# DV360 Reporting & Insights Guide ## Enterprise Operating System Guide **Purpose:** Enterprise framework for extracting, validating, interpreting, communicating, governing, and improving reporting from Display & Video 360. **Companion guides:** Use alongside the related operating and specialist guides in this knowledge library. **Core rule:** Reporting explains what happened. Insight explains why it matters. Recommendation defines what should happen next. Attribution assigns credit. Experimentation estimates causal impact. These jobs must not be confused. --- # 1. Reporting Mandate The DV360 Agent must turn reliable evidence into clear decisions without overstating certainty. Every report should answer: 1. What happened? 2. What changed? 3. What caused or may have caused the change? 4. Why does the change matter to the business? 5. What action is recommended? 6. What evidence, owner, guardrail, and review trigger support the action? A report is incomplete when it gives metrics without interpretation. An insight is incomplete when it gives interpretation without evidence. A recommendation is incomplete when it lacks ownership, expected effect, risk, and review logic. --- # 2. Reporting Principles - Start from the approved business objective. - Use the correct source of truth for the question. - Separate observed facts from calculations, assumptions, interpretations, and recommendations. - Reconcile material differences between DV360, Campaign Manager 360, analytics, CRM, sales, finance, publisher, and verification sources. - State attribution models, conversion windows, counting methods, currencies, timezones, and cost bases. - Use approved benchmarks only. - Report business quality, not merely platform volume. - Explain uncertainty and missing evidence. - Limit recommendations to the few highest-impact decisions. - Never claim causality from correlation or attribution alone. --- # 3. Reporting Decision Tree ## If the stakeholder asks "What happened?" Use validated reporting for the approved date range, source, scope, and metric definitions. ## If the stakeholder asks "Why did it happen?" Use diagnostic analysis. Segment by campaign structure, audience, supply, creative, device, market, time, frequency, and measurement changes. ## If the stakeholder asks "What should we do?" Use the Optimization Guide. Provide evidence, expected trade-off, owner, guardrail, review trigger, and rollback condition. ## If the stakeholder asks "Which touchpoint receives credit?" Use attribution reporting. State the model, eligible interactions, Floodlight, and windows. ## If the stakeholder asks "Did media cause the outcome?" Use an appropriate incrementality method. Do not answer from attributed conversions alone. ## If the stakeholder asks "Did the partner deliver?" Use ad-server, platform, partner, deal, and verification evidence. Reconcile definitions before judging delivery. --- # 4. Source-of-Truth Framework ## Business outcomes Use the advertiser's validated CRM, commerce, booking, application, subscription, sales, or finance system for qualified outcomes, fulfilled sales, net revenue, margin, and customer quality. ## Website or app activity Use the approved analytics or mobile measurement platform for sessions, engagement, product exploration, funnel steps, and app events. ## Cross-channel ad serving and attribution Use Campaign Manager 360 and Floodlight for shared activities, ad exposure and click history, assisting touchpoints, and conversion credit under configured rules. ## DV360 operations Use DV360 for spend, pacing, impressions, reach, frequency, inventory, audience, creative, bids, and platform-attributed outcomes. ## Verification Use the approved verification source for its supported viewability, invalid-traffic, suitability, and geographic measures. ## Reporting rule The source of truth is objective-dependent. A single platform should not be declared the universal truth for every business and media question. --- # 5. Reconciliation Framework Differences between systems are expected. The reporting job is to explain them. Check: - Reporting period - Timezone - Currency - Cost basis - Attribution model - Click and view windows - Conversion event - Counting method - Cross-device or modelled treatment - Consent and observability - Data latency - Refunds and cancellations - New versus existing customers - Gross versus net revenue ## Reconciliation workflow 1. Define the two values being compared. 2. Confirm date range, timezone, and filters. 3. Confirm metric and conversion definitions. 4. Confirm attribution and windows. 5. Confirm cost and currency. 6. Check tagging, consent, and recent releases. 7. Quantify the difference. 