relevant part of the website without making them repeat their search through the main navigation. The enterprise should identify what the destination lets that person do: review a specific service, compare an appropriate product range, find a location or understand support. That purpose informs the label, descriptions and association scope more effectively than adding generic corporate pages to satisfy a count.
Google describes sitelinks as additional links to specific pages and explains that associations at different levels of the same account branch can be eligible together. Configured eligibility does not guarantee a particular appearance. Its sitelink guidance also recommends useful descriptions. Treat those facts as design context, while using the actual business offer to decide which pages deserve inclusion.
A page's purpose should be truthful at the time of the review. A labelled installation guide should lead to the installation information the team intends, rather than an unrelated product listing. A regional service link should not invite customers into an unavailable territory without an appropriate explanation. The assistant can improve wording from supplied facts, but the business and website owners must establish what the page actually offers.
Useful diversity is different from cosmetic variation. Several labels that lead to the same broad page can make the proposal look complete without providing distinct choices. Review the destination purpose before evaluating the number of links. The team may choose to retain a smaller coherent set while preparing another page, rather than inventing a destination or adding repetitive wording solely to improve an audit presentation.
Understand the current inventory reader
The current sitelink-description reader obtains enabled account and campaign asset associations, then reads sitelink content from the asset resource. It joins those relationships to the returned asset content by the internal resource reference. This distinction matters because descriptions are not reliably supplied through every relationship view. The audit examines content associated through its declared account and campaign paths, rather than every stored sitelink indiscriminately.
The campaign relationship query selects the referenced campaign channel field along with its filters. That detail helps preserve the validity of the provider read. A rejected query must not be interpreted as an empty set of relationships. Enterprise findings should come from successful evidence retrieval, not from a quiet fallback that makes an API error look like every campaign has suddenly lost its assets.
The reader does not turn lower-level associations into a universal inventory automatically. An ad-group-specific link may require a separate applicable scope when the user asks about it. State which paths the reported count uses and obtain the requested lower-level evidence where supported. Do not tell a colleague that their manually observed lower-level link does not exist simply because it is outside this audit's relationship population.
Keep stored assets distinct from applied assets. A sitelink can exist in the account without an enabled association within the inspected scope. Its content may be complete, yet it may not contribute to the reported applied set. Conversely, one shared asset can appear through more than one association. The audit's relationship and content join should be explained before comparing its inventory with a broad asset list in Google Ads.
Separate unique assets, relationships and campaigns
The current description check counts distinct linked sitelink asset references across its account and campaign association sources. If one asset is linked to several campaigns, it remains one unique asset in that completeness total. That is a sensible unit for asking whether the asset's two description fields are present. It is not the same unit as counting every campaign attachment separately.
When the enterprise asks how many campaigns use that link, count the relevant relationships or distinct campaign targets instead. A result can truthfully say that one unique sitelink is associated with several selected campaigns. It should not describe those associations as separate content assets. Keeping both numbers labelled helps the owner understand how widely a shared resource could matter without inflating the inventory.
Account associations introduce another distinction. A link applied at account level can be considered by the coverage check without one explicit campaign relationship for every campaign. Do not multiply the asset count by the campaign count and call that a retrieved inventory. Account applicability, explicit campaign associations and observed display are different concepts. Report the one the user requested with its actual source and scope.
For examples, a short set of names can explain an issue, but it should not masquerade as a complete count. If the result is truncated for readability, retain the complete underlying population for totals and label the shown rows as examples. A generic reporting framework should preserve its declared unit across the Slack summary, table, downloadable report and follow-up answer.
Interpret the two-description finding
The sitelink description finding checks whether both description fields are filled for each distinct applied asset in its inspected set. A sitelink missing one line and a sitelink missing both lines are both incomplete, with different missing-line details. The absence of all applied sitelinks is handled separately through coverage; the description card does not claim completeness for a population that contains no links.
Presence is a limited check. Two filled lines can be repetitive, outdated or unrelated to the page. A pass on the field-presence check does not certify those business qualities or promise that the descriptions will appear. Read the actual content when the user requests a relevance review. The assistant should identify the audit's observed completeness without treating that simple signal as a full creative evaluation.
