Turn exclusion findings into decisions the client can maintain
Google Ads negative-keyword audits for consultants: connect query evidence, commercial boundaries and approved changes
Explain the audit's presence check, review observed searches in the client's business context and prepare precise exclusions without treating every disappointing result as unwanted demand.
Agree the exclusion review's actual question
A consultant should establish what the client wants to decide before asking for negative-keyword data. “Do our campaigns have exclusions?” is a configuration question. “Should this search be blocked?” is a commercial judgement. “Where does this shared list apply?” is a relationship question. These can appear in one engagement, but answering one does not answer the others. A useful brief identifies the immediate decision, its target and the evidence needed to reach it.
Define a practical deliverable. The client might need a missing-presence list, a review of several observed phrases or an approved change to one campaign. A broad request to clean up traffic should become an agreed sequence of evidence and decisions before execution. This prevents a familiar audit failure label from becoming an excuse to add a generic list that changes the score without establishing relevance.
Confirm who can decide the commercial boundary. The person operating the account may understand campaign structure while a sales colleague understands which enquiries the business accepts. A consultant needs both kinds of context when they affect the proposal. Record the source of the business rule so that a later reviewer can distinguish a confirmed offer restriction from the consultant's hypothesis about an ambiguous search.
Keep the engagement proportionate. A question about one campaign does not require a complete account inventory, every historical search or every shared list. Start with the narrow evidence that answers the brief and broaden only when the unresolved decision requires it. This is faster for the client and reduces provider calls without substituting guesses for account facts.
Document the client's offer and exceptions
Write a short account of what the campaign sells and what it deliberately does not sell. Include service area, product compatibility, delivery method and relevant customer constraints. These facts give an exclusion a purpose. A search for a different product might be clearly unsuitable, while a differently worded search for the same service might be useful demand the client should retain.
Ask for exceptions before turning the description into a rule. A business might supply replacement parts only with an installation, offer trade pricing to eligible buyers or accept a small service as part of a larger appointment. Blocking a generic word associated with those situations can remove an exception the client values. The review should preserve the distinction between an unavailable standalone offer and an acceptable combination.
Separate permanent restrictions from temporary capacity. A discontinued service is different from a full diary this week. A missing stock item is different from a category the business never supplies. Temporary circumstances need a review condition and appropriate remedy; otherwise a short operational problem can become a lasting demand restriction that nobody remembers to revisit.
The assistant can help organise this supplied context without an API call. It can draft a concise offer brief or identify questions that need the owner's answer. That creative assistance becomes useful evidence only when the client confirms the relevant business facts. It should not invent an unavailable service or present a suggested qualification rule as an established account setting.
Understand the current presence audit
The current reader selects enabled Search, Shopping and Performance Max campaigns. It also reads campaign negative keywords, ad-group negative keywords and enabled negative-keyword-list attachments for enabled campaigns. It combines returned campaign identities from those exclusion surfaces and lists the selected campaigns without a recorded presence signal.
This is a presence check rather than a query-quality assessment. A missing signal identifies a campaign for investigation. It does not identify the words to exclude, calculate preventable cost or show that the business has received unsuitable enquiries. Those conclusions need their own evidence and commercial interpretation.
The audit's repair explanation requires account-specific search-term research instead of offering a universal automatic fix. That is an important capability boundary. A consultant should not manufacture negatives merely to produce a pass. The next useful step is to identify observed searches and decide which, if any, contradict the client's offer.
The selected campaign population also matters. A paused campaign is outside this enabled-campaign presence audit even if it has historical cost. A different campaign type can be outside the selected check. The finding therefore needs its stated scope rather than a claim that every campaign in the account has been comprehensively reviewed.
Keep presence and coverage separate
A single returned ad-group exclusion can supply a presence signal for its campaign. It does not demonstrate that every other ad group has appropriate protection. Likewise, a returned list attachment does not establish the usefulness of that list's contents. A pass says the audit did not find an enabled selected campaign lacking its recorded signal.
The present query does not require the ad group supplying a negative to be enabled. Explain that reader scope if the client asks why a campaign passes despite the active groups appearing empty. A consultant should examine the requested live group directly rather than use the campaign-level presence result as evidence of that group's current exclusions.
Completeness has several meanings here. The presence reader can finish successfully while the consultancy review of content remains unfinished. The client may still need a list-content check, an attachment review or a commercial judgement. Name the unresolved layer so that “complete audit” does not conceal an incomplete decision.