8. Separate explained and unexplained components. 9. Assign an owner to unresolved gaps. 10. State which source governs the decision. Do not force systems to match when they measure different populations or rules. --- # 6. Reporting Architecture ## Executive report Purpose: support leadership decisions. Include: - Business outcome - Material performance change - Key reason or hypothesis - Principal risk - Recommended decision - Measurement confidence ## Operational report Purpose: manage delivery and optimization. Include: - Budget and pacing - Delivery health - Audience - Supply and deals - Creative - Reach and frequency - Quality - Floodlights and business outcomes - Actions and owners ## Measurement appendix Include: - Source definitions - Attribution and windows - Conversion taxonomy - Reconciliation notes - Benchmarks and approval status - Experiment design and limitations ## Campaign closure report Include final delivery, reconciled outcomes after lag, commitments, learnings, unresolved issues, and implications for the next plan. --- # 7. Reporting Cadence ## Daily control view Use for high-risk delivery, pacing, status, tracking, creative approval, deal participation, and critical anomalies. ## Weekly performance view Use for business KPI movement, audiences, supply, creative, frequency, bids, budget, and approved actions. ## Monthly business review Use for trends, marginal returns, partner and deal value, measurement reconciliation, experiments, and strategic changes. ## Quarterly review Use for portfolio learning, incrementality, market differences, creative pipeline, operating maturity, and future investment. Cadence should reflect spend, flight, risk, conversion lag, commitments, and stakeholder needs. Do not impose one universal schedule. --- # 8. Report Design Standards A strong report should be: - Accurate - Comparable - Concise - Decision-oriented - Transparent - Traceable - Appropriate for the audience ## Page hierarchy 1. Executive decision 2. Business outcome 3. Key drivers 4. Risks and unknowns 5. Recommended actions 6. Supporting detail 7. Methodology appendix Do not place a large metric table before explaining what the stakeholder needs to know. --- # 9. Facts, Calculations, Assumptions, and Interpretation ## Fact Directly observed from a validated source. ## Calculation Derived from stated inputs and formula. ## Assumption A value or interpretation used because direct evidence is unavailable. ## Interpretation A reasoned explanation supported by evidence but not directly observed. ## Recommendation A proposed action based on the finding, implication, and risk. The Agent must label these categories. Do not present interpretation as fact. --- # 10. Insight Generation Framework Use this sequence: ## Finding What materially changed or differs? ## Evidence Which metrics, segments, periods, and sources support the finding? ## Root cause What factor caused or may have caused the movement? ## Business implication Why does the finding matter to revenue, customers, reach, risk, efficiency, or learning? ## Recommendation What should happen next? ## Confidence How strong is the evidence? ## Example structure - **Finding:** Qualified conversion rate declined after a delivery shift. - **Evidence:** The decline is concentrated in a specified audience and supply segment. - **Interpretation:** The change may reflect weaker audience or inventory quality. - **Implication:** Additional spend may increase volume without equivalent business value. - **Action:** Hold further scaling, validate the source, and run a controlled comparison. --- # 11. Insight Quality Test Before publishing an insight, ask: - Is it material? - Is it new or decision-relevant? - Is the evidence sufficient? - Is the comparison fair? - Does it explain a driver rather than restate a metric? - Does it connect to the business objective? - Can the stakeholder act on it? - Is uncertainty disclosed? "CTR increased" is an observation. "CTR increased after a new creative entered delivery" is a relationship. "The creative caused the increase" requires stronger evidence. --- # 12. Recommendation Writing Framework Every recommendation should contain: - Action - Evidence - Business rationale - Expected direction - Trade-off - Owner - Approver - Guardrail - Review trigger - Rollback condition Avoid vague recommendations such as: - Optimize further - Monitor performance - Increase budget - Refresh creative Replace them with a scoped action and decision condition. --- # 13. Executive Summary Framework The executive summary should answer: - Are business and media outcomes on track? - What is the most important change? - What is driving it? - What is the main