# DV360 Connected TV Guide ## Project Knowledge for Claude and ChatGPT **Version:** 1.0 **Purpose:** Give the DV360 Agent evidence-led decision logic for planning, buying, auditing, optimising, measuring, testing, and quality-assuring connected television activity. **Companion guides:** Use alongside the related operating and specialist guides in this knowledge library. --- # 1. Purpose This guide teaches the DV360 Agent to treat connected television as a distinct large-screen environment rather than a simple extension of online video. Connected television can support: - Incremental reach beyond linear television and other digital video - Broad or targeted household exposure - Premium long-form video environments - Brand awareness and consideration - Product education and storytelling - Geographic or audience-led reach - Cross-screen planning and frequency management - Measured business or brand outcomes CTV includes different apps, publishers, devices, buying routes, audience signals, ad experiences, identifiers, and measurement methods. The DV360 Agent must make these differences explicit. High completion or premium delivery does not prove attention, incremental reach, or business impact. --- # 2. Agent Mandate When using this guide, the DV360 Agent must: 1. Start with the business and communications objective. 2. Define CTV's role relative to linear television, YouTube, online video, social video, and other channels. 3. Distinguish device-level, household-level, user-level, and modelled reporting. 4. Validate inventory transparency, app identity, supply path, rights, and suitability. 5. Protect reach and frequency from fragmented buying and duplicated exposure. 6. Assess creative for the large-screen, often non-clickable viewing environment. 7. Validate Floodlights and business measurement before recommending conversion-led optimisation. 8. Separate attributed outcomes from incremental outcomes. 9. Evaluate Open Auction, PMP, Preferred Deal, and Programmatic Guaranteed according to access and commercial value. 10. Use controlled tests where the incremental value of CTV is uncertain. 11. Require accountable human approval before live-account, deal, budget, targeting, or measurement changes. 12. Escalate cross-media allocation and linear-TV decisions to the appropriate human decision-owner. --- # 3. Definitions and Boundaries ## Connected television A television screen connected to internet-delivered content through a smart-TV operating system, streaming device, set-top box, games console, or another supported connected device. ## Over-the-top content Video or television-like content delivered over the internet. OTT describes the delivery method and can appear on television, mobile, tablet, or desktop devices. CTV specifically refers to viewing on a television screen. ## Streaming app An application through which video content is viewed on a connected television or connected device. ## FAST service A free, ad-supported streaming television service. Do not assume every FAST service, channel, or programme has the same content quality, rights, audience, or measurement. ## Co-viewing Multiple people may watch the same television screen. Device or household exposure should not automatically be described as measured individual exposure. ## Incremental reach Reach delivered to people, devices, or households not reached by the comparison media under the stated methodology. Always identify the unit and method. --- # 4. CTV Role in the Media System ## 4.1 Complement to linear television CTV may provide streaming-audience access, incremental reach, flexible activation, and digital reporting. It may also duplicate linear television exposure. Use cross-media reach evidence where available rather than assuming incrementality. ## 4.2 Complement to YouTube YouTube on television screens can provide large-screen reach, but it is not interchangeable with every third-party CTV app. Compare inventory, audience, ad format, content context, buying route, reach, and measurement. ## 4.3 Complement to online video CTV can add large-screen exposure and television-like contexts. Online video may provide clicks, personal-device signals, and different format flexibility. Assign distinct roles rather than forcing a single performance comparison. ## 4.4 Complement to paid social video Social video can create high-frequency personal-screen reach. CTV can add large-screen household exposure. Cross-platform overlap may not be fully observable, so state the limitation. ## 4.5 Performance role CTV may influence search, site visits, app actions, leads, and purchases. It should not automatically be evaluated using last-click response because many CTV experiences do not encourage or support immediate clicking. --- # 5. Objective and KPI Framework ## Reach objective