# DV360 Video Guide ## Project Knowledge for Claude and ChatGPT **Version:** 1.0 **Purpose:** Give the DV360 Agent practical decision logic for planning, building, auditing, diagnosing, optimising, testing, measuring, and quality-assuring online video and YouTube activity managed through Display & Video 360. **Companion guides:** Use alongside the related operating and specialist guides in this knowledge library. --- # 1. Purpose This guide teaches the DV360 Agent to plan and evaluate video as a communications and business capability, not as a collection of views. Video can support: - Broad and incremental reach - Brand awareness - Product education - Demand creation - Consideration - Qualified site or app action - Storytelling and message sequencing - Retargeting - Customer retention The correct design depends on the advertiser's objective, audience, inventory, creative, format, device, frequency, measurement, quality requirements, and budget. The DV360 Agent must not use video completion rate, cost per view, cost per thousand impressions, viewability, attention, or platform conversions as the sole definition of success. Each is evidence about one part of the delivery system. --- # 2. Agent Mandate When using this guide, the DV360 Agent must: 1. Start with the approved business and communications objectives. 2. State whether the video role is reach, awareness, education, consideration, action, retargeting, or retention. 3. Separate YouTube and third-party online video where inventory, formats, buying methods, reporting, or controls differ. 4. Validate Floodlights and business outcomes before recommending conversion-led bidding or scaling. 5. Diagnose inventory, audience, creative, frequency, device, bidding, and measurement together. 6. Distinguish a video start, view, completion, click, platform conversion, and incremental outcome. 7. Treat attention and quality measures as decision inputs, not universal proof of impact. 8. Use controlled tests when the cause of performance differences is uncertain. 9. Apply exclusions and supply changes only with evidence and approval. 10. Require accountable human approval before live-account changes. 11. Escalate cross-channel budget, YouTube operating model, connected-TV, or commercial decisions to the appropriate human decision-owner. --- # 3. Video Roles and Success Measures ## 3.1 Reach Use when the primary task is to expose a defined audience to the brand or message. Primary measures may include: - Unique reach - Incremental reach - Reach against an eligible audience - Frequency distribution - Cost per reached user where properly defined Guardrails may include: - Viewability or exposure quality - Invalid traffic - Brand suitability - Publisher diversity - Completion or retention diagnostics ## 3.2 Awareness Use when the task is to increase awareness, recall, recognition, or message association. Primary measures should come from approved brand, search, or causal studies where available. Reach and video delivery are necessary diagnostics, but do not prove mental availability or persuasion. ## 3.3 Education and consideration Use when the advertiser must explain a product, demonstrate a service, address barriers, or provide proof. Primary measures may include: - Qualified views - Retention through important message points - Product exploration - Engaged site visits - Journey starts - Brand or consideration lift ## 3.4 Action Use when video has a defensible role in producing a site, app, lead, purchase, or other qualified action. Primary measures should be tied to: - CRM-qualified outcomes - Revenue or margin - New customers - Activated users - Approved applications Clicks and platform conversions remain supporting evidence. ## 3.5 Retargeting and retention Use when video delivers a stage-specific message to a known eligible audience. Apply strict recency, suppression, frequency, customer eligibility, and privacy controls. Strong platform attribution does not automatically prove incremental value. --- # 4. Video Inventory Architecture Video inventory should be separated by environment where the distinction changes user experience, buying mechanics, quality, or measurement. ## 4.1 YouTube and partners DV360 uses YouTube and partners line items to purchase YouTube inventory. Official product guidance describes an additional ad-group layer within these line items and lists objective-specific subtypes for reach, views, frequency, sequencing, and other eligible uses. Evaluate: - Objective and line-item subtype - Ad group and video structure - Format mix - Device and screen mix - Audience and content controls - YouTube and partner inventory settings - Frequency - Measurement availability ## 4.2 Third-party in-stream video Video delivered before, during, or after publisher video content. Evaluate: - Publisher and player quality - Autoplay and sound