risk or limitation? - What decision is required? ## Leadership audience variants ### Marketing leadership Emphasise customer, growth, brand, and channel role. ### Finance Emphasise spend, value, margin, forecast, and risk. ### Procurement Emphasise commitments, fees, delivery, partner value, and commercial transparency. ### Activation teams Emphasise root cause, operational action, dependencies, and review timing. Do not use one identical summary for every stakeholder. --- # 14. Metric Interpretation: Cost ## CPM Use to understand media cost per thousand impressions under a stated cost basis. CPM can change because of market, seasonality, audience restriction, format, device, inventory quality, floors, deals, bids, and competition. CPM does not show whether media produced valuable outcomes. ## CPC Use to understand cost per recorded click. CPC combines CPM and CTR effects. A low CPC may reflect relevant engagement, click-heavy placements, accidental clicks, or weak-quality supply. ## CPA and CPL Use only with an approved conversion definition and cost basis. A lower CPA can hide weaker lead quality, lower margin, retargeting concentration, or excess attribution credit. ## ROAS State revenue and cost definitions. ROAS does not equal profit and may not reflect refunds, cancellations, margins, or new-customer value. --- # 15. Metric Interpretation: Delivery and Engagement ## Impressions Impressions indicate served opportunities under the platform definition. They do not prove attention or unique reach. ## Click-through rate CTR indicates recorded clicks relative to impressions under the selected definitions. It does not prove qualified interest, destination quality, conversion, or incremental impact. ## View rate State the format-specific view definition. Do not compare unlike views. ## Completion rate Completion is affected by duration, skippability, autoplay, device, and inventory. It does not prove comprehension or persuasion. ## Viewability Separate eligible, measurable, viewable, and unmeasured populations. Higher viewability does not automatically create better business performance. --- # 16. Metric Interpretation: Reach and Frequency Before reporting reach, state whether it represents: - User - Device - Household - Account - Deduplicated population - Modelled estimate Review frequency distribution, not only average frequency. ## Useful interpretations - Spend growth with flat reach can indicate saturation or restriction. - Rising reach with stable frequency can indicate scalable expansion. - Rising reach and rising frequency require distribution analysis. - High average frequency may be caused by a small heavy-exposure group. Do not sum incompatible reach figures across platforms. --- # 17. Awareness and Brand Reporting Primary evidence may include: - Reach - Frequency - Viewable or completed exposure - Brand Lift - Search Lift - Incremental reach Supporting evidence may include cost and quality. Do not judge awareness solely by clicks or last-click conversions. Do not claim brand impact without a valid study or approved proxy interpretation. --- # 18. Consideration Reporting Primary evidence may include: - Qualified sessions - Engaged visits - Product exploration - Journey starts - Consideration lift - Search behaviour Supporting evidence may include CTR, CPC, view rate, and creative engagement. Separate media click quality from landing-page and journey quality. --- # 19. Lead Generation Reporting Report the funnel: - Platform-attributed lead - Analytics submission - CRM-created lead - Qualified lead - Opportunity - Sale or approved customer Use validated denominators for stage conversion rates. Do not call raw form submissions qualified leads. Report the latency between submission and qualification. --- # 20. Commerce Reporting Compare: - Platform transactions - Analytics purchases - Order-system orders - Fulfilled or net orders - Gross and net revenue - Margin where approved - New versus existing customers State refunds, cancellations, currency, tax, and revenue basis. --- # 21. Audience Reporting Playbook For every material audience, report: - Role - Source - Spend - Reach and frequency - Inventory and creative mix - Qualified outcomes - New-customer contribution where available - Data cost - Measurement confidence Do not rank audiences without controlling for role, media conditions, and attribution. Keep prospecting and retargeting separate. --- # 22. Supply and Deal Reporting Playbook Review: - Exchange or SSP - Publisher - Domain or app - Deal - Direct or reseller route - Format and device - Cost - Quality - Reach contribution - Qualified outcome A higher-cost deal may provide unique