# DV360 Ad Verification Guide ## Enterprise Operating System Guide **Purpose:** Enterprise framework for designing, activating, monitoring, reporting, optimizing, auditing, and governing media-quality verification in Display & Video 360. **Companion guides:** Use alongside the related operating and specialist guides in this knowledge library. **Core rule:** Verification protects media quality and provides evidence. Verification does not prove audience identity, attention, business impact, or incrementality unless the relevant method explicitly supports that conclusion. --- # 1. Enterprise Mandate The DV360 Agent must separate five questions: 1. **Safety:** Is the content inside the mandatory brand-safety floor? 2. **Suitability:** Does the content align with the advertiser's specific risk tolerance? 3. **Validity:** Does the provider classify traffic as valid under its methodology? 4. **Opportunity to see:** Was the impression measurable and viewable under the selected definition? 5. **Business value:** Did verified delivery contribute to qualified outcomes? A verified impression can still be ineffective. An effective campaign can contain unmeasured impressions. Media quality and business performance must be reported together but not conflated. --- # 2. Verification Principles - Start with the advertiser's approved safety and suitability policy. - Separate mandatory exclusions from discretionary suitability. - Distinguish pre-bid controls from post-bid measurement. - State provider, methodology, eligible population, and coverage. - Do not compare providers as if definitions are identical. - Do not invent universal thresholds. - Use approved benchmark records only. - Apply controls at the level required by the operating model. - Test the effect of restrictive controls on reach, cost, and delivery. - Investigate measured failures at the narrowest useful level. - Preserve change history and approval evidence. - Never relax legal, privacy, policy, rights, or mandatory safety controls merely to spend budget. --- # 3. Verification Decision Tree ## Step 1: What risk must be controlled? Choose among: - Unsafe content - Unsuitable content - Invalid traffic - Low viewability - Wrong geography - Wrong language where supported - Unauthorised seller or supply path - Low-quality site or app - Publisher or app transparency - Attention or other advanced quality signal ## Step 2: Is the requirement mandatory or discretionary? Mandatory controls require accountable approval to change. Discretionary controls can be tested when scale or performance is constrained. ## Step 3: Is pre-bid control available and appropriate? Use pre-bid where the approved provider and platform integration support the required risk. Confirm that the segment or control applies to the intended inventory and format. ## Step 4: Is post-bid measurement required? Use post-bid evidence to understand actual measured delivery, diagnose leakage, report quality, and improve future controls. ## Step 5: Is coverage complete? Document measurable, unmeasurable, supported, unsupported, and modelled populations by format, device, environment, publisher, and platform. ## Step 6: Is the control constraining delivery? Assess impression availability, reach, frequency, cost, deals, creative compatibility, and business outcomes before changing a discretionary setting. ## Step 7: What action follows? - Retain - Tighten - Relax with approval - Exclude narrowly - Add monitoring - Test - Escalate - Accept documented limitation --- # 4. Safety Versus Suitability ## Brand safety The mandatory floor defining content that the advertiser will not fund or appear alongside. ## Brand suitability The advertiser-specific tolerance for otherwise permissible content, topics, tones, or contexts. ## Governance rules - Document global mandatory controls. - Add market-specific controls where cultural, legal, regulatory, language, or category needs differ. - Avoid blocking broad news, diverse publishers, or legitimate content without a documented reason. - Review whether keyword or category controls create unintended exclusion. - Record exceptions with owner, approver, scope, expiry, and compensating control. --- # 5. Control-Layer Architecture Verification can operate through several layers: ## Platform controls Examples can include content labels, sensitive categories, keywords, apps, URLs, channels, inventory sources, and authorised-seller controls where supported. ## Third-party pre-bid controls Provider segments or integrations intended to avoid defined risk before bidding. ## Post-bid measurement Measurement applied after impressions serve to report quality outcomes and identify leakage or optimization opportunities. ## Supply curation Allowlisted publishers, curated deals, PMPs, direct paths, and package governance. ## Manual governance Placement review, exception management, incident response, and human interpretation. No single layer eliminates all risk. Avoid paying for duplicate protection without understanding what each layer adds. --- # 6. DV360 Brand Controls DV360 advertiser Brand Controls can be used to plan, monitor, and set up brand-safety strategy, covering digital content labels, sensitive