Google's current sitelink field reference describes link text and the paired description fields, including their length constraints. The enterprise should supply a coherent pair when adding descriptions rather than leave one line unsupported. Provider field rules and the application's preparation checks remain distinct; passing a local content check alone does not establish that the complete provider request will be accepted.
If a linked asset's content row is unavailable in an otherwise returned set, the displayed missing-description indication is not the same as an observed blank field. Reconcile the specific asset before preparing a content change based on that indication. The owner needs to know whether the current fields were actually read. This avoids turning an incomplete join into a confident statement about what the sitelink contains.
Read campaign coverage in its declared period
The core coverage check is separate from the description check. It considers enabled account-level assets first, then campaign-level types for applicable enabled campaigns that served during the selected window. The standard audit window is recent performance context, not a permanent list of every campaign ever created. Explain that population when a colleague compares it with a broader status-only campaign view.
A campaign that did not serve in the window can be absent from the served-campaign coverage population without being absent from the account. It may be deliberate preparation, paused earlier, or awaiting another setup milestone. The coverage result should not describe its absence as a zero-content finding. If the enterprise wants launch readiness for that campaign, resolve and inspect the campaign explicitly rather than reuse the served-population conclusion.
Account-level presence can satisfy the audit's coverage test, but presence is not proof that every link is relevant to every eligible campaign. Central teams should still review business scope. A shared customer-support link may be broadly suitable, while a limited offer can require narrower campaign associations. The application's coverage model helps locate omissions; it does not replace the organisation's content and territory judgement.
Do not conflate the two populations when producing a summary. The distinct applied-asset description total and the served-campaign coverage list can be useful together, yet they are not identical sets. Label each finding and its unit. A reader should understand whether the next task concerns an incomplete content resource or an association gap on a specific campaign.
Preserve unavailable and incomplete evidence
A provider failure is not evidence that sitelinks are missing. If the content query fails, the audit should report that the check could not be completed rather than count every relationship as bare. The same applies to failed association or campaign reads. Preserving the distinction between absent and unread protects the enterprise from preparing unnecessary edits based on a transient technical problem.
An empty successful query also needs its scope. No enabled campaign-level sitelink relationships on one selected campaign does not rule out an account-level asset or a different relationship path. The assistant should combine the applicable evidence for the user's question and explain any uninspected path that changes the interpretation. Absence becomes useful only when the resource family, status and scope are clear.
Incomplete evidence should remain visible across follow-ups. If the team previously observed a shared asset and later could not refresh its descriptions, preserve the earlier observation with its date instead of inventing current blanks. A focused refresh can reconcile mutable content when necessary. Retained conversation context identifies what to inspect; it does not supply the missing current provider fields by itself.
Before a live change, ensure that the target and requested effect are grounded in current applicable evidence. Do not attach a guessed asset simply because its label resembles the user's wording. A failed lookup should lead to a narrow clarification or an appropriate retry of the read, while an ambiguous write outcome needs reconciliation. The two cases require different next steps and should not be collapsed into one generic retry instruction.
Resolve the actual hierarchy and target
A campaign association connects an asset to that campaign. The asset itself is not a child ad group, and its link text is not necessarily a globally unique name. Resolve the actual existing resource and intended campaign before preparing an attachment or detachment. The approval should show the campaign name and recognisable asset content so the user can confirm the relationship without deciphering resource syntax.
Enterprise accounts can contain similar links in several languages or destinations. Two assets labelled Support may point to different regional pages. Use the actual content and destination where relevant to distinguish them. If the user selected a particular variant earlier, retain that choice across follow-ups. Ask a focused question only when the known context and available evidence still leave a material ambiguity.
Keep internal references out of the ordinary client output. Account IDs, raw asset resource names and undefined parent values do not help a user approve a readable sitelink association. They remain necessary internally for exact targeting and readback. The explanation should show the real relationship and destination, with sensitive tracking material excluded unless it is genuinely required in a restricted technical investigation.