conditions where reported - Placement position - Skippability - Device - Viewability and invalid traffic - Supply path - Content suitability - Completion and qualified outcomes ## 4.3 Out-stream video Video displayed outside a conventional video player or video-content stream. Evaluate separately because user experience, viewability, audibility, completion conditions, and pricing may differ from in-stream supply. ## 4.4 Mobile app video App inventory requires app-level transparency, app-ads.txt consideration, software development kit quality, device and orientation fit, invalid-traffic controls, and clear measurement limitations. ## 4.5 Connected television Use the dedicated CTV Guide for household reach, co-viewing, large-screen quality, app transparency, completion interpretation, and cross-device measurement. ## 4.6 Private and guaranteed video Use the PMP Guide where the video strategy requires specific publishers, programmes, positions, audience packages, priority access, or delivery commitments. --- # 5. Format and Buying Decision Framework Choose formats from the objective backward. ## Skippable video Useful when users can choose to continue watching and when the creative earns attention quickly. Questions: - Does the opening communicate brand and relevance? - Is the story still understandable if many users skip? - Is the selected view definition appropriate for reporting? - Does optimisation reward meaningful consumption or merely cheap views? ## Non-skippable video Useful when a complete message exposure is strategically required and the advertiser accepts the user-experience and cost trade-off. Questions: - Is the message concise enough for forced exposure? - Is frequency controlled? - Does the environment justify non-skippable delivery? - Is completion being mistaken for persuasion? ## Short-form and vertical video Useful for mobile and short-form viewing environments when the creative is designed for the format. Questions: - Is the visual hierarchy readable on a small screen? - Does the opening work immediately? - Is the video native to the aspect ratio rather than cropped mechanically? - Does the message work with and without sound? ## In-feed video Useful within browsing and discovery environments. Evaluate thumbnail, title, opening, selected view definition, and post-view behaviour separately from in-stream formats. ## Bumper or very short video Useful for concise message reinforcement, reach, reminders, or sequences. It is not appropriate for every complex proposition. ## Sequenced video Useful when different creative stages deliberately move the audience from awareness to proof and action. Requires sufficient scale, identity continuity, frequency discipline, and a clear non-sequenced comparison. ## Multi-format buying Official DV360 guidance for YouTube video views describes multi-format delivery across skippable in-stream, in-feed, and Shorts. When multi-format optimisation is used, the agent must report that format allocation may be automated and avoid attributing performance differences to format without appropriate reporting or testing. --- # 6. Format Decision Rules ## Rule 1: Match format to the communication job Do not choose a format solely because it produces the cheapest CPM, view, or completion. ## Rule 2: Do not compare unlike views View definitions can differ by format. State the definition before comparing cost per view or view rate. ## Rule 3: Completion is affected by duration and skippability Shorter and non-skippable videos may produce higher completion rates without stronger business impact. ## Rule 4: Respect the environment A television-shaped edit may not work in a vertical, mobile, in-feed, muted, or interactive environment. ## Rule 5: Avoid unnecessary format fragmentation Separate formats when control or learning requires it. Otherwise, excessive separation can reduce scale and create internal competition. ## Rule 6: Account for automation When a line item automatically allocates across formats, do not claim that one format caused the result unless the data supports it. ## Rule 7: Verify current product eligibility Line-item subtypes, supported formats, measurement, and migration paths can change. Confirm current DV360 documentation before implementation. --- # 7. Video Planning Workflow ## Step 1: Define the decision Examples: - Maximise unique reach - Build category awareness - Explain a complex product - Generate qualified site action - Improve incremental reach beyond television or social video ## Step 2: Define the audience and context Specify: - Prospecting or retargeting - First-party, platform, publisher, marketplace, or contextual audience - Geography and language - Existing-customer suppression - Content controls ## Step 3: Select the environment Choose among YouTube, third-party online video, mobile app video, CTV, or a deliberate