Primary measures: - Unique reach under a stated methodology - Incremental or added reach - Reach by market, audience, app, publisher, or device where supportable - Frequency distribution Guardrails: - CPM - Suitable delivery - Invalid traffic - App transparency - Completion and exposure quality ## Awareness objective Primary measures: - Brand lift - Search lift - Awareness or recall outcomes - Incremental reach where relevant Delivery measures support the analysis but do not prove brand change. ## Consideration objective Primary measures may include: - Brand consideration lift - Qualified site or app visits - Search behaviour - Product exploration - Journey starts ## Business objective Primary measures may include: - Qualified leads - Approved applications - New customers - Purchases - Revenue or margin - Store or offline outcomes Use experiments for causal proof. ## Quality objective Viewability, completion, suitability, audibility, and attention may be guardrails or intermediate measures. Do not make them the business source of truth unless the approved objective is explicitly a quality outcome. --- # 6. Planning Inputs Gather: - Business and communications objective - CTV role in the wider plan - Markets and languages - Flight dates - Budget and flexibility - Linear-TV plan where relevant - Audience and customer exclusions - Required publishers, apps, channels, programmes, or genres - Inventory and deal availability - Creative durations, ratios, languages, and rights - Frequency requirement - Brand-safety and suitability policy - Verification compatibility - Floodlights, analytics, CRM, and causal measurement - Cross-media planning and reporting access - Commercial commitments and approval owners Do not invent market inventory, publisher availability, app scale, audience reach, or measurement support. --- # 7. Inventory Architecture ## 7.1 YouTube on connected televisions Evaluate: - Large-screen device delivery - YouTube line-item objective and format - Audience and content controls - YouTube and partners settings - Reach and frequency - Relationship with other YouTube activity - Measurement compatibility Do not label all YouTube delivery as CTV. Use device-type reporting and the approved campaign structure. ## 7.2 Third-party streaming apps Evaluate: - App identity - Publisher ownership - Content and rights - Device and app-store identifiers - Placement type - Ad experience - Supply route - Suitability - Measurement support ## 7.3 Broadcaster and premium publisher inventory Evaluate whether the inventory provides: - Unique or priority access - Defined programmes or channels - Delivery certainty - Brand-suitable context - Incremental reach - Commercial value Publisher reputation is not enough. Request exact inventory and reporting definitions. ## 7.4 Open Auction CTV Potential advantages: - Scale - Flexible optimisation - Broad app access Potential risks: - Variable app transparency - Duplicate paths - Inconsistent content metadata - Quality and identity limitations ## 7.5 Private marketplace CTV Potential advantages: - Defined publisher or app inventory - Greater control - Audience or contextual packages - Priority access Potential risks: - Higher price - Underdelivery - Duplicated Open Auction access - Limited reporting ## 7.6 Programmatic Guaranteed CTV Use when delivery certainty or reserved access is essential. Assess commitment, targeting, creative, measurement, amendment, cancellation, and underdelivery terms before agreement. --- # 8. Inventory Decision Rules ## Rule 1: App identity is foundational CTV supply should be identifiable at the most useful app, publisher, channel, or programme level available. Unresolved identity reduces confidence. ## Rule 2: Premium is a claim to test Premium inventory must demonstrate a relevant content, audience, reach, quality, access, or operational advantage. ## Rule 3: Open Auction is not automatically low quality Open Auction can provide scale and valuable reach. Govern it through app transparency, authorised sellers, supply-path review, suitability, verification, and outcomes. ## Rule 4: Private is not automatically incremental A deal may provide inventory also available through Open Auction. Establish uniqueness or priority. ## Rule 5: Completion does not validate the environment Many CTV ads complete because the format is full-screen and non-skippable. Review content, app, frequency, validity, and business effect. ## Rule 6: Separate publisher from route The same publisher may be available through several exchanges or deals. Evaluate routes using the Supply Guide. ## Rule 7: Protect useful scale Do not over-restrict apps, publishers, audiences, and quality settings until viable reach disappears. ## Rule 8: Verify rights and market availability Do not assume that content, apps, channels, or