A useful follow-up is specific: show the exclusions for this ad group, show the selected list's contents or identify the campaigns using that list. Those requests target different resources. Keeping them separate makes the evidence easier to explain and prevents a campaign presence flag from being stretched into a certificate of account-wide relevance.
Handle failed reads honestly
The presence reader relies on four related queries. They run together, and a rejected request propagates rather than returning a fabricated empty result. That behavior matters because an empty missing-campaign result can otherwise look like a clean pass even when the provider evidence was never obtained.
An unavailable check should say that it could not be completed. It should not report no exclusions or complete coverage. The consultant can identify the affected evidence and retry through the supported read workflow when appropriate. A temporary provider failure is an availability issue until a successful response establishes the configuration.
A successful empty result has a different meaning. It can indicate that no selected enabled campaign lacks the presence signal, or that there are no campaigns within the selected population. The broader account context helps the reviewer interpret that result. It still does not prove that every observed search is appropriate or that the account has no commercial waste.
Preserve successful related evidence without treating it as a replacement for the failed part. A known campaign name can support a readable error message, but it cannot supply missing negative-list relationships. This keeps the conversation useful while respecting the boundary between a partially known target and an established configuration conclusion.
Work from returned query evidence
A reported search term is an observed phrase from the selected resource and period. It is different from a positive keyword, a proposed creative idea and a broad intent category. Keep the row's campaign context and measured values attached when reviewing it. Removing that context can make an otherwise sensible phrase look unsuitable for the whole business.
Do not promise a complete transcript of everything people searched. A report can have provider visibility limits, a selected resource population and a row cap. A displayed shortlist is not automatically the total eligible population. The assistant should preserve the evidence coverage and distinguish a count from examples whenever the client asks how many searches meet a condition.
An invented candidate can still be useful in brainstorming, but label it as a suggestion rather than retrieved traffic. The consultant might discuss which job-related themes could be irrelevant before obtaining actual observations. That conversation should not acquire fictional cost, conversion or impression values merely to make the suggestion seem more authoritative.
When the brief needs an observed shortlist, request the relevant dates, dimensions and metrics through the supported reporting framework. Reuse that dataset for a table or summary when it already contains the needed evidence. A format change should not cause the tool to retrieve the same period repeatedly or quietly substitute another resource with different coverage.
Separate unsuitable intent from weak results
A search can be commercially irrelevant even with very little measured activity. A business that never offers equipment rental may have a clear reason to exclude an observed rental enquiry from a sales campaign. The conclusion comes from the offer boundary. It does not need a claim that the search has already wasted a large amount of money.
A relevant search with disappointing performance needs a different review. The visitor may have encountered unclear pricing, an unsuitable destination or an incomplete measurement path. These are questions to investigate rather than explanations to invent. An exclusion closes that demand route; it may be the wrong remedy if the client still wants the underlying customer intention.
Keep sample size and outcome maturity visible. No recorded conversion in a short period is evidence about that report, not proof that the phrase can never produce a customer. A later sale may not yet be recorded, and the selected metric may not capture the business's accepted enquiry. Establish the intended outcome before treating a zero as a final commercial judgement.
Use a clear decision label and reason: retain because the offer fits, exclude because the service is unavailable, or investigate because the meaning remains ambiguous. This is more actionable than a numerical rank called bad traffic. The client can then approve a boundary it understands and evaluate the decision later against that stated purpose.
Challenge inherited exclusion assumptions
Common exclusion templates contain price, advice, recruitment and education themes. They are starting points for discussion, not universal business rules. A training provider may sell the course another business wants to block. A retailer may legitimately promote a free accessory with a paid order. The word alone does not establish the searcher's fit with this client's offer.
Review inherited entries alongside their original reasons when available. An exclusion can outlive a discontinued campaign structure, a former service restriction or a temporary stock problem. The consultant should identify whether the reason still applies rather than keep every existing entry because it has been present for years.
Channel routing deserves similar care. A support query might belong outside one acquisition campaign while remaining useful to another destination. A recruitment phrase might be unsuitable in a customer campaign and appropriate in a hiring effort. Turning a local decision into an account-wide rule can suppress legitimate activity elsewhere.
Document unresolved assumptions. If the client has not confirmed whether it accepts trade enquiries, do not quietly add a trade-related negative. Ask the narrow commercial question that changes the decision. This makes the review faster than debating every possible interpretation and safer than approving a broad template whose exceptions nobody examined.