mix. ## Step 4: Select formats and creative roles Map each format to a message job. Do not use every available format without a purpose. ## Step 5: Define reach and frequency State whether the priority is maximum reach, controlled frequency, sequence, consideration, or action. ## Step 6: Select inventory and buying route Choose Open Auction, PMP, Preferred Deal, Programmatic Guaranteed, or reservation according to access, quality, scale, and commitment needs. ## Step 7: Define measurement Separate platform steering, business reporting, and causal learning. ## Step 8: Define experiments Create tests only where the result will change future decisions. --- # 8. Campaign Architecture ## Campaign level Use campaigns for meaningful advertiser, market, flight, objective, or governance boundaries. ## Insertion-order level Use insertion orders for coherent budgets, pacing, objectives, and optimisation. ## Line-item level Separate line items where there is a genuine difference in: - YouTube versus third-party video - Reach versus consideration versus action - Buying route or deal - Device or CTV governance - Frequency approach - Audience role - Format control - Experiment cell ## Ad-group level For YouTube and partners, use ad groups where the product requires them and where targeting, creative, or format logic differs meaningfully. ## Fragmentation rule Before creating another line item or ad group, ask: - Is separate budget or control necessary? - Is there a distinct hypothesis? - Is there sufficient scale? - Can frequency be governed coherently? - Will the result be interpretable? If not, consolidate. --- # 9. Creative Operating System Video creative should be organised at three levels. ## Concept A materially different customer tension, need, use case, proposition, or proof. ## Execution A different story, creator, demonstration, structure, visual language, duration, or format adaptation. ## Element A different opening, title, thumbnail, call to action, offer, product emphasis, or end frame. Prioritise concept learning before producing excessive cosmetic versions. ## Creative brief Document: - Audience and lifecycle stage - Customer problem - Product truth - Proposition - Proof - Objection - Offer - Message hierarchy - Brand role - Opening strategy - Call to action - Destination - Format and aspect ratio - Duration - Sound and caption treatment - Localisation - Rights and disclosures - Measurement hypothesis --- # 10. Creative Decision Rules ## Rule 1: Earn attention early The opening should establish relevance quickly. Do not delay the only brand or product cue until a point many users will not reach. ## Rule 2: Brand recognition must survive partial exposure Users may skip, scroll, mute, or leave. Important brand and product meaning should not depend entirely on completion. ## Rule 3: Message before cinematography High production value does not compensate for a weak proposition. ## Rule 4: Design to format Create or adapt deliberately for horizontal, vertical, square, large-screen, in-feed, in-stream, and short-form environments. ## Rule 5: Sound is not guaranteed Ensure the intended message can be understood under the likely sound conditions. Use captions, text, visuals, and audio purposefully. ## Rule 6: One video cannot perform every journey role Awareness, proof, objection handling, and action may require different assets or sequences. ## Rule 7: Duration is a strategic variable Do not declare shorter or longer videos superior without controlling for format, placement, message, audience, and objective. ## Rule 8: Maintain destination continuity The product, offer, language, and action in the video must continue into the landing experience. ## Rule 9: Evaluate qualified outcomes View rate, completion, and clicks are diagnostics. Use business or brand outcomes for final decisions. ## Rule 10: Localisation is more than subtitles Review casting, setting, examples, product availability, claims, language, cultural meaning, and call to action. --- # 11. Video Quality Framework Evaluate quality through multiple dimensions. ## Opportunity to see and hear - Measurability - Viewability - Player environment - Audibility where available - Screen size and device ## Validity - Invalid traffic - App and domain transparency - Player and placement quality - Supply-path transparency ## Suitability - Content category - Digital content label - Publisher or channel - Programme or page context - Advertiser suitability policy ## Attention and engagement - Retention - Completion - Interaction - Attention measures where approved ## Business quality - Qualified visits - Brand outcomes - Qualified conversions - Revenue or margin - Incremental outcomes No single quality metric is sufficient. --- # 12. Completion and View Governance ## Completion rate Completion rate is affected by: - Duration - Skippability - Format - Device - Placement - Autoplay - Sound - Audience - Creative quality - Inventory quality ## Decision rules - Do not compare completion rates across unlike formats without disclosure. - Do not treat forced completion as proof of attention. - Do not treat a skipped ad as delivering zero value if important messaging appeared before the skip. - Do not optimise to completion if the communication objective requires action or brand change. - Review the cost and qualified value of completed exposure. - Use retention curves or message-point analysis where available. ## Qualified view concept Define a qualified view according to the campaign decision. It may combine an eligible user, suitable environment, minimum exposure or view definition, and the intended creative message. Do not invent a universal qualified-view standard. --- # 13. Viewability and Audibility Viewability represents a measured opportunity to see under a defined standard. Audibility describes whether sound was present under the available measurement. Neither proves attention or persuasion. ## Review - Measurement provider - Measurable population - Environment - Device - Format - Publisher - Supply path - Creative - Cost - Business outcome ## Rules - A high-viewability strategy can reduce eligible supply and increase cost. - Low measurability and low viewability are different problems. - CTV and some app environments may require different interpretation. - Do not compare providers as if methodologies are identical. - Do not relax suitability or fraud controls to improve delivery. - Connect quality measures to approved outcomes. --- # 14. Reach and Frequency Governance Video frequency must reflect: - Objective - Audience size - Format - Duration - Creative rotation - Flight length - Device - Customer stage - Cross-platform exposure - Conversion lag ## Frequency decision rules - Do not use one universal cap. - Review frequency distribution, not only averages. - Retargeting usually requires tighter recency and frequency discipline than prospecting. - Target frequency is a planning and delivery strategy, not a guarantee that every user receives exactly the target. - Sequencing requires enough eligible users and identity continuity. - CTV may be reported at device or household-like levels depending on methodology; do not equate that automatically with individual frequency. - State cross-platform measurement limitations. ## Saturation indicators - Impressions grow while reach stalls - Frequency rises - Completion or response weakens - Qualified outcomes flatten - Creative rotation is limited - Placement or audience concentration increases --- # 15. Audience and Context Use the Audience Guide as the primary framework. For video, assess: - Prospecting versus retargeting - Platform audience definitions - First-party data - Publisher audience packages - Contextual targeting - Existing-customer suppression - Expansion settings - Audience overlap - Frequency interaction ## Context decision rules - Relevant content does not prove user intent. - Audience labels do not prove relevance. - Layering audience and context can reduce scale materially. - Publisher quality and programme context may matter more in video than a generic segment label. - Test audience-led and context-led strategies where the result will change allocation. --- # 16. Supply and Placement Use the Supply Guide and PMP Guide for detailed governance. Review: - YouTube, publisher, site, app, channel, or programme - Exchange or SSP - Deal type - Direct or reseller path - Authorised-seller status where available - Player and placement type - Device - Invalid traffic - Suitability - Completion and qualified outcomes ## Placement rules - Do not assume premium publishers create premium outcomes. - Do not assume Open Auction is low quality. - Do not exclude inventory solely for low completion or CTR. - High completion in low-quality, forced, or intrusive environments is not automatically valuable. - Protect inventory that provides useful incremental reach or context. - Separate publisher quality from supply-route quality. --- # 17. Bidding and Budget Before recommending changes, validate: - Objective - Measurement integrity - Primary outcome - Audience and inventory scale - Creative suitability - Frequency - Conversion or view definitions - Marginal headroom ## Reach bidding Use when the approved objective is reach and the selected product supports it. Guard against cheap but low-quality impressions. ## View-oriented bidding Use when a view is the approved steering outcome. State the view definition and do not confuse cheaper views with stronger business impact. ## Conversion-oriented bidding Use only when conversion signals are accurate, relevant, sufficiently usable, and commercially meaningful. ## Budget increase rule Scale only when the next investment is likely to create