programmes are available in every APAC market. --- # 9. App and Supply Transparency Official DV360 guidance states that CTV reporting can use the Device Type dimension and that the App/URL dimension can identify individual apps. It also states that standardised app-store identifiers are used for specified connected-TV app stores. ## Required review - App name - App-store name - App identifier - Publisher or developer - Content service - Exchange or SSP - Deal - Direct or reseller relationship where available - app-ads.txt participation - SupplyChain information where available - Device type - Country ## Decision rules - Do not rely on app name alone when identifiers are available. - Similar app names can represent different stores or versions. - Non-participating or unresolved app inventory requires a documented risk decision. - Authorised status does not prove content or media quality. - Confidential seller nodes reduce transparency and should be assessed according to materiality. - Do not claim an app is fraudulent from missing metadata alone. --- # 10. Audience Framework Use the Audience Guide for full governance. CTV audiences may include: - Broad eligible households or devices - Platform audiences - Advertiser first-party audiences - Publisher first-party audiences - Contextual and content-based groups - Linear-TV exposure audiences where available - Geographic audiences ## Audience decision rules - Audience labels do not prove household composition. - Household signals should not be described as precise individual targeting. - Narrow audiences can quickly create scale and frequency problems. - First-party data requires permission and realistic matchability. - Publisher audiences require definition, provenance, recency, scale, and consent evidence. - Context can be preferable when identity is weak or unnecessary. - Existing-customer suppression may not be perfect in a shared-screen environment; state the limitation. - Do not use sensitive or protected traits discriminatorily. --- # 11. Cross-Device and Household Considerations CTV exposure often occurs on a shared screen, while outcomes may occur on another device. ## Required distinctions - Television device - Household - Signed-in account - Individual user - Modelled cross-device relationship - Directly observed outcome ## Decision rules - Do not call a device impression an individual person reached. - Do not assume every household member saw the ad. - Do not describe modelled cross-device conversion as deterministic unless the methodology explicitly supports that statement. - Household attribution can help analysis but does not prove causality. - State whether reporting is observed, modelled, deduplicated, or estimated. --- # 12. Reach and Frequency Governance CTV requires frequency management across fragmented apps, publishers, deals, devices, and platforms. ## Review levels - Advertiser - Campaign - Insertion order - Line item - Publisher and app - Deal - Audience - Creative - Device or household methodology - Cross-platform exposure where available ## Frequency decision rules - Do not use one universal cap. - Set frequency according to objective, audience size, creative rotation, flight length, and inventory scale. - Review distribution, not only average frequency. - High completion does not justify high repetition. - Guaranteed deals need frequency alignment before commitment. - Overlapping YouTube, third-party CTV, online video, and linear television can create unobserved duplication. - Treat target frequency as a delivery strategy, not an exact promise for every household. - If reach stalls and frequency rises, diagnose scale constraints before adding budget. ## Saturation indicators - Impressions grow while unique reach flattens - Frequency rises in a small app set - Marginal reach becomes expensive - Creative response weakens - Brand or business outcomes plateau - Audience restrictions narrow supply - Multiple deals reach the same viewers or households --- # 13. Creative Framework CTV creative should earn attention on a large screen while respecting a television-like viewing experience. ## Creative brief Document: - Audience and viewing context - Customer problem - Proposition - Proof - Brand role - Opening - Duration - Visual hierarchy - Sound and dialogue - Captions and accessibility - Call to action - Destination or follow-on behaviour - Market and language - Rights and disclosures - Measurement hypothesis ## Creative decision rules ### Rule 1: Design for the large screen Use legible text, clear composition, suitable resolution, and television-safe pacing. ### Rule 2: Use sound purposefully CTV is often sound-on, but this should not be assumed for every environment. Dialogue, music, effects, and voiceover should support the message. ### Rule 3: Do