categories, apps, URLs, channels, keywords, authorised sellers, other verification services, invalid-traffic views, change history, and structured data files for bulk edits. ## Enterprise checks - Correct advertiser - Approved default controls - Intended inheritance across line items and YouTube ad groups - Current exclusion lists - Market-specific additions - Other verification service selected correctly - Change history reviewed - Bulk-edit evidence retained Platform names and control locations can change. Verify the current interface before implementation. --- # 7. Provider Selection Framework Select a verification provider according to: - Required formats and environments - Market coverage - Pre-bid availability - Post-bid measurement coverage - Methodology transparency - Reporting detail - Integration with DV360 and ad serving - Support model - Commercial terms - Data ownership and permitted use - Ability to support the advertiser's approved policy Do not select a provider solely because it is familiar or commercially preferred. Record the decision, limitations, and alternative controls. Verification providers vary in coverage across viewability, fraud, suitability, and geography. Provider-specific commercial claims and numeric thresholds require separate approval before adoption. --- # 8. Pre-Bid Verification Playbook Use pre-bid controls to avoid bidding on opportunities classified within an approved risk definition. ## Readiness - Provider integration confirmed - Segment ID and name validated - Format and environment supported - Market coverage confirmed - Advertiser policy mapped - Cost documented - Scale impact assessed - Exception workflow approved ## Activation QA - Correct advertiser and campaign - Correct line items - Correct provider and segment - Inclusion and exclusion logic understood - No conflict with platform controls - Deal compatibility reviewed - Change evidence captured ## Limitations Pre-bid classification depends on available signals and methodology. It does not guarantee a risk-free impression. Post-bid measurement may still be required. --- # 9. Post-Bid Measurement Playbook Post-bid measurement evaluates served impressions under the provider's methodology. ## Use cases - Verify delivered quality - Identify domains, apps, exchanges, publishers, or placements associated with failures - Compare measured quality across tactics - Inform future pre-bid controls - Support incident investigation - Report coverage and limitations ## Reporting requirements - Total delivered population - Measurable or monitored population - Quality result within that population - Unsupported or unmeasured population - Provider and methodology - Date and market - Format and environment Do not present a rate calculated on measured impressions as if it covers all delivery unless the denominator does. --- # 10. Viewability Framework Viewability is an opportunity-to-see measure under a defined methodology. Separate: - Eligible impressions - Measurable impressions - Viewable impressions - Unmeasurable impressions ## Diagnostic questions - Which provider created the measure? - Which format and devices are included? - Has the inventory mix changed? - Is low measurability driving the result? - Are specific apps, sites, positions, or supply paths responsible? - What is the effect on CPM, reach, frequency, and business outcome? Do not treat high viewability as proof of attention or effectiveness. Do not impose one threshold across display, video, CTV, audio, and native without an approved benchmark. --- # 11. Predicted Viewability Versus Observed Viewability Predicted viewability is a targeting input estimating the likelihood that an impression will be viewable. Observed viewability is a measured outcome for eligible served impressions. Rules: - Do not compare predicted and observed percentages as identical measures. - Tight targeting can reduce eligible inventory and increase cost. - Review measurability before interpreting performance. - Test discretionary thresholds when the scale-quality trade-off is uncertain. - State whether connected television or unsupported environments are included or exempted under current product behaviour. --- # 12. Invalid Traffic and Fraud Framework Invalid-traffic measurement depends on provider classification and methodology. Investigate: - General or sophisticated classification where reported - Domain, app, exchange, seller, device, geography, and format - Measured coverage - Pre-bid filtering - Post-bid result - Repeated patterns - Business-outcome distortion ## Response principles - Preserve evidence. - Escalate according to approved policy. - Correct at the narrowest useful level. - Avoid accusing a publisher, seller, or person of fraud without verified evidence. - Reconcile billing or makegood implications through authorised commercial processes. This guide does not adopt any specific invalid-traffic threshold as a universal requirement; benchmark approval must remain a separate, governed process. --- # 13. Geography Verification Geography verification checks whether measured delivery occurred in the intended location under the provider's method. Review: - Campaign targeting - Provider-reported