access, certainty, context, audience, or quality. Report whether the promised advantage was delivered. Do not describe premium supply as effective without evidence. --- # 23. Creative Reporting Playbook Report creative at four levels: - Concept - Execution - Format adaptation - Element Review: - Delivery opportunity - Audience and inventory mix - Reach and frequency - Viewability or completion - Landing continuity - Qualified outcomes - Brand outcomes - Rights and expiry Do not declare a creative winner when rotation or delivery conditions are materially unequal. --- # 24. Format-Specific Reporting ## Display Report placements, supply, audience, creative size and type, viewability, frequency, destination quality, and qualified outcomes. ## Video Report format, duration, view definition, quartiles, completion, reach, frequency, inventory, creative, and brand or business outcomes. ## Connected television Report app, publisher, deal, device or household reach method, frequency, completion, incremental reach where measured, and cross-device limitations. ## Audio Report inventory type, publisher, reach, frequency, completion where supported, creative, companion response, and brand or business outcomes. ## Native Report rendering, publisher context, component performance, viewability coverage, click quality, destination, and qualified outcomes. --- # 25. Attribution Reporting Every attribution report should state: - Model - Floodlight activity - Counting method - Click window - View window - Eligible interaction types - Cross-device or modelled treatment - Reporting source Separate: - Click-through conversions - View-through conversions - Other eligible interaction types - Total attributed outcome Attribution reports answer how credit is assigned. They do not establish counterfactual impact. --- # 26. Path Reporting Path analysis may show: - Interaction sequence - Impression and click roles - Channel combinations - Path length - Time lag - Assisting touchpoints Use paths to form hypotheses about demand creation and capture. Do not assume the first touch caused the conversion or that each recorded touch was necessary. --- # 27. Incrementality Reporting Clearly separate: - Observed outcomes - Attributed outcomes - Measured incremental outcomes For every experiment, report: - Hypothesis - Control and treatment - Assignment method - Primary outcome - Guardrails - Study period - Result - Uncertainty - Contamination - Scope of inference Do not translate a directional or inconclusive study into a definitive business claim. --- # 28. Benchmark Reporting Use only benchmark records with an approved status and valid review date. Before comparison, validate: - Market - Currency - Objective - Format - Environment - Audience role - Buying route - Cost basis - Attribution - Conversion definition - Reference period - Confidence A benchmark variance triggers diagnosis, not an automatic optimization. --- # 29. Variance Analysis Report variance using: - Absolute change - Percentage change - Contribution to total change - Appropriate baseline - Materiality Possible baselines: - Plan - Approved benchmark - Previous comparable period - Control cell - Forecast Do not use a previous period when seasonality, budget, market, offer, measurement, or structure makes the comparison misleading. --- # 30. Root-Cause Scenario Library ## High CPM Check market, seasonality, audience restriction, inventory quality, floors, deals, device, format, bids, and competition. ## Low CTR Check objective, creative, audience, placement, device, rendering, and whether clicks are strategically relevant. ## High CTR with poor engagement Check clickbait, accidental clicks, intrusive supply, slow landing pages, tracking, and expectation mismatch. ## Low conversion rate Check audience intent, destination, offer, product availability, form or checkout, tracking, and conversion definition. ## High CPA Decompose cost, conversion rate, volume, attribution, customer quality, and lag. ## Weak ROAS Check revenue basis, cost basis, product mix, margin, refunds, cancellations, and customer type. ## Strong platform results with weak CRM Check event quality, view-through share, attribution windows, duplicate firing, qualification criteria, and CRM latency. ## Weak platform results with strong CRM Check attribution gaps, cross-device journeys, offline outcomes, tagging coverage, and business-system timing. ## Low reach Check audience size, inventory, geography, device, format, bids, frequency, and exclusions. ## High frequency Check audience overlap, retargeting pools, inventory concentration, separate caps, flight, and budget growth. ## High viewability with weak outcomes Check audience relevance, creative, cost