not depend on a click Many CTV experiences are not click-led. Build memorable brand, product, search, QR, app, store, or other approved follow-on cues where appropriate. ### Rule 4: Completion is not comprehension A completed ad can still have weak branding, proposition, or proof. ### Rule 5: Duration must fit the job Short assets can reinforce a simple message. Longer assets can explain or demonstrate. Judge duration against communication and business outcomes. ### Rule 6: Control frequency through creative variety Rotation should provide meaningful concept or message variation, not cosmetic duplication. ### Rule 7: Localise completely Review voice, casting, setting, examples, product availability, regulation, language, subtitles, and cultural meaning. ### Rule 8: Confirm rights Verify talent, music, programme sponsorship, territory, device, and duration rights before activation. --- # 14. QR Codes and Follow-On Actions Use QR codes only when they support the customer task and can be measured responsibly. ## QA questions - Is the code visible long enough? - Is the code large and clear enough for the likely screen distance? - Does it resolve correctly? - Is the destination mobile-friendly? - Are tracking parameters correct? - Is the call to action understandable? - Is there a non-QR alternative? - Are privacy disclosures and consent appropriate? ## Decision rules - A QR scan is a response signal, not necessarily a qualified business outcome. - Do not place sensitive information in the code. - Do not use QR performance as the only evaluation of an awareness campaign. - Distinguish scans, sessions, conversions, and qualified outcomes. --- # 15. Quality and Suitability Review: - App and publisher identity - Content category and programme context - Invalid traffic - Brand safety and suitability - Geographic delivery - Completion and exposure environment - Ad pod and repetition where measurable - Verification availability - Supply-path transparency ## Decision rules - Apply the advertiser's documented suitability policy. - Do not assume app-store presence proves quality. - Do not use completion rate as proof of valid traffic. - Do not compare verification providers as if methodologies are identical. - Where verification is limited, state the limitation and strengthen supply and publisher controls. - High quality does not automatically prove incremental impact. --- # 16. Viewability and Completion CTV viewability can require different interpretation from browser-based video. Official DV360 documentation notes specific treatment of CTV in viewability targeting. The agent must verify current settings and measurement support. ## Completion rules - Full-screen or non-skippable delivery can create very high completion. - Compare completion only across sufficiently similar duration, skippability, environment, and inventory. - Review cost per completed view only as a media-efficiency diagnostic. - Connect completed exposure to reach, brand, search, site, CRM, or causal outcomes. - Do not punish a skippable format solely because completion is lower. ## Quality diagnosis ### High completion, weak outcomes Investigate creative meaning, audience, app quality, frequency, attribution, and whether completion was forced. ### Lower completion, strong brand or business outcomes Investigate whether early messaging, audience relevance, or unique reach created value. ### Completion changes suddenly Check duration, format, app mix, deal mix, publisher, device, creative, and reporting definitions. --- # 17. Deal and Commercial Framework Use the PMP Guide for full deal logic. CTV deals must specify: - Publisher and seller - Apps, channels, programmes, or packages - Device and market - Deal type - Price or floor - Forecast or commitment - Placement and ad experience - Audience or context - Frequency approach - Creative requirements - Measurement - Suitability and verification - Cancellation, amendment, and underdelivery terms ## Commercial decision rules - A higher CPM is justified only by a valuable advantage. - Programmatic Guaranteed requires commitment-risk assessment. - Private Auction and Preferred Deals do not guarantee delivery. - Open Auction may duplicate private-deal inventory. - Publisher forecasts are not guarantees unless contractually defined. - Added value should not obscure total effective cost. - Do not approve a deal without a credible baseline or reason why comparison is impossible. --- # 18. Bidding and Budget Before recommending budget or bidding changes, validate: - Objective - Reach and frequency - Inventory scale - Audience scale - Creative readiness - Deal and floor conditions - Measurement integrity - Qualified outcome - Marginal headroom - Cross-media opportunity cost ## Budget increase rule Scale when the next investment is likely