Choose the intended matching boundary
Google's negative-keyword guidance distinguishes Search negative broad, phrase and exact matching. Broad considers all the terms in any order; phrase preserves their order with other words allowed; exact concerns the ordered query without additional words. Negative matching does not inherit positive-keyword expansion behavior, and synonyms or singular and plural forms need deliberate consideration.
The current convenience actions in ppctor add exact-match negatives to campaigns, ad groups or shared lists. They should not accept a phrase or broad instruction and silently substitute exact while claiming the requested broader restriction was applied. A consultant can offer a supported exact proposal, but the client must understand and select its narrower intended effect.
Distinguish one multi-word entry from several entries. “Equipment rental” as one negative is a different request from equipment and rental as separate objects. The proposal should state each intended text and its match type in readable language. Punctuation alone is an inadequate explanation for a client who does not work with matching rules every day.
Verification must concern the entry actually prepared. A successful exact-match readback cannot certify that every related expression will be blocked. The commercial review can identify further forms worth considering, but those need their own deliberate treatment. Keep the observed configuration result separate from a broader prediction about future search eligibility.
Resolve local ad-group targets correctly
An ad-group negative can enforce a local boundary while leaving demand available elsewhere. Start with the group's offer and destination. A query for buying a new item may be unsuitable for a repair group but valuable to a separate sales group. The local exclusion should follow that distinction rather than become an account-wide prohibition.
Resolve the ad group within its real campaign. Repeated group names are possible, and a follow-up such as “in that group” needs established conversation context. The confirmation should show the group and parent campaign by readable names. Internal identifiers stay available for precise execution without becoming client-facing labels.
Ask when ambiguity remains. If the client has two groups named Repairs in different campaigns, selecting the intended parent is part of the target decision. A consultant should not infer the parent merely because one group appeared earlier in an unrelated report. The tool needs canonical ownership, rather than a guessed relationship assembled from similar names.
The parent hierarchy also must remain valid. An ad group does not have another ad group as its parent. A keyword belongs under the appropriate group and campaign when that relationship is relevant. Clear labels make the proposal easier to review and expose resolution mistakes before a live request changes the wrong object.
Assess the campaign-wide effect
A campaign negative affects a broader boundary than an ad-group negative. Confirm that the exclusion belongs across the campaign's intended offers. The same phrase can be inappropriate for one group yet useful in another. The consultant should examine that distinction before choosing the wider resource simply because its name is easier to resolve.
Campaign structure can reflect geography, service, brand or another business purpose. Do not assume that a shared label means all campaigns should receive the same restriction. A campaign serving existing customers can have different acceptable intentions from one acquiring new customers. The client's actual structure gives the exclusion its scope.
An existing matching entry can make the requested addition unnecessary. The preparation code reads current relevant keywords, compares intended text and match type, and prepares missing entries. This supports a precise proposal rather than duplicate creation for every conversational request. An existing exact entry still does not prove the broader exclusion effect the user might have intended.
Keep campaign performance and configuration scope distinct. A high-cost search from a historical period does not establish that the campaign currently remains enabled or unchanged. The proposal needs current ownership and relevant settings. The historical row supplies decision evidence; the current read supplies the live target state.
Review shared-list contents
A shared negative-keyword list is a maintained collection, not proof that any particular campaign uses it. Review its name, type and contents according to the client's request. A descriptive name can suggest a purpose, but the actual entries establish what restriction the list contains.
Adding an entry to a shared list can affect the campaigns already using that list. The consultant should understand that propagation before recommending the change. A reusable list is useful when the commercial rule genuinely belongs across its attached population; it is risky when a local exception is mistakenly expressed as a shared rule.
The current shared-list convenience addition uses exact match and checks existing contents before preparing missing entries. Its proposal explicitly states that adding contents does not attach the list to a campaign. This distinction should remain visible in a multi-step plan rather than disappear behind a generic message that the negatives are now active everywhere.
Choose a maintainable list purpose. A mixed collection of unrelated historical fixes is harder for the next consultant to understand than a clearly owned rule set. The decision record can explain why the list exists and where exceptions belong. It should not claim that a tidy list name alone makes the contents commercially appropriate.
Review shared-list attachments separately
The attachment connects a shared list to a campaign. It is a different resource and action from creating the list or adding its keywords. A three-step setup can therefore mean create an empty list, add intended entries and attach it to the selected campaign. All three effects need clear presentation before one confirmation authorises execution.