inflation, reach loss, frequency, and destination. ## High completion with weak outcomes Check forced exposure, duration, branding, message, audience, inventory, and measurement. ## Prospecting appears inefficient Review new-customer outcomes, assisted contribution, qualified engagement, reach, and incrementality. ## Retargeting appears highly efficient Review saturation, existing demand, suppression, view-through attribution, and marginal conversion. ## PMP costs more than Open Auction Report the promised and delivered value: unique access, priority, certainty, context, audience, or quality. ## One publisher dominates outcomes Check spend, audience, creative, route, device, attribution, and customer quality before recommending concentration. ## Budget increase weakens efficiency Check marginal headroom, inventory quality, audience expansion, frequency, bids, and conversion lag. ## Results change after creative launch Separate creative effects from allocation, audience, inventory, destination, and seasonal changes. ## Conversions drop after consent change Separate reduced observability from actual business demand using analytics, CRM, and implementation evidence. ## Report totals change unexpectedly Check filters, dimensions, metric compatibility, timezone, attribution, data latency, and platform field changes. --- # 31. Insight Prioritization Prioritize insights by: - Business impact - Evidence strength - Urgency - Scalability - Risk - Actionability Use three categories: ## Now Critical corrections or decisions requiring immediate action. ## Next Structural improvement requiring preparation or approval. ## Test Uncertain opportunity requiring controlled evidence. Avoid more than five major recommendations in one executive report unless completeness is explicitly required. --- # 32. Reporting QA Framework Before publication, confirm: - Date range and timezone - Advertiser and campaign scope - Currency and cost basis - Filters - Metrics and dimensions - Formula calculations - Attribution model and windows - Conversion event and counting method - Reach, view, and completion definitions - Benchmark status - Analytics and CRM reconciliation - Partner and verification definitions - Facts, assumptions, and interpretations labelled - Confidentiality and privacy - Recommendations supported by evidence --- # 33. Reporting QA Template ```markdown # DV360 Reporting QA - Report ID: - Report title: - Reporting period: - Advertiser and campaign scope: - Audience: - Author: - Reviewer: - Approver: - Primary source of truth: - Supporting sources: - Currency and cost basis: - Timezone: - Attribution model: - Click window: - View window: - Conversion definition: - Benchmarks used and status: - Reconciliation completed: - Calculations validated: - Known limitations: - Assumptions: - Confidentiality status: - Critical issues: - Approved corrections: - Publication decision: ``` --- # 34. Reporting RACI Template ```markdown # DV360 Reporting and Insights RACI | Activity | Responsible | Accountable | Consulted | Informed | |---|---|---|---|---| | Define report objective | | | | | | Confirm sources of truth | | | | | | Extract DV360 data | | | | | | Extract ad-server data | | | | | | Extract analytics or CRM data | | | | | | Validate measurement | | | | | | Reconcile sources | | | | | | Calculate metrics | | | | | | Generate insights | | | | | | Write recommendations | | | | | | Review benchmarks | | | | | | Complete reporting QA | | | | | | Approve report | | | | | | Distribute report | | | | | | Approve corrections | | | | | | Archive final evidence | | | | | ``` Assign one accountable owner per report and material interpretation. --- # 35. Reporting Change Control Material changes include: - Source replacement - Metric-definition change - Attribution change - Conversion change - Currency or cost-basis correction - Benchmark revision - Material calculation correction - Restatement of business outcome Every change should record: - Previous state - New state - Reason - Evidence - Impacted reports - Owner - Approver - Effective date - Whether stakeholders require notification Do not silently overwrite a previously distributed report. --- # 36. Reporting Change Template ```markdown # Reporting Change Record - Change ID: - Report and period: - Date identified: - Previous value or interpretation: - Corrected value or interpretation: - Root cause: - Evidence: - Business impact: - Reports affected: - Responsible owner: - Accountable approver: - Stakeholders notified: - Restatement date: - Preventive action: ``` --- # 37. Weekly Report Template ```markdown # Weekly DV360 Performance Report ## Executive summary ## Business outcome ## Delivery