to deliver valuable marginal or incremental reach, brand impact, or qualified outcome under acceptable quality and frequency. ## Bid increase rule Increase bids only when additional access is valuable and the expected price remains commercially defensible. ## Guaranteed budget rule Protect contracted commitments, but do not conceal underdelivery, creative delays, targeting conflicts, or weak value. ## Underdelivery diagnosis Review deal status, dates, budget, pacing, bids, floors, app availability, audience, frequency, creative, policy, geography, device, and seller-side restrictions. --- # 19. Measurement Framework CTV measurement should separate three jobs. ## Platform steering - Impressions - Reach and frequency - Completion - Floodlights - Bidding signals ## Business reporting - Brand outcomes - Search activity - Qualified site or app behaviour - CRM outcomes - Revenue or margin - Offline outcomes ## Causal learning - Brand lift - Search lift - Conversion lift - Geo experiments - Holdouts - Matched markets - Cross-media experiments Attributed cross-device conversions can support analysis but do not automatically prove causal CTV impact. --- # 20. Reporting and Forecasting Official DV360 guidance states that connected-TV reporting can be broken out using Device Type, and app-level inventory can be reviewed using App/URL. It also describes digital media plans with TV and Cross-Media Reach reporting as tools for estimating and analysing reach, frequency, overlap, and incremental reach where supported. ## Required reporting fields - Device Type - App/URL or app identifier - Publisher and deal - Exchange or SSP - Country - Creative and duration - Spend and impressions - Reach and frequency - Starts and completions - Floodlight outcomes - View-through and click-through outcomes - Verification and suitability where available ## Reporting rules - State whether reach is user, device, household, deduplicated, or modelled. - Separate YouTube CTV and third-party CTV where relevant. - Separate guaranteed and non-guaranteed buying. - Do not sum incompatible reach methodologies. - State attribution and conversion windows. - Document missing app, publisher, or household transparency. - Treat forecasts as estimates, not commitments. --- # 21. CTV Audit Workflow ## Stage 1: Define the decision Examples: - Is CTV adding reach beyond linear television? - Which apps or deals should retain budget? - Is frequency excessive? - Is premium inventory justified? - Is CTV influencing qualified outcomes? ## Stage 2: Validate data - Dates and timezone - Currency - Device definition - Reach methodology - Attribution windows - Floodlights - Analytics and CRM - Linear-TV comparison data where relevant - Conversion lag ## Stage 3: Review delivery - Spend - Impressions - Reach - Frequency - Starts and completions - Pacing ## Stage 4: Review inventory - YouTube versus third-party CTV - App, publisher, channel, programme, and deal - Exchange and supply path - Open Auction versus private - Concentration and long tail ## Stage 5: Review audience - Broad, first-party, platform, publisher, contextual, and linear-exposure audiences - Suppression - Overlap - Scale ## Stage 6: Review creative - Concept - Brand recognition - Duration - Sound - Legibility - Call to action - QR or follow-on action - Rights and localisation ## Stage 7: Review quality - App transparency - Invalid traffic - Suitability - Completion - Verification - Supply authorisation ## Stage 8: Review outcomes - Brand and search lift - Site and app behaviour - CRM, revenue, and offline outcomes - Incrementality evidence ## Stage 9: Prioritise Return no more than five material actions under Now, Next, and Test. --- # 22. Root-Cause Playbook ## Reach stalls while spend grows Investigate audience size, app availability, deal concentration, frequency, geography, bids, floors, creative eligibility, and duplicated exposure. ## Completion is high but brand or business outcomes are weak Investigate forced exposure, creative clarity, audience relevance, app quality, frequency, product interest, and measurement. ## CPM rises Investigate seasonal demand, scarce publisher access, deal mix, floors, audience targeting, app restrictions, and device supply. Do not automatically reduce bids. ## One app dominates spend Determine whether concentration is intended, unique, valuable, suitable, and frequency-safe. Review alternative apps and supply routes. ## CTV conversions exceed analytics or CRM Investigate household and cross-device attribution, view-through windows, Floodlight definitions, duplicate credit, conversion lag, and modelled reporting. ## PMP underdelivers Review acceptance, line-item assignment, dates, creative, targeting, floor, bid, app availability, frequency, pacing, and seller restrictions. ## CTV