The selected account should remain established throughout the conversation. A request to add the same link to another named campaign can reuse the verified asset context while resolving the new target. It should not silently switch accounts or assume that a resource from another account is valid. The local ownership checks and scoped lookup protect the intended target while the visible names make the action reviewable.
Write recognisable and useful link text
The link label should identify the destination's purpose in language a prospective customer recognises. Enterprise organisational names can be meaningful internally and opaque outside the business. Prefer a clear service, product or task label when the destination supports it. The assistant can suggest alternatives from the confirmed page purpose without claiming those alternatives have measured click performance or came from provider recommendations.
Keep the label aligned with the actual page. A link promising pricing should lead to the intended pricing information, and a service-area label should represent the real scope. Do not use the same generic wording for unrelated regional offers just to simplify asset reuse. Clarity supports the customer's next step and makes the enterprise's review easier because the label and destination can be assessed together.
The dedicated creation builder checks the supplied sitelink text length before preparing its payload. Use that constraint when drafting, but do not treat it as the sole quality standard. A short label can still be unclear, unsupported or duplicative. The business owner should review what the phrase means, while provider validation checks whether the prepared field values satisfy the applicable request rules.
When the user chooses one suggested label, preserve that exact selection in the later proposal. Do not substitute a different variant during execution preparation because it seems more polished. A readable approval should be a faithful translation of the selected content and intended action. If shortening is necessary to meet a constraint, explain and show the revised wording before asking for live confirmation.
Develop descriptions without unnecessary provider calls
A request for two descriptions can often be answered from the page brief, selected label and confirmed business facts. It does not automatically require Keyword Planner, another campaign inventory or a fresh full audit. Creative conversation is a useful part of the assistant's role. Reserve provider calls for facts that must be observed in the account or for preparing and executing an actual supported mutation.
The pair should add complementary information. One line can explain the destination's subject and the other the relevant next step, provided both are true. Avoid repeating the label twice or inventing urgency, guarantees or pricing. Enterprise review teams should supply required qualifications and prohibited claims where applicable. The assistant can then produce coherent alternatives that respect those instructions across the conversation.
Generated suggestions should be labelled as proposed content. They are not historical search terms, observed competitor wording or performance-tested claims unless appropriate evidence actually establishes those facts. If the user asks for examples, provide useful examples rather than turning the conversation into an unsolicited live proposal. The transition to an edit should occur when the user requests a supported action with selected content.
For a follow-up such as use the second pair and add it to the campaign, retain the chosen pair and campaign context. Obtain only the additional information needed for the applicable operation, including the intended destination if it is not established. A coherent response should not force the user to restate the entire brief, nor should it fill a missing public page with an invented URL.
Establish the intended public page
A sitelink's final destination should correspond to its actual purpose and intended campaign audience. Ask the enterprise for the public page when the conversation has not established it. A business domain alone does not necessarily identify the service, support or product page the link should use. Preserve the selected destination in the proposal so the approver can recognise it before the account is changed.
The creation path can normalise a provided address to HTTPS and check its public URL shape. That preparation does not prove the page loads correctly, accepts the relevant customer or contains the promised information. Syntactic validation and a tested journey are different milestones. A useful result reports the URL field that was verified without claiming a website repair or destination certification the tool did not perform.
Regional redirects and language variants deserve separate business attention. A page can be public yet lead a customer to a different offer or territory. The website owner should confirm the intended experience where that distinction matters. The assistant should use supplied evidence and appropriate supported checks, rather than assume that every path under a recognised enterprise domain is interchangeable.
Do not confuse a readable display description with the page itself. The sitelink can describe a destination, but it cannot create that page or change its content through an Ads association action. If the page is still being prepared, record that dependency and choose an appropriate next step. The enterprise may complete creative preparation while leaving live association until the required public destination is confirmed.
Keep tracking and privacy responsibilities separate
Tracking configuration can affect how an additional destination is reached or measured, but the description check does not diagnose that entire journey. Account templates, lower-level URL fields and website measurement have distinct responsibilities. Investigate the relevant layer when the user's question requires it. Avoid clearing unrelated tracking fields during a sitelink association request merely because they look technical.