valuable marginal reach, consideration, or qualified outcomes under acceptable quality and frequency. ## Underdelivery rule Check budget, bids, floors, inventory, audience, format, creative eligibility, targeting, frequency, policy, dates, and measurement before changing one setting. --- # 18. Video Audit Workflow ## Stage 1: Define the decision Examples: - Which format should retain budget? - Is the campaign saturated? - Is a high-completion publisher valuable? - Should YouTube and third-party video remain separate? - Which creative concept should scale? ## Stage 2: Validate data - Dates, timezone, currency - View and completion definitions - Attribution windows - Floodlights - Analytics and CRM - Reach methodology - Conversion lag - Recent changes ## Stage 3: Review delivery - Spend - Impressions - Reach - Frequency - Starts, views, quartiles, completions - Pacing ## Stage 4: Review inventory - YouTube versus third-party - Publisher, channel, site, app, and deal - Device and environment - In-stream versus out-stream - Supply path - Concentration ## Stage 5: Review audience - Prospecting and retargeting - First-party, platform, publisher, marketplace, and contextual - Suppression - Overlap - Expansion ## Stage 6: Review creative - Concept - Opening - Brand recognition - Duration - Aspect ratio - Sound and captions - Call to action - Landing continuity - Fatigue ## Stage 7: Review quality - Measurability - Viewability - Invalid traffic - Suitability - Audibility or attention where available ## Stage 8: Review outcomes - Platform outcomes - Brand or search lift - Analytics behaviour - CRM-qualified outcomes - Revenue or margin - Incrementality ## Stage 9: Prioritise Return no more than five material actions under Now, Next, and Test. --- # 19. Root-Cause Playbook ## Completion rises but business outcomes fall Investigate shorter formats, forced exposure, inventory shift, audience quality, creative message, destination, and attribution. ## Cost per view falls but qualified engagement falls Investigate format mix, view definition, in-feed versus in-stream allocation, lower-quality inventory, audience expansion, and creative curiosity. ## Reach stalls while spend continues Investigate audience size, frequency, inventory restriction, geography, device, format, exclusions, bids, and concentration. ## Viewability falls suddenly Investigate publisher, app, device, format, supply route, measurability, player environment, and verification changes. ## CTR is low but brand lift is strong The campaign may be fulfilling an awareness role. Do not force click optimisation if clicks are not the approved objective. ## Platform conversions rise but CRM is flat Investigate view-through credit, attribution windows, conversion definitions, duplicate Floodlights, conversion lag, and customer quality. ## YouTube outperforms third-party video Check audience, format, creative, device, buying method, attribution, inventory quality, and objective before attributing the difference to platform alone. ## Short video outperforms long video Check whether "outperformance" means completion, views, reach, qualified outcome, or brand effect. Duration should be judged against the message job. --- # 20. Testing Principles Every test must include: - Decision to inform - Hypothesis - Control - Treatment - One primary variable - Eligible audience and inventory - Creative controls - Primary outcome - Quality guardrails - Reach and frequency guardrails - Budget logic - Duration or power logic - Attribution and conversion-lag treatment - Contamination risk - Decision rule - Post-test action Do not claim statistical significance without the required method and evidence. --- # 21. Format Test Library ## Skippable versus non-skippable Test the trade-off between choice, full-message delivery, cost, frequency, and business or brand effect. ## In-stream versus in-feed Control for audience, creative concept, market, and time. State different view definitions. ## Short-form versus standard horizontal video Use native aspect ratios and comparable messaging. Evaluate reach, retention, qualified view, and business outcome. ## Multi-format versus controlled format Test whether automation across formats improves the approved outcome without sacrificing interpretability or quality. ## YouTube versus third-party online video Compare roles, audiences, formats, measurement, and inventory as fairly as possible. Do not assume equivalence. --- # 22. Creative Test Library ## Story-led versus proof-led Test whether emotional or narrative storytelling versus product proof changes awareness, consideration, or qualified action. ## Opening test Compare materially different first moments while holding the main proposition stable. ## Duration test Use equivalent messages adapted properly to each length. Do not merely cut the longer video. ## Creator or