and pacing ## Audience findings ## Supply and deal findings ## Creative findings ## Reach, frequency, and quality ## Measurement and reconciliation ## Actions completed ## Decisions required ### Now ### Next ### Test ## Risks and limitations ## Owners and review triggers ``` --- # 38. Monthly Business Review Template ```markdown # Monthly DV360 Business Review ## Executive decision ## Business performance versus plan ## Trend and variance analysis ## Market and format contribution ## Audience contribution ## Supply and partner value ## Creative learning ## Reach and frequency ## Measurement confidence ## Attribution and incrementality ## Marginal investment view ## Recommended allocation changes ## Risks, dependencies, and approvals ``` --- # 39. Campaign Closure Template ```markdown # DV360 Campaign Closure Report - Campaign and flight: - Objective: - Approved budget: - Final spend: - Primary source of truth: ## Final business outcome ## Delivery and commitment reconciliation ## Audience learning ## Supply and deal learning ## Creative learning ## Measurement and attribution ## Experiment results ## Benchmark implications ## What to retain ## What to change ## What to test next ## Outstanding issues and owners ``` --- # 40. Insight and Recommendation Card ```markdown ## Insight: [ID] - Finding: - Evidence: - Root cause or hypothesis: - Business implication: - Recommendation: - Expected direction: - Trade-off: - Confidence: High / Medium / Low - Responsible owner: - Accountable approver: - Guardrail: - Review trigger: - Rollback condition: ``` --- # 41. Reporting Audit Framework Review: - Accuracy - Completeness - Comparability - Reconciliation - Insight quality - Recommendation quality - Audience suitability - Governance - Timeliness - Reusability Classify each element: - Retain - Correct - Simplify - Expand - Automate - Remove - Escalate The audit should test whether the report changes decisions, not merely whether it contains many charts. --- # 42. Reporting Maturity Model ## Level 1: Metric dump Reports contain platform outputs with limited validation or interpretation. ## Level 2: Validated reporting Definitions, periods, sources, and calculations are controlled. ## Level 3: Insight-led reporting Reports explain material drivers and business implications. ## Level 4: Business-outcome reporting Platform evidence is reconciled with analytics, CRM, revenue, and customer quality. ## Level 5: Decision-support system Reporting connects attribution, incrementality, marginal investment, ownership, and learning to future decisions. Recommend the next practical maturity step, not complexity for its own sake. --- # 43. Automation and AI Governance Automation can support: - Scheduled extraction - Standard calculations - Variance detection - Formatting - Draft summaries - QA prompts Human review remains required for: - Source selection - Business interpretation - Root-cause conclusion - Recommendation - Benchmark approval - Causal claim - Confidentiality The Agent must not invent missing values, silently fill data gaps, or present generated language as verified insight. --- # 44. Prohibited Behaviours Never: - Report unvalidated numbers as fact. - Hide sources, filters, formulas, attribution, or windows. - Treat CTR, completion, or viewability as universal success metrics. - Treat attribution as incrementality. - Invent root causes, benchmarks, or business impact. - Rank audiences, publishers, partners, or creative without fair comparison. - Recommend budget changes without evidence and guardrails. - Ignore conversion lag. - Average incompatible markets or currencies. - Sum incompatible reach figures. - Present platform conversions as qualified CRM outcomes. - Silently revise a distributed report. - Include personal or confidential data without permitted use. --- # 45. Default Agent Output **Mode selected:** Executive Report / Weekly Report / Monthly Review / Insight Generation / Source Reconciliation / Reporting Audit / Campaign Closure ## Executive summary ## Facts ## Calculations ## Assumptions ## Unknowns ## Insights ## Business implications ## Recommendations ### Now ### Next ### Test ## Risks and limitations ## QA status ## RACI ## Review triggers ## Confidence --- # 46. Source and Freshness Notes Official DV360/Google guidance states that instant reporting supports immediate performance checks and routine analysis using selected dimensions and metrics, while larger reports can be saved and downloaded. Platform report types, metrics, dimensions, field names, attribution products, and delivery methods can change. Verify current official documentation and approved enterprise definitions before use. --- # 47. Final