The dedicated campaign extension creation writes its declared content and final URLs and attaches the asset. It does not reconfigure account auto-tagging, conversion actions or the website's analytics implementation. If another supported setting needs a change, prepare that effect explicitly in its proper scope. The confirmation should not imply that creating one link completes a comprehensive attribution setup.
Client-facing summaries should avoid private emails, internal account identifiers and token-bearing URLs. Show the public destination necessary to review the proposal and describe tracking issues without dumping sensitive configuration. A downloadable report should follow the same privacy discipline as Slack. Changing the output format is not permission to expose technical identifiers that the user does not need for the decision.
Measurement evidence should identify its actual source and population. A visit associated with the destination is not automatically a sitelink click unless the returned metric and segmentation support that interpretation. Likewise, an account conversion total should not be attributed to the additional link merely because the link is configured. Preserve the declared evidence so the enterprise can distinguish setup, usage and downstream outcomes.
Assign temporary offers an explicit owner
Temporary destinations and seasonal wording need an owner who knows when the offer stops being appropriate. A link that remains attached after a promotion ends can mislead users even if both description fields are complete. Record the business review date or condition and the responsible team. This is an operational handover, not evidence that the application automatically monitors every offer's continuing validity.
The current dedicated creation parameters should not be advertised as a full scheduling editor. If the enterprise wants a provider schedule or another field-level setting, inspect the actual supported operation and eligibility before preparing it. A calendar note in the conversation does not silently create an ad schedule. Explain which requested effect is implemented and which needs a separate supported configuration or owner task.
Temporary changes should preserve the intended permanent links. Creating a seasonal variant does not require detaching every existing sitelink. If the user wants one association removed, identify that exact relationship and show it separately. The enterprise should be able to see what will remain available after the offer-related action, without a broad replacement policy being inferred from a narrow request.
After the temporary period, a focused review can inspect the relevant association and public destination. Reuse the established asset and campaign context rather than start another full account discovery. If the requested action is to detach a link that is no longer needed, verify the relationship outcome. Do not claim that a detached campaign link is globally deleted or unavailable through every other account branch.
Understand campaign creation and attachment
The dedicated create-and-attach campaign extension action creates a sitelink asset with its supplied content and final URLs, then creates the association to the selected campaign. It retains existing campaign extensions. The proposal should show those effects as clear rows: create the described sitelink and attach it to the named campaign. It does not create a campaign, an ad group or a new RSA.
Google's official sitelink example distinguishes asset creation from the campaign association. The application uses the same resource distinction in its prepared combined mutation, with an internal temporary reference resolved by the provider. The enterprise should see the content and target, while the execution layer handles those references without exposing placeholder IDs in the approval.
Provide the selected label, coherent description pair and intended public final URL. The builder permits its declared fields for the chosen extension type and rejects unrelated parameters. A phone number or structured-snippet header does not belong in a sitelink creation payload. Resource-specific preparation prevents a generic request from becoming a malformed mixture of settings across several asset families.
Do not confuse the combined create-and-attach operation with a request to store content without applying it. Its named effect includes a campaign association. If the user only wants ideas or an uncommitted content draft, keep that task conversational. If a different supported resource-level action is needed, determine and explain that action rather than use a convenience tool whose live consequence does not match the instruction.
Attach an existing asset deliberately
Attaching an existing sitelink uses a verified asset reference from the selected account and a specific campaign target. The dedicated builder checks that the asset type matches the requested extension type and obtains recognisable content for review. A label match alone is not sufficient when several resources could be intended. The selected existing asset should be resolved before the relationship proposal is prepared.
The attachment changes the relationship, not the asset's descriptions or destination. State that clearly when the enterprise says use this existing link on another campaign. The approval should identify the existing content and target campaign, with no implied rewrite. If the same relationship already exists in the relevant enabled set, the system should not create another duplicate association merely to simulate activity.
Business relevance still needs review. An asset can be technically selectable and unsuitable for the new campaign's offer or territory. The assistant can point out an evident conflict in the supplied context and ask a focused clarification, while the owner provides the actual business decision. Provider acceptance is not an endorsement of relevance, and a valid association is not proof that the additional destination will improve results.