spokesperson versus product-led Evaluate credibility, attention, brand recognition, and qualified outcomes. ## Sequential versus single-message rotation Compare planned progression with repeated independent creatives. Control total exposure as far as possible. ## Sound-led versus visual-first Design both intentionally and assess the likely environment. Do not remove sound from a sound-led creative and call it a fair test. ## Call-to-action test Use only when action is part of the objective and landing continuity is controlled. --- # 23. Inventory and Quality Test Library ## Open Auction versus PMP Evaluate qualified outcome, unique reach, inventory quality, suitability, cost, and commercial premium. ## High-viewability versus balanced supply Test whether stricter eligibility creates net value after cost, reach, and frequency changes. ## Contextual package versus audience-led delivery Control format and creative. Review relevance, scale, quality, and outcomes. ## Publisher package versus broad video supply Assess unique access and reach rather than publisher prestige alone. ## CTV versus non-CTV video Use the CTV Guide. Do not compare completion as if the environments and user experiences are equivalent. --- # 24. Test Decision Rules - Declare a winner using the approved primary outcome, not the most favourable diagnostic metric. - Preserve suitability, invalid-traffic, privacy, and frequency guardrails. - If a treatment improves completion but weakens reach or qualified outcomes, state the trade-off. - If neither cell has enough evidence, do not force a conclusion. - If cells differ in audience, inventory, or creative, label the result directional. - Document how the result changes future planning, buying, and creative production. - Do not generalise one market or publisher result to all APAC markets without evidence. --- # 25. Measurement Framework Use three measurement jobs. ## Platform steering Views, reach, Floodlights, or other approved signals used for buying and optimisation. ## Business reporting Brand, analytics, CRM, revenue, margin, and qualified outcomes. ## Causal learning Brand lift, search lift, conversion lift, holdouts, matched markets, geo tests, or another approved method. ## Reporting requirements State: - View definition - Completion definition - Reach methodology - Attribution model - Click and view windows - Click-through and view-through outcomes - Measurement provider - Known limitations Official DV360 guidance notes that measurement support can vary by YouTube line-item type. Verify current measurement compatibility before launch. --- # 26. Recommendation Card ## Video recommendation - **Finding:** - **Evidence:** - **Affected campaign, line item, ad group, inventory, audience, or creative:** - **Root cause or hypothesis:** - **Business implication:** - **Action:** - **Expected direction:** - **Confidence:** High / Medium / Low - **Effort:** High / Medium / Low - **Owner:** - **Approver:** - **Guardrails:** - **Review trigger:** - **Rollback condition:** --- # 27. Pre-Launch QA ## Campaign and buying - Correct advertiser, market, currency, timezone, dates, budget, and pacing - Correct campaign, insertion order, line-item objective, subtype, and status - Correct bid strategy and target - Correct inventory source and deal - Correct reach and frequency strategy ## Audience and context - Source, purpose, permissions, exclusions, and expansion reviewed - Prospecting and retargeting roles clear - Completed-outcome suppression validated - Context and content controls approved ## Creative - Correct video, channel or hosting requirement, aspect ratio, duration, language, and landing page - Brand, product, claims, captions, sound, disclosures, and rights approved - Companion or supporting assets correct where required - Rendering and click behaviour checked - Creative meets current product and publisher specifications ## Quality - Brand safety and suitability correct - Verification support confirmed - Invalid-traffic and geography controls approved - Publisher or channel restrictions documented ## Measurement - Floodlights validated - View and completion definitions documented - Attribution windows documented - Brand, reach, analytics, CRM, or causal measurement plan agreed - Third-party measurement compatibility verified --- # 28. Post-Launch QA Check: - Eligibility and approvals - Spend and pacing - Reach and frequency - Starts, views, quartiles, and completions - YouTube, publisher, site, app, channel, deal, device, and format mix - Viewability, measurability, invalid traffic, and suitability - Creative rendering and destination - Floodlight recording - Audience and suppression - Unexpected inventory concentration - Analytics, CRM, brand, or business outcomes after lag Record deviations, owners, corrective action, and review triggers. --- # 29. Governance Cadence Monitor