The attachment adapter checks that the selected list is enabled and of the negative-keyword type in the same account. It reads the campaign's current enabled relationships and prepares an attachment or detachment only where the intended relationship differs. These checks establish the live structural prerequisites rather than relying on the list name alone.
Detaching a list from one campaign does not delete its contents or remove its relationships elsewhere. Explain that exact effect to the client. Likewise, adding another attachment broadens where the existing contents apply without rewriting those contents. A readable proposal should preserve the distinction between changing a rule and changing its audience of campaigns.
If the client asks where a list applies, that is a read request. It does not authorise detachment from campaigns the consultant considers unsuitable. Report the relationships and reasons for further review, then prepare any explicit requested edit separately. Keeping diagnosis and execution apart preserves a deliberate approval boundary.
Distinguish account-wide intent
An account-wide instruction is a broader business decision than a campaign attachment. Google documents account-level negatives for eligible search and shopping inventory in relevant campaign types. A shared list attached to selected campaigns should not be described as an account-level rule merely because several campaigns use it.
The presence audit reads its specified surfaces and should not become a universal certificate of every account-level exclusion mechanism. If the client asks specifically about an account-wide rule, select the relevant supported resource and explain the evidence scope. A campaign presence result cannot stand in for that distinct query.
Before proposing a broad boundary, ask whether any legitimate campaign needs the intention being excluded. A business might advertise both sales and training, or acquire customers and recruit staff separately. The decision needs confirmed exceptions rather than an assumption that one commercial purpose describes every account activity.
Capability constraints remain important. The exact-negative convenience actions target campaigns, ad groups and shared sets. A schema field being visible does not establish that the requested account-wide operation is prepared and verified through that convenience path. Explain the supported effect precisely instead of choosing a nearby operation and claiming equivalent account coverage.
Preserve inventory and channel context
A negative keyword is different from excluding a product in a listing-group tree. The keyword concerns a supported matching boundary, while the listing setting concerns product participation in its campaign and group context. The consultant should resolve which object the client actually wants to restrict before selecting an action.
Performance Max can have several asset groups with distinct listing structures. Standard Shopping uses ad groups. A product occurrence can be excluded in one enabled group and included in another. That state cannot be inferred from a campaign-level negative presence check or a search phrase containing the product's name.
Channel-specific reporting resources can also have different coverage. A returned category is not an individual observed query, and a keyword row is not a product row. Keep the resource meaning attached to the evidence. The assistant should not label an intent category as an exact customer search merely to populate a requested exclusion shortlist.
A complete commercial review may consider several remedies, including destination clarification, group organisation or a supported listing change. These remain different decisions. Choose the action that matches the confirmed problem and present its own target hierarchy, rather than call every restriction a negative-keyword fix.
Define the outcome behind performance rules
Before filtering queries by performance, specify the outcome being measured. All primary conversions, primary purchases and a named conversion action can produce different totals. The client should recognise the measure in the request, result and proposed decision. An unlabeled conversions column can conceal a scope change that alters the shortlist.
Standard Ads conversion value can include assigned value from primary non-purchase actions, including calls or leads. It is not automatically purchase revenue. Several primary actions contribute within the chosen definition; secondary actions remain outside a primary-only measure. The consultant should not silently remove primary contributions because another audit recommends a simpler configuration.
An Ads lead comparison with Analytics needs a corresponding GA4 event selected when its meaning is account-specific. The assistant can suggest likely names and seek confirmation. A named action can be compared with a specified event, while broader all-primary conversions and selected key events need their clearly stated aggregate definitions.
All-traffic revenue divided by advertising cost is blended ROAS or MER. That can be useful context, but it is broader than a Google-Ads-only return measure. Keep the definition visible rather than treat a favorable aggregate ratio as proof that each observed query is commercially suitable.
Make thresholds and logic exact
“Cost above 100” excludes the boundary value; “cost at least 100” includes it. “No conversions and cost above 100” differs from a rule joined by OR. The framework should preserve the requested comparisons and logical grouping instead of simplifying every condition into a familiar underperformance template.
Ratios need valid base metrics. Undefined return from zero cost is not a measured zero return, and unavailable conversion evidence is not no conversions. Preserve those states so that missing data does not qualify a phrase for exclusion. Display rounding should not alter the underlying value used to evaluate a threshold.
Use the appropriate population and completeness for counts. A shortlist of the ten highest-cost examples does not establish how many total rows meet the rule. The report definition should carry filters and coverage through aggregation and presentation. If the selected evidence cannot establish the total, explain that constraint rather than promote a preview count into a complete answer.