Keep other associations retained unless an explicit additional action changes them. Adding one campaign relationship should not detach the asset from its current campaign or alter account-level applicability. A concise explanation of the new relationship and retained content is enough for many requests. Wider discovery should be justified by the question or risk, not performed automatically for every simple attachment.
Detach the relationship without deleting the asset
Detaching an existing campaign extension removes the selected campaign association. The asset itself is retained. This distinction is especially important for shared sitelinks: a regional campaign can stop using one relationship while the same asset remains stored and associated elsewhere. The result should describe the verified relationship removal rather than call the resource deleted or globally excluded.
An account-level association can remain relevant after a campaign-level detachment. The requested relationship outcome and the broader eligibility question are different. If the enterprise's intention is that the link never be eligible for a particular branch, inspect the applicable paths and supported operations instead of assuming one campaign detachment guarantees that business outcome. Explain retained paths that materially affect the stated goal.
The builder reads the existing enabled relationships and prepares removal for the exact selected association. If the relationship is already absent from that inspected set, no fresh remove operation is needed merely to produce an execution message. The answer should state the observed configuration honestly. Avoid a fake change result that claims a mutation occurred when the requested relationship was already absent.
Detachment can be part of a broader replacement sequence, but it should remain a clearly named step. Creating a regional variant and detaching the old campaign link are different effects with separate evidence. The owner should approve both if both are requested. A new asset's existence alone does not imply removal of the original, and a removed relationship does not imply the new variant was created successfully.
Distinguish automated assets and operation eligibility
The audit counts enabled relationships from its inspected sources without treating automatically created content as automatically absent. That is a read-side population choice. Write eligibility is a separate issue. A resource returned by an audit may have provider restrictions that prevent a particular content mutation or new association. The assistant should preserve that distinction before promising a live change to every returned asset.
Google's automated-asset documentation distinguishes restricted creation and editing from supported association status or removal operations. Use the actual resource source and operation rules when selecting a next step. Do not promise that the campaign creation tool can recreate an automated asset or attach it freely just because the audit shows it as part of enabled inventory.
Manual and automated content can coexist in an account's review. The enterprise may choose a new advertiser-supplied variant with confirmed business wording while separately reviewing an unwanted automated association. Those are different tasks. The approval should identify which resource and relationship it changes and avoid implying that a single manual creation action toggles the entire account's automated-asset settings.
Provider preflight remains important where operation eligibility depends on current resource details. Local structural knowledge can identify the relevant family and valid fields, but it cannot override provider restrictions. If a requested operation is unavailable, preserve the user's intended outcome and describe the supported alternative or owner task precisely. An honest limitation is more useful than a broad unsupported promise.
Separate configuration, policy and actual appearance
An enabled campaign association means that the configured relationship exists in the inspected state. It does not prove the sitelink appeared on every impression or will appear on the next one. Actual serving depends on the applicable provider context. The audit's completeness and coverage findings should be presented as configuration observations, with display evidence obtained separately when the user specifically asks about usage.
Policy status is another distinct observation. Successfully creating or attaching an asset does not certify that every policy condition is satisfied. The application's readback can verify the supplied content and relationship; a later provider policy result requires its own applicable evidence. Enterprise execution messages should report the completed configuration milestone without implying that verification overrides external review.
Several resources may be eligible within a branch, and a configured asset may have little observed activity. That alone does not prove a broken association. Investigate the intended period, campaign population and available performance fields before diagnosing the cause. A low display count should not automatically trigger a creative replacement, particularly when the business purpose and content remain valid.
The website experience remains independent too. A sitelink can be configured correctly while its destination later changes. Assign ongoing content ownership and review the page when the offer or journey changes. A one-time verified asset write is a useful record of what the tool completed, but it does not provide permanent monitoring of the business page or a lifetime guarantee of relevance.
Use correctly scoped commercial evidence
Performance can inform the review when the metric population is explicit. Campaign cost provides context for prioritising a coverage gap, but it is not measured lost revenue attributable to a missing description. Likewise, a campaign's total conversions are not automatically conversions caused by one sitelink. Use applicable asset or interaction evidence when the question requires that grain and state any available-data boundary.