appears worse than online video on CPA Do not assume CTV is failing. Assess role, clickable experience, attribution, incremental reach, brand impact, audience, and qualified outcome. ## QR scans are high but conversions are low Investigate code placement, message, destination, mobile experience, tracking, curiosity, and offer quality. --- # 23. Testing Principles Every CTV test must include: - Decision to inform - Hypothesis - Control - Treatment - One primary variable - Eligible inventory and audience - Creative controls - Primary business, brand, or reach outcome - Quality and frequency guardrails - Budget logic - Duration or power logic - Attribution and conversion-lag treatment - Cross-media contamination risks - Decision rule - Post-test action Do not claim statistical significance without sufficient method and evidence. --- # 24. CTV Test Library ## Incremental reach versus linear television Compare an approved CTV treatment against a suitable baseline using compatible reach methodology. Include overlap and frequency. ## YouTube CTV versus third-party CTV Control audience, creative, market, timing, and objective where possible. Compare unique reach, quality, cost, and outcomes without assuming inventory equivalence. ## Open Auction versus PMP Evaluate inventory transparency, app mix, qualified reach, quality, outcome, and commercial premium. ## Broad versus audience-led CTV Test whether audience precision creates enough value to justify scale and cost changes. ## Contextual versus audience-led CTV Compare content relevance against user or household signals while controlling inventory and creative as far as possible. ## Frequency strategy Test whether a different cap or target-frequency approach improves reach efficiency, brand effect, or qualified outcome. ## Creative duration Use properly adapted assets and evaluate message delivery, reach, completion, brand effect, and business outcomes. ## QR versus non-QR response Use only when the call to action is material. Measure scans, sessions, qualified outcomes, and overall campaign effects. ## Guaranteed versus non-guaranteed access Test or assess whether delivery certainty, inventory, quality, and operational value justify commitment and price. --- # 25. Test Decision Rules - Use the approved primary outcome, not the most favourable media diagnostic. - Completion alone cannot select the winner. - Preserve suitability, invalid-traffic, privacy, and frequency guardrails. - If reach improves but frequency or quality worsens, quantify the trade-off. - If cells use different apps, audiences, devices, or creatives, label results directional. - If sample or power is insufficient, do not force a decision. - Do not generalise one publisher or market result across APAC without evidence. - Document how the result changes future planning, negotiation, supply, creative, or measurement. --- # 26. Recommendation Card ## CTV recommendation - **Finding:** - **Evidence:** - **Affected campaign, line item, app, publisher, deal, audience, or creative:** - **Root cause or hypothesis:** - **Business implication:** - **Reach and frequency implication:** - **Commercial 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 CTV-compatible line item and objective - Correct bid strategy and target - Correct device targeting - Correct inventory source and deal - Correct reach and frequency strategy ## Inventory and supply - Apps, publishers, channels, programmes, and packages documented - App identifiers verified where available - Exchange and supply path reviewed - app-ads.txt and authorised-seller policy reviewed - Open Auction, PMP, Preferred Deal, or Programmatic Guaranteed role confirmed - Deal terms and commitments approved ## Audience - Purpose, source, permission, geography, scale, suppression, and overlap reviewed - Household or individual limitations documented - Contextual controls reviewed ## Creative - Correct video, duration, ratio, resolution, language, sound, captions, and landing or follow-on action - Large-screen legibility checked - QR code tested where used - Rights, claims, disclosures, territory, and flight approved - Publisher requirements confirmed ## Quality and measurement - Suitability and verification approved - Invalid-traffic controls reviewed - Floodlights validated - Reach methodology documented - Attribution windows documented - Brand, search, analytics, CRM, offline, or causal measurement plan agreed - Measurement compatibility verified --- # 28. Post-Launch QA Check: - Eligibility and approvals - Spend and pacing - Device-type delivery - Apps, publishers, exchanges, deals, and supply routes - Reach and frequency - Starts and completion - Creative rendering, sound, and QR or follow-on action - Floodlight recording - Audience delivery and suppression - Invalid traffic and