A numerical rule identifies candidates, not commercial approval. The consultant should still examine whether the client accepts the intention and whether the selected outcome is mature enough to support the decision. Keeping filter qualification separate from the final judgement makes the shortlist useful without turning analytics into an automatic exclusion policy.
Separate historical evidence from current state
Use the account's time zone and the requested period boundaries. Last thirty days is not last month, and a completed-day range should not silently omit another available day. The chart and table should retain the same resolved dates so that the consultant and client are discussing one evidence population.
Historical traffic can belong to a campaign that is paused today or has since changed its offer. An exclusion added now does not rewrite those earlier rows. Explain the current target and the period that supplied the recommendation as separate facts. This avoids claiming that a live setting read reconstructs the entire historical configuration.
A recent period can have incomplete outcome maturity. An enquiry received near the end may become a recorded sale later. The consultant should inspect that business process where it affects the decision rather than treat every recent non-converter as permanently unproductive demand.
Record the change date for later evaluation. A comparison should maintain the outcome definition and identify relevant offer or measurement changes. The client can learn from subsequent results without attributing every difference to the exclusion alone. Configuration verification and commercial evaluation use different evidence and should remain distinct.
Follow an illustrative consultancy decision
This example is invented. Consultant June Ellis reviews the Workshop Equipment campaign for Pine Workshop Supply. The business sells equipment and replacement components but does not rent equipment. The audit flags missing recorded negative coverage, and the owner asks June to investigate before changing the account.
The selected observed-query report includes a rental-intent phrase and a parts-related phrase. The owner confirms that rental is outside the offer, while parts are commercially useful even though that row has no recorded conversion in the selected period. June keeps the latter as an investigation item rather than automatically blocking both phrases.
The owner requests an exact exclusion for the observed rental phrase in Workshop Equipment. June presents the campaign name, exact text and match boundary. She does not describe the action as blocking every possible rental-related expression. The proposal follows the confirmed commercial restriction and the supported operation.
After approval, a successful relevant readback establishes the added configuration. The handover records the reason, scope and date, while the parts-related question remains open for destination and outcome review. The audit can later observe presence without becoming proof of perfect query quality or a promised cost saving.
Use creative discussion and focused reads
The client can ask the assistant to brainstorm possible irrelevant themes from a supplied offer brief. That is conversational work. It does not need Keyword Planner or an account-wide performance scan unless the user asks for provider-backed demand or actual measured evidence. Suggestions should remain visibly distinct from retrieved facts.
When the client asks which observed searches meet a condition, a tool read is appropriate. Select the narrow period, population and fields needed for the answer. The same dataset can support a concise summary, table and subsequent format change if it already contains those facts.
Follow-up context should retain the established campaign and proposed phrase. “Add that one to this campaign” should resolve through the conversation's confirmed objects, then obtain relevant fresh state for the live proposal. It should not forget the target and send the client through an unrelated creative workflow.
Reuse is bounded by the question. A different period, changed filter or current-state edit prerequisite can justify another read. Rewording a recommendation usually cannot. The goal is an evidence-driven conversation with efficient provider usage, rather than a new full audit for every short reply.
Prepare changes through shared capabilities
The shared action catalogue describes valid targets and parameters. Local dependency definitions establish structural relationships without provider calls merely to discover that a shared-list attachment needs a list and campaign. Actual ownership and current settings still need relevant live evidence before an approved mutation.
For exact negative additions, preparation reads the current relevant entries and constructs missing changes. It records a verification check for the wanted keyword set. This separates selecting the business intention from building and checking the API request. A consultant can understand the former without reviewing raw mutation JSON.
Creating a list, adding its contents and attaching it are distinct logical actions even when they serve one overall objective. Their dependency order should follow the available local definitions. A newly created resource needs its resolved result before a dependent operation can refer to it; that relationship is not a reason to perform redundant broad discovery queries.
A supported operation must preserve its documented effect. Exact-only convenience actions cannot become broad matching through explanatory wording. An account-level request cannot silently become a campaign attachment. When capability does not match the user's intention, clarify the available effect before preparing approval.
Present evidence without a technical wall
A client-facing shortlist should show the observed phrase, relevant campaign or group, selected performance evidence and plain-language reason for review. It does not need resource names, internal identifiers or query syntax. Keep enough hierarchy to identify the target while avoiding labels that imply nonexistent parents.