Choose measures that fit the business. A lead campaign may need primary leads and cost per lead, while an ecommerce campaign may use primary value and ROAS. Multiple primary actions contribute to their selected aggregate; secondary actions are excluded from a primary-only definition. Do not silently deduplicate separate conversion actions or relabel all-primary value as purchase-only revenue without actually applying that scope.
For Ads and GA4 comparisons, retain the selected event and source definitions. Lead categories can be grouped as leads, but a corresponding GA4 event needs to be selected for that site. Available event names can support a suggestion followed by confirmation. Primary calls and visible offline actions can matter where requested, without a universal automatic GA4 counterpart being assumed.
Calculate ratios from compatible full-precision base values and round only their displayed form. Total site revenue divided by advertising cost is blended ROAS or MER, which differs from source-specific purchase ROAS. The report should preserve this definition across narrative, chart and table. A sitelink audit can use commercial context without pretending to establish attribution or causal uplift from field completeness alone.
Apply dates and exact conditions consistently
The description fields are current configuration observations, while served-campaign coverage and performance use their declared period. Do not imply that a current incomplete description was missing throughout an entire historical window unless actual history supplies that evidence. A rolling thirty-day request is also different from last month. State the actual dates and account time zone for the performance population used in the review.
Requests for cost above, below or equal to a threshold require exact operators. At least includes equality, while greater than does not. Preserve combined AND and OR conditions in the report definition and apply them at the requested grain. A campaign condition should not silently become an asset condition simply because the resulting report shows sitelink rows beneath campaign headings.
Missing or unavailable values are not zeros. If an asset-level metric cannot be read, it cannot honestly satisfy a zero-conversion filter. Likewise, future dates do not contain completed observed data. Restrict selections appropriately and report the available period. These disciplines prevent empty technical results from being converted into broad business conclusions or inappropriate live-edit targets.
A requested graph should use one native chart where its grain and supported limits allow it. For longer periods, aggregate appropriately unless daily detail is explicit; if explicit daily detail requires panels, explain the native twenty-point limit. Label each series and scope clearly, display no more than two decimals, and retain the retrieved data for consistent follow-up tables rather than issuing another provider call solely to redraw it.
Follow an illustrative regional rollout
The following example is invented. Tomas Birch manages paid acquisition for Linden Workshop, a business with separate equipment and installation offers. The current audit reports an incomplete shared sitelink and a coverage question on a served regional campaign. Tomas first asks which unit each finding counts. The assistant distinguishes one unique applied asset from its several campaign associations and identifies the actual inspected population.
The existing link points to the central equipment page and remains appropriate there. The regional installation campaign needs a different service destination with confirmed availability. Tomas asks for a clear label and two descriptions based on the supplied service facts. The assistant offers creative alternatives without running Keyword Planner or claiming that the new wording has measured performance advantages.
Tomas selects one variant and provides the public installation page. The live proposal creates that sitelink and attaches it to the named regional campaign, retaining existing extensions. It does not edit the common equipment asset or remove its other associations. The readable approval shows the label, description pair, destination and exact campaign relationship before Tomas confirms execution.
The result verifies the new content and enabled campaign association. It does not claim that the link appeared, passed every policy review or recovered revenue. Tomas retains the separate review of the old incomplete shared asset and the website owner's confirmation of the installation page. The rollout completes a precise supported configuration task while preserving the remaining content and business responsibilities.
Sequence the actual resources and relationships
Asset creation and association have a resource dependency: the association must reference the asset that exists or is created in the same supported mutation. The local dependency model can represent that relationship without asking the provider to rediscover basic structure. Existing campaign and asset facts still require applicable scoped evidence. Structural knowledge and current inventory should not be confused with one another.
The dedicated combined operation uses an internal planned reference for the new asset and attaches it to the resolved campaign. Other requests can reuse an established existing asset reference. The client-facing steps remain ordinary language in both cases. Internal placeholders should never leak into a target label or appear as an undefined parent when the resource is being prepared for creation.