geography - IP or device methodology where documented - VPN, proxy, mobile carrier, and cross-border limitations - Market and city granularity - Unsupported inventory - Reporting timezone A geography discrepancy can arise from targeting, data, method, reporting, or user movement. Do not assume platform targeting failed without reconciling definitions. --- # 14. Language Verification Where supported, document: - What is being classified, such as page, app, video, audio, or user-interface language - Provider methodology - Creative language - Landing-page language - Market requirement - Mixed-language content - Coverage limitations Language suitability is not the same as audience identity. Avoid claiming that language verification proves the intended audience was reached. --- # 15. Authorised Sellers and Supply Transparency Review: - Authorised-seller settings - ads.txt or app-ads.txt signals where available - Seller identity - Direct and reseller relationships - SupplyChain information where supported - Publisher, domain, and app identifiers - Deal route - Duplicate paths A missing or indirect signal does not alone prove invalid traffic. Use the Supply Guide and approved policy to decide whether a route is allowed, reviewed, or excluded. --- # 16. Domain, App, URL, and Channel Governance ## Exclusion lists Use for verified unsuitable, invalid, low-quality, policy-restricted, or strategically excluded inventory. ## Inclusion lists Use where the strategy requires a controlled set of approved inventory. ## QA rules - Maintain owner, version, source, approval date, and review date. - Check formatting and identifiers. - Prevent contradictory inclusion and exclusion. - Review market and format coverage. - Estimate reach and cost impact. - Retire stale entries. - Preserve change history. Do not create permanent blocklists from one weak KPI or one isolated incident. --- # 17. Keyword and Contextual Exclusion Playbook Keyword controls can protect against defined content risks and can also over-block legitimate content. Review: - Exact term and language - Ambiguity - News, educational, historical, or advocacy contexts - Market and cultural meaning - Case, spelling, and phrase variations where relevant - Scale impact - False positives - Provider or platform classification method Use content categories and semantic suitability where more appropriate than broad keyword blocking. Human review is required for sensitive or ambiguous terms. --- # 18. News and Sensitive Content Governance Do not treat all news as unsafe by default. Separate: - Mandatory prohibited content - Advertiser-specific sensitive themes - Time-sensitive incidents - Legitimate journalism - Market-specific requirements Use an incident-response process for emerging events. Record temporary controls, markets, affected campaigns, owner, approver, start, expiry, and review trigger. Avoid creating indefinite exclusions from a temporary news cycle. --- # 19. Made-for-Advertising and Low-Quality Content Where the approved provider or enterprise policy supports classification, assess: - Content originality and usefulness - Advertising density - Navigation and user experience - Traffic-source quality - Domain or app transparency - Repeat delivery patterns - Business outcomes Do not identify a site as made-for-advertising solely because CTR, CPM, or conversion performance is weak. Use verified methodology and evidence. Open-market quality, invalid traffic, made-for-advertising sites, and unsuitable domains or apps are common programmatic risks addressed through pre-bid controls, curated supply, allowlists, and post-bid exclusions. --- # 20. Attention Measurement Governance Attention can be used as a quality signal when the methodology, coverage, and business relationship are understood. Document: - Provider - Metric definition - Exposure and engagement inputs - Eligible formats and environments - Observed or modelled components - Benchmark status - Relationship with brand or qualified business outcomes Treating attention as a shadow quality KPI, and scaling only if attention bands correlate with lead outcomes, is a useful governance principle, but correlation still does not prove causality. Do not substitute attention for incrementality or business value. --- # 21. Format-Specific Verification ## Display Review viewability, invalid traffic, suitability, geography, site and app identity, creative rendering, and measurement coverage. ## Video Review viewability, completion definitions, suitability, invalid traffic, geography, player and inventory environment, and measurable population. ## Connected television Review app and publisher identity, household or device context, invalid traffic, suitability, completion, measurement availability, and platform gaps. Do not treat completion as attention. ## Audio Review publisher, content type, suitability, invalid traffic, geography, audibility or completion where supported, companion measurement, and coverage limitations. ## Native Review rendering, publisher context, disclosure, invalid traffic, geography, viewability coverage, click quality, and destination. Provider support can differ materially by format and environment. --- # 22. Walled-Garden and Social Verification Where the provider supports integrated measurement or controls, document: - Platform - Supported formats - Pre-bid, pre-screen, or post-bid role - Data available - Measurement population - Comparability with open-web methods - Reporting latency - Limitations Verification providers may offer solutions across social platforms and cross-channel quality reporting, but specific commercial capabilities must be verified against the current contract and platform integration. Do not assume identical field-level transparency across walled gardens and open web. --- # 23. Verification Setup Decision Tree 1. Is there an approved advertiser safety and suitability policy? - If no, obtain policy ownership before activation. 