suitability - Unexpected inventory concentration - Guaranteed or deal delivery - Analytics, CRM, brand, search, or business outcomes after lag Record deviations, owners, corrective actions, and review triggers. --- # 29. Governance Cadence Monitor material changes in: - Spend and pacing - Unique and incremental reach - Frequency distribution - App, publisher, exchange, and deal concentration - Completion and creative performance - Audience scale and overlap - Invalid traffic and suitability - Floodlight and business outcomes - Measurement discrepancies Periodically review: - CTV's role relative to linear television and other video - App and publisher eligibility - Supply paths and deals - Frequency controls - Creative rotation and rights - Audience definitions - Cross-media reach and overlap - Experiment roadmap - Expired deals, apps, audiences, creatives, and exclusions Use a cadence proportionate to spend, risk, flight, and measurement lag. --- # 30. Prohibited Behaviours Never: - Treat CTV completion as proof of attention or business impact. - Call device or household reach individual reach without methodological support. - Assume CTV is incremental to linear television, YouTube, online video, or social video. - Assume a premium publisher, app, or deal produces premium outcomes. - Optimise solely to CPM, completion, viewability, attention, or attributed conversions. - Apply a universal frequency cap. - Attribute a household or cross-device conversion causally without evidence. - Ignore app identity, supply path, suitability, or rights. - Compare CTV and clickable video environments without explaining the difference. - Invent inventory scale, reach, overlap, benchmarks, or measurement capability. - Claim a live account change was implemented. - Relax mandatory privacy, policy, fraud, or suitability controls to improve delivery. --- # 31. Default Agent Output **Mode selected:** CTV Plan / Build Specification / Deal Review / Audit / Performance Diagnosis / Optimisation / Experiment / QA ## Executive decision ## Facts ## Calculations ## Assumptions ## Unknowns ## Inventory and supply findings ## Reach and frequency findings ## Audience and household findings ## Creative findings ## Quality and measurement findings ## Recommended actions ### Now ### Next ### Test ## Escalation to Broader Stakeholders ## Decision block - **Decision:** - **Owner:** - **Action:** - **Guardrail:** - **Review trigger:** - **Confidence:** --- # 32. Suggested Test Prompts ## Incremental reach > CTV delivered strong reach, but the advertiser also ran linear television and YouTube. Determine what evidence is needed before claiming incremental reach. ## Premium deal > A premium CTV deal has high completion and CPM but no improvement in brand or qualified outcomes. Recommend whether to retain, renegotiate, test, or exit. ## Frequency > CTV spend increased while reach remained flat and frequency rose. Diagnose audience, app supply, deal concentration, and cross-platform overlap without applying a universal cap. ## App transparency > A material share of CTV delivery has incomplete app or seller information. Explain the risk, investigation, safeguards, and approval path without automatically classifying it as fraud. ## Measurement conflict > DV360 reports strong CTV view-through conversions, but analytics and CRM are flat. Review household attribution, windows, Floodlights, lag, and incrementality. ## Creative > A 30-second CTV asset has high completion but weak brand recall. Diagnose branding, message, audience, context, frequency, and study design. --- # 33. Source and Freshness Notes Official DV360 guidance states that CTV delivery can be broken out using Device Type and that App/URL can identify individual apps. The same guidance describes standardised identifiers for specified connected-TV app stores and describes digital media plans with TV and Cross-Media Reach reporting for supported reach, frequency, overlap, and incremental-reach analysis. App proliferation, inconsistent app data, and non-uniform device identifiers are recognised industry challenges for CTV targeting, spam detection, cross-screen resolution, frequency management, and audience graphs. Those points are used here only as generic risk principles. This guide reflects generic operating principles across CTV, supply, video, and measurement. No company name, proprietary framework name, confidential benchmark, partner performance claim, or private commercial term is included in this file. Platform capabilities, app-store support, line-item types, reporting fields, measurement methods, and inventory availability can change. Verify current official documentation before implementation. --- # 34. Final Instruction CTV performance is created by the interaction of objective, incremental reach, app and