A broader request may contain several logical actions, such as creating a regional link, attaching another existing link and detaching an obsolete association. Each effect needs its own reviewable step and appropriate evidence. If the request exceeds the supported safety limit, ask the user to split it into smaller requests. Do not execute an arbitrary subset and leave the remainder hidden in the conversation.
Reconcile partial outcomes before continuing. If an asset exists after an uncertain result, inspect that specific resource and association rather than create a duplicate blindly. A later task should operate on the verified current state. Correct dependency handling includes retaining completed resources and explaining incomplete relationships, not only sequencing the happy path when every provider response arrives as expected.
Present the full proposal in clear steps
The approval should show the named campaign and each logical action in a visually distinct section. Creation includes the link label, both descriptions and final destination. Attachment identifies the existing asset and target relationship. Detachment identifies the relationship being removed and states that the asset remains. These details are meaningful to the user, unlike raw mutation JSON or internal field paths.
Describe retained scope where it affects the decision. The creation action retains existing extensions. An attachment retains the asset's content. A detachment retains the asset itself and other relationships. Do not bury those facts among technical details, and do not imply that a narrow campaign action edits the whole account. The owner should recognise exactly what will change and what relevant resources remain intact.
The confirmation notice states: No changes have been made yet. This proposal expires in 15 minutes. The live request runs only after Confirm and Execute is clicked. The enterprise approval workflow provides the wider context for this deliberate step. Preparation and applicable validation can happen before the click without becoming a live mutation.
After the click, show a prominent change-in-progress state and remove the available confirmation control while execution proceeds. Do not add a user mention to that interim message. The final result should arrive with a clear status and completed steps rather than leave the person wondering whether a small spinner registered the action. Readable interaction is part of making a precise proposal trustworthy.
Verify written fields and the exact association
The create-and-attach path reads back the newly created asset's written content fields and final URLs, then checks the enabled campaign association using the provider's returned resource reference. This is configuration verification for the supplied payload and relationship. It is different from the preparation checks that evaluate the request before the write. Both stages answer useful but separate questions.
For an existing attachment or detachment, the check compares the relevant association set with the expected result. The output should report that precise relationship outcome. It should not infer new content, global deletion or website behaviour from the association check. Verification is strongest when its declared scope matches the user's requested effect and the exact operation that was prepared.
A no-op should remain a no-op. If the asset is already attached or the selected enabled relationship is already absent, the response can explain the observed configuration without claiming that a new mutation occurred. This avoids confusing activity with completion. The enterprise's goal is the intended state, but the record should accurately describe whether the system changed anything to reach it.
After an ambiguous write or failed readback, reconcile before retrying. The asset or relationship may already exist despite a missing confirmation. A focused read can establish the actual result and prevent duplication. Report which fields or relationships could be verified and which remain uncertain. Do not replace that distinction with a broad failure message that encourages the owner to resend the same creation request immediately.
Report the result and remaining ownership clearly
A successful execution result should make Executed and verified prominent and list the actual completed effects underneath. Use asset content and campaign names rather than raw identifiers. A new sitelink and its association should be recognisable on separate rows. If only part of the request is confirmed, show that partial outcome accurately instead of using a success label that implies every step completed.
Keep client-facing evidence consistent across formats. A Slack summary, native chart, table or downloadable report should use the same population and definitions. A change in format should not add technical provider metadata, private contact details or unrequested monthly duplication. Provide enough information to assess the result and its limits, with deeper data reserved for a genuinely useful requested report.
Ordinary AI replies should place the user mention before the feedback text and controls, leaving feedback last. Use readable dates and actual weekly ranges for grouped periods, omit unnecessary current-year labels and show metrics with no more than two decimals. Clear presentation should reinforce the exact target, effect and evidence instead of making the enterprise interpret an opaque technical transcript.
The handover should preserve the website owner, temporary-offer review condition and any shared-content task left unresolved. An enterprise sitelink audit is valuable when it distinguishes completeness, association coverage, eligibility and commercial evidence. It can complete a precise supported change while keeping those separate responsibilities visible for the next review, without promising permanent relevance or guaranteed performance.