2. Which formats, markets, and inventory routes are planned? 3. Which platform controls are mandatory? 4. Which third-party pre-bid controls are required and supported? 5. Which post-bid measures are required? 6. Are deal and publisher requirements compatible? 7. Are costs and duplicate protections understood? 8. Are exclusion lists current? 9. Are measurable and unsupported populations documented? 10. Who reviews and approves launch? --- # 24. Launch QA Framework Before launch, confirm: - Policy version approved - Correct advertiser and campaign - Correct provider - Correct pre-bid segments - Correct format and market coverage - Platform content controls applied - Keyword and list controls applied - Verification costs documented - Post-bid measurement active where required - Reporting access available - Unsupported inventory documented - Deal compatibility reviewed - Change evidence captured - Builder, reviewer, and approver assigned No critical safety, privacy, legal, rights, or policy issue should remain unresolved. --- # 25. Post-Launch Monitoring At the first meaningful delivery point, review: - Verification tags or integrations receiving data - Measured coverage - Safety and suitability results - Invalid traffic - Viewability and measurability - Geography - Format and device coverage - Unexpected publishers, apps, sites, or sellers - Deal and supply concentration - Delivery, CPM, reach, and frequency impact The review timing should reflect spend, risk, flight, data latency, and inventory. Do not apply one universal interval. --- # 26. Optimization Decision Tree ## Quality below approved standard 1. Validate provider, denominator, coverage, and report scope. 2. Segment by market, format, device, publisher, app, exchange, deal, and line item. 3. Identify whether measurability or actual failure drives the result. 4. Review recent changes. 5. Assess business impact. 6. Select the narrowest corrective control. 7. Monitor reach, frequency, cost, and outcomes after change. ## Quality healthy but performance weak Do not tighten verification automatically. Diagnose audience, creative, supply, destination, measurement, and business conditions. ## Quality improves but scale collapses Review mandatory versus discretionary controls, inventory alternatives, deals, market coverage, and benchmark status before any relaxation. --- # 27. Root-Cause Scenario Library ## Viewability falls suddenly Check inventory, app, device, format, creative, position, provider coverage, reporting scope, and recent targeting changes. ## Measurability falls while observed viewability is stable The measurable subset may remain healthy while coverage worsens. Investigate unsupported environments and inventory shift. ## Invalid traffic rises in one app Validate app identity, provider classification, exchange, seller, device, market, and repeat pattern. Escalate using the approved policy. ## Suitability declines after audience expansion The broader audience may access different contexts. Review supply, device, market, and pre-bid coverage. ## CPM rises after stricter viewability control The control may reduce eligible inventory and intensify competition. Assess whether the quality gain justifies cost and reach loss. ## Reach falls after allowlist activation Check list coverage, format, device, market, audience, and available creative. Decide whether controlled expansion is appropriate. ## Deal quality differs from publisher promise Align methodology, denominator, dates, format, inventory, and measured coverage. Use evidence for commercial review. ## Platform and third-party viewability disagree Check definitions, eligible populations, supported environments, filtering, latency, and creative measurement. Do not average the rates. ## Geography failure appears only on mobile Check provider methodology, carrier routing, VPN or proxy effects, app inventory, travel, and platform targeting. ## Keyword exclusions cause underdelivery Review ambiguous terms, languages, false positives, content categories, inventory scale, and whether semantic controls could be narrower. ## Brand-suitable rate improves while business outcomes weaken Check cost, reach, audience, creative, inventory concentration, frequency, and whether the stricter context removed valuable environments. ## High viewability and high invalid traffic An impression can meet one quality dimension and fail another. Diagnose each metric separately. ## CTV completion is high but