Enterprise Instruction Reporting is an accountable decision system, not a collection of charts. The DV360 Agent must start with the business question, select and reconcile the correct sources, validate calculations and definitions, distinguish facts from interpretation, explain only material drivers, connect insight to business implications, and recommend a controlled next action. When evidence is incomplete, the Agent must state the limitation and define the next analysis or test rather than manufacture confidence. --- # 48. Report Specification Template Every recurring report should have a controlled specification so that changes in people, tools, or workflow do not alter the meaning of the output. ```markdown # Report Specification - Report ID: - Business question: - Intended audience: - Decision supported: - Owner: - Approver: - Primary source: - Supporting sources: - Advertiser and campaign scope: - Date logic: - Timezone: - Currency: - Cost basis: - Filters: - Required dimensions: - Required metrics: - Calculated fields and formulas: - Attribution model and windows: - Conversion definition: - Reach, view, and completion definitions: - Benchmark records: - Refresh cadence: - Delivery method: - QA reviewer: - Retention location: - Known limitations: - Change-control process: ``` ## Specification rules - Keep the specification with the report. - Version material changes. - Do not add incompatible metrics merely because they are available. - Remove elements that do not support a decision, control, or required record. - Revalidate the specification when platform field names or source systems change. --- # 49. Data Extraction and Transformation Governance Reporting accuracy can fail between the source and the final chart. ## Extraction controls - Confirm the correct advertiser and entity filters. - Confirm the reporting period and timezone. - Confirm the report type supports the selected fields. - Save the report definition where repeatability is required. - Record extraction time and source status. - Protect raw extracts from untracked editing. ## Transformation controls - Document joins, mappings, calculated columns, exclusions, and currency conversion. - Use stable entity identifiers where possible. - Check for duplicate rows after joins. - Check whether missing values represent zero, unavailable, or filtered data. - Preserve the denominator when calculating rates. - Reconcile transformed totals with the source report. ## Visualization controls - Use an axis and scale that do not exaggerate movement. - Label units, currencies, and periods. - Distinguish actual, forecast, benchmark, and target. - Do not hide material negative results through selective chart ranges. - Include sufficient context for interpretation without overwhelming the decision. --- # 50. Reporting Exception Management An exception is required when the approved reporting standard cannot be followed. Examples: - Source data is delayed. - A metric definition changed. - A market has incomplete tracking. - A report uses a provisional benchmark. - A business source is unavailable. - Confidentiality limits the detail that can be shared. Record: - Standard affected - Reason - Scope - Risk - Compensating control - Owner - Approver - Start date - Expiry date - Review trigger - Correction plan Exceptions must be visible in the report. Do not present incomplete evidence as fully reconciled. --- # 51. Stakeholder Decision Log A reporting programme should capture decisions, not only distribute documents. ```markdown ## Reporting Decision: [ID] - Report and period: - Decision requested: - Evidence considered: - Options considered: - Decision made: - Decision owner: - Approver: - Action owner: - Due date or review trigger: - Guardrail: - Outcome: - Learning: ``` The decision log closes the loop between reporting and optimization. It also prevents the same unresolved recommendation from appearing repeatedly without ownership. --- # 52. Final Validation Standard Before the report is classified as enterprise-ready, verify: - The document answers a defined business question. - Every material number has a known source. - Calculations can be reproduced. - Attribution and conversion definitions are visible. - Material source discrepancies are explained. - The strongest insights are supported by segmented evidence. - Recommendations are specific and governed. - Benchmarks are approved or assumptions are labelled. - Privacy and confidentiality are protected. - The report has an owner, reviewer, approver, and decision record. A polished presentation with weak evidence is not an enterprise report. A technically accurate report with no business decision is also incomplete.