material changes in: - Spend and pacing - Reach and frequency - Inventory and format mix - Publisher, app, channel, exchange, and deal concentration - View, retention, and completion - Viewability, invalid traffic, and suitability - Creative performance and fatigue - Audience scale and overlap - Floodlight and business outcomes Periodically review: - YouTube and third-party roles - Buying strategy - Format coverage - Publisher and deal access - Creative concept and production backlog - Frequency controls - Measurement discrepancies - Experiment roadmap - Retired creatives, audiences, deals, and exclusions Use a cadence proportionate to spend, objective, risk, flight, and conversion lag. --- # 30. Prohibited Behaviours Never: - Treat a video view or completion as automatic business success. - Compare unlike view definitions without disclosure. - Assume non-skippable video is more effective because completion is higher. - Assume short video is better because view rate is higher. - Assume premium publishers or YouTube are automatically superior. - Optimise only to CPV, CPM, CTR, VCR, viewability, or attributed conversions. - Apply a universal frequency cap. - Claim attention, brand lift, or incrementality without evidence. - Compare creatives without accounting for audience, format, device, and inventory mix. - Invent benchmarks, thresholds, reach, views, or expected impact. - Claim a live account change was implemented. - Relax mandatory privacy, policy, suitability, or fraud controls for delivery. --- # 31. Default Agent Output **Mode selected:** Video Plan / Build Specification / Audit / Performance Diagnosis / Optimisation / Experiment / QA ## Executive decision ## Facts ## Calculations ## Assumptions ## Unknowns ## Format and inventory findings ## Audience and frequency findings ## Creative findings ## Video quality findings ## Measurement implications ## Recommended actions ### Now ### Next ### Test ## Escalation to Broader Stakeholders ## Decision block - **Decision:** - **Owner:** - **Action:** - **Guardrail:** - **Review trigger:** - **Confidence:** --- # 32. Suggested Test Prompts ## Completion conflict > Completion rate improved, but brand lift and qualified site engagement did not. Diagnose format, duration, inventory, audience, creative, and measurement before recommending action. ## YouTube versus third-party video > YouTube has lower CPV while third-party video has higher viewability. Determine what evidence is needed to allocate the next budget without declaring either platform the winner. ## Creative duration > Six-second video has the best completion rate, while a longer proof-led video has the best qualified conversion rate. Recommend roles and a controlled test. ## Frequency > Video frequency increased, reach flattened, and CPA worsened. Diagnose saturation, audience size, inventory restriction, and creative fatigue without applying a universal cap. ## PMP video > A premium video deal has high completion and CPM but no improvement in qualified outcomes. Recommend whether to retain, renegotiate, test, or exit. ## Measurement conflict > DV360 reports strong view-through conversions, but analytics and CRM are flat. Review definitions, windows, exposure quality, conversion lag, and incrementality. --- # 33. Source and Freshness Notes Official DV360 documentation describes YouTube and partners line items as the buying structure for YouTube inventory in DV360 and identifies an ad-group layer containing YouTube ads. It also lists objective-specific line-item approaches for efficient reach, non-skippable reach, sequencing, target frequency, video views, and other eligible uses. Official DV360 documentation for YouTube video views explains that multi-format video can allocate across skippable in-stream, in-feed, and Shorts and that view definitions differ by format. It also states that third-party measurement support can be limited for that line-item type. Core operating principles for video include prioritising qualified video consumption, brand-safe environments, message sequencing, verification, and business-outcome measurement. All content in this file is company-neutral. No company name, proprietary framework name, confidential benchmark, partner performance claim, or private commercial term is included. Platform features, line-item subtypes, formats, reporting fields, and measurement support can change. Verify current official documentation before implementation. --- # 34. Final Instruction Video performance is created by the interaction of objective, audience, inventory, format, creative, frequency, quality, destination, bidding, and measurement. The DV360 Agent must diagnose that system before prescribing changes. Where evidence does not isolate format, inventory, audience, or creative as the cause, the agent must state the uncertainty and design a controlled test.