Use at most two displayed decimal places and readable dates. Current-year daily labels can omit the year; weekly ranges should use actual dates rather than an unexplained provider week code. A total or ratio should retain its meaning, and grouped ratios should be calculated from combined base metrics rather than averaged row ratios.
When a chart is requested, try one compatible native chart with up to three metrics or segments. This workflow allows twenty points per series. Longer periods can use a stated weekly or monthly breakdown unless the user explicitly requires daily detail; explicit longer daily output needs split native charts with an explanation of that limit.
The ordinary conversational response should put the user's mention before feedback so feedback remains last. A proposal should give each action a clear separate step. These presentation rules help the client understand what is evidence, what is a recommendation and what a confirmation will actually authorise.
Confirm distinct steps and exact scope
One confirmation can cover a supported sequence with each action prominently described. If the full instruction cannot be prepared within the safeguards, explain why it needs narrower proposals before writing. The app should not execute an arbitrary subset silently or hide an additional attachment behind a keyword-addition description. The client approves the complete displayed scope.
Each step should name the target and exact effect. Creating an empty list is different from populating it; populating it is different from attaching it. A campaign negative should show its campaign, and an ad-group negative should show its group and parent campaign. The client should not need to infer relationships from an arrow or a wall of technical prose.
The proposal states “No changes have been made yet. This proposal expires in 15 minutes.” Confirm and Execute authorises the presented scope. A report request, a suggestion discussion or a prior approval for another target does not authorise the new live mutation.
After the click, make the controls inactive and show prominent change-in-progress feedback without mentioning the user. Relevant state is rechecked before execution. This immediately acknowledges the interaction while allowing the tool to validate the exact current request instead of leaving the client unsure whether the button worked.
Verify settings and reconcile uncertain writes
Validation checks that the prepared request and its prerequisites are acceptable. Verification checks the resulting configuration through relevant readback. A request can be valid without having executed, and an accepted write can need further reconciliation before the app can report its complete result.
A successful negative addition should be checked for the intended text and match type at the intended resource. A successful attachment should be checked as a relationship to the selected campaign. Neither result proves that every future related search will be blocked or that the business has achieved a particular financial outcome.
Ambiguous and partial outcomes need reconciliation before replay. A list may have been created even if a later step could not be verified. Repeating the entire sequence can create duplicates or encounter an existing resource. State established changes and the remaining uncertain step, then inspect the affected state through the supported workflow.
The final success message should make executed and verified status prominent and describe each achieved effect clearly. If verification is incomplete, say what is known rather than declaring everything failed or retrying blindly. The client needs a reliable configuration record, especially when several dependent actions were approved together.
Evaluate the commercial effect separately
A verified configuration change is one milestone. The later commercial effect requires another period of observed results with the same relevant outcome definition. The consultant should not present the cost of an excluded historical phrase as guaranteed future savings, because future traffic and customer behavior can differ.
Review the original purpose. If the aim was to avoid enquiries for an unavailable service, examine that intention and the business feedback rather than judge success only by an overall conversion rate. A broader campaign ratio can change for many reasons unrelated to the single excluded phrase.
Keep opportunity cost visible in the discussion. A wide exclusion may reduce unsuitable traffic while also restricting useful demand. That possibility is why commercial exceptions and matching boundaries were reviewed before approval. Later evidence can support a revised decision, but it should not rewrite the original proposal's effect or conceal its scope.
A reversal is another requested action with its own supported preparation and approval. The original conversational change does not imply an automatic future undo. Maintain the decision date and review condition so the owner can ask for the appropriate next step when stock, service availability or strategy changes.
Leave a durable exclusion decision record
The handover should preserve the offer boundary, observed evidence period, selected phrase, matching rule, target scope and execution result. It should also identify unresolved commercial questions and the person who can answer them. Another consultant can then continue the review without reconstructing every conversation or inspecting raw API payloads.
For shared lists, record both the rule's purpose and its relevant attachment decisions. A future colleague needs to understand why a list serves several campaigns and where exceptions remain. Do not treat the mere existence of a list as proof that its current scope was deliberately approved for every possible offer.
Keep ongoing monitoring separate from a one-time review. A periodic discussion can revisit temporary restrictions and changed offers, but an ordinary confirmed edit does not install an unmentioned scheduled automation. Any maintained workflow needs its actual supported configuration and clear owner responsibilities.
The consultant audit guide provides the broader account-review context. A strong negative-keyword engagement joins the presence finding to real commercial evidence, chooses a precise supported boundary and leaves the client with a readable decision that can be maintained after the consultant's work is complete.