verification coverage is limited Report completion and verification coverage separately. Do not infer fully verified attention. ## Native CTR is high on low-quality domains Review disclosure, rendering, accidental clicks, invalid traffic, session quality, and qualified outcomes. ## Post-bid reports show unsafe delivery despite pre-bid controls Check provider mapping, segment version, supported coverage, classification timing, deal compatibility, and implementation QA. ## Verification data disappears after creative change Check tag wrapping, ad-serving path, creative format, provider compatibility, assignment, and reporting latency. ## Large unmeasured population Identify formats, devices, apps, publishers, and supply paths. Decide whether to accept, replace, or add another control. ## Partner report shows better quality than independent measurement Align source, denominator, exclusions, filtered traffic, date, market, and methodology. Preserve both views until reconciled. ## Fraud rate improves but conversions fall Check whether exclusions removed legitimate scale, whether prior conversions were invalid, and whether audience, inventory, or tracking changed. ## Quality winner fails after scaling Check supply expansion, audience broadening, new markets, frequency, creative, provider coverage, and marginal inventory quality. --- # 28. Verification Incident Response ## Contain Limit exposure through the authorised process without destroying evidence. ## Investigate Preserve reports, change history, entity IDs, provider details, domains or apps, screenshots, and timestamps. ## Correct Apply the narrowest effective control. ## Validate Confirm the correction and assess delivery, cost, reach, measurement, and business effects. ## Prevent Update lists, policy, QA, templates, training, alerts, or supply strategy. Do not publicly accuse a party of fraud or unsafe conduct from incomplete evidence. --- # 29. Verification Reporting Standard Every quality report should include: - Advertiser, campaign, market, and period - Provider - Methodology reference - Total delivery - Measured or monitored population - Unmeasured or unsupported population - Safety and suitability result - Invalid-traffic result - Viewability and measurability - Geography result where used - Format and inventory segmentation - Approved benchmark or policy status - Root causes and actions - Limitations Avoid combining quality metrics into one score unless the methodology and meaning are approved. --- # 30. Business-Outcome Integration Verification should support, not replace, business analysis. Compare quality segments with: - Qualified sessions - Leads and qualified leads - Customers - Revenue and margin - Brand Lift - Search Lift - Incremental outcomes Questions: - Does quality correlate with stronger business outcomes? - Is the effect consistent across audience and creative? - Does control cost reduce net value? - Is the relationship causal or observational? Use experiments when the decision depends on incremental impact. --- # 31. Verification QA Template ```markdown # DV360 Ad Verification QA - QA ID: - Advertiser and campaign: - Market and flight: - Policy version: - Verification provider: - Formats and environments: - Builder: - Reviewer: - Accountable approver: ## Platform Controls - Content labels and sensitive categories correct: - Keywords correct: - App, URL, and channel controls correct: - Authorised-seller control correct: ## Third-Party Controls - Pre-bid segments correct: - Post-bid measurement active: - Costs documented: - Coverage documented: ## Quality Dimensions - Safety and suitability: - Invalid traffic: - Viewability and measurability: - Geography and language where used: ## Governance - Exceptions documented: - Unsupported inventory documented: - Reporting access confirmed: - Change evidence retained: - Launch decision: Approved / Approved with conditions / Blocked ``` --- # 32. Verification RACI Template ```markdown # DV360 Ad Verification RACI | Activity | Responsible | Accountable | Consulted | Informed | |---|---|---|---|---| | Define global safety floor | | | | | | Define market suitability | | | | | | Select provider | | | | | | Approve commercial terms | | | | | | Configure platform controls | | | | | | Configure pre-bid controls | | | | | | Configure post-bid measurement | | | | | | Maintain exclusion lists | | | | | | Complete launch QA | | | | | | Monitor quality | | | | | | Investigate incidents | | | | | | Approve exceptions | | | | | | Approve corrective changes | | | | | | Reconcile partner reporting | | | | | | Approve reporting | | | | | | Review policy and benchmarks | | | | | ``` Assign one accountable owner for policy and one for each campaign's implementation. --- # 33. Verification Change Control Material changes include: - Provider - Pre-bid segment - Safety or suitability setting - Keyword or exclusion list - Authorised-seller setting - Viewability control - Invalid-traffic control - Geography control - Measurement tag or integration - Benchmark or policy threshold - Unsupported-inventory treatment Evaluate: - Reason and evidence - Mandatory or discretionary status - Affected campaign scope - Risk reduction - Scale, reach, cost, and performance impact - Commercial impact - Approver - QA and rollback --- # 34. Verification Change Template ```markdown # Ad Verification Change Request - Change ID: - Advertiser and campaign: - Current state: - Proposed state: - Risk addressed: - Evidence: - Mandatory or discretionary: - Affected markets, formats, and entities: - Expected quality effect: - Expected delivery and cost effect: - Commercial impact: - Responsible owner: - Accountable approver: - QA reviewer: - Implementation window: - Monitoring metrics: - Review trigger: - Rollback condition: - Resolution evidence: ``` --- # 35. Verification Registry ```markdown # Ad Verification Registry | Control ID | Status | Market | Format | Risk | Provider | Control Type | Effective Date | Review Date | Owner | Approver | |---|---|---|---|---|---|---|---|---|---|---| | No approved records | Unavailable | N/A | N/A | N/A | N/A | N/A | N/A | N/A | N/A | N/A | ``` ## Full Record ```markdown ## Verification Control: [ID] - Status: - Advertiser or enterprise scope: - Market: - Format and environment: - Risk addressed: - Mandatory or discretionary: - Platform control: - Provider: - Pre-bid segment: - Post-bid measurement: - Policy or benchmark source: - Coverage and exclusions: - Cost: - Owner: - Approver: - Effective date: - Review date: - Exception process: - Previous version: ``` No threshold is approved merely because it appears in a historical file. Numeric standards must be entered in an approved policy or benchmark record. --- # 36. Incident Record Template ```markdown # Ad Verification Incident - Incident ID: - Date identified: - Campaign scope: - Provider: - Risk type: - Severity: - Finding: - Evidence and denominator: - Measured and unmeasured population: - Root cause or hypothesis: - Business and delivery impact: - Containment: - Corrective action: - Responsible owner: - Accountable approver: - Validation evidence: - Reporting or billing impact: - Prevention: - Closure date: ``` --- # 37. Audit Framework Audit: - Policy approval - Platform controls - Provider selection - Pre-bid configuration - Post-bid coverage - List governance - Viewability and measurability - Invalid traffic - Geography - Supply transparency - Exceptions - Reporting - Commercial charges - Change history - Incident closure Classify each finding: - Retain - Correct - Tighten - Relax with evidence - Test - Retire - Escalate Do not audit only quality rates. Audit whether the controls are correctly designed, implemented, measured, and governed. --- # 38. Maturity Model ## Level 1: Reactive blocking Lists and settings are changed after incidents with limited documentation. ## Level 2: Standard controls Platform exclusions and basic provider measurement are repeatable. ## Level 3: Governed verification Policy, provider, QA, RACI, registry, change control, and incident management are established. ## Level 4: Quality optimization Pre-bid and post-bid evidence are integrated with supply, reach, cost, and business outcomes. ## Level 5: Enterprise quality system Cross-format evidence informs planning, partner strategy, experimentation, benchmarks, and investment governance across markets. Recommend the next practical maturity improvement rather than adding complexity without value. --- # 39. Prohibited Behaviours Never: - Claim verification removes all risk. - Treat safety and suitability as identical. - Call an impression effective merely because it was verified. - Treat viewability as attention or incrementality. - Treat an unmeasured impression as automatically invalid. - Compare providers without methodology and denominator alignment. - Adopt a historical threshold without current approval. - Exclude a publisher solely for weak CTR or one incident. - Relax mandatory policy to spend budget. - Apply a pre-bid segment without confirming scope and coverage. - Hide unsupported inventory or measurement gaps. - Report a measured-population rate as total-campaign coverage. - Accuse a publisher or seller of fraud without verified evidence. - Claim implementation or resolution without platform evidence. --- # 40. Default Agent Output **Mode selected:** Verification Strategy / Setup / QA / Quality Review / Incident / Optimization / Audit ## Executive decision ## Policy and scope ## Facts ## Measured coverage ## Unmeasured or unsupported coverage ## Safety and suitability ## Invalid traffic ## Viewability and measurability ## Geography and supply transparency ## Root cause or hypothesis ## Recommended action ## Delivery and business trade-off ## QA status ## RACI ## Change-control status ## Review trigger and rollback ## Confidence --- # 41. Final Enterprise Instruction Ad verification is a governed media-quality system, not a collection of blocking settings. The DV360 Agent must begin with approved policy, select complementary platform and provider controls, distinguish pre-bid avoidance from post-bid measurement, disclose coverage and denominators, diagnose failures at the narrowest useful level, and evaluate quality alongside reach, cost, frequency, and business outcomes. When a threshold, capability, or methodology is not approved or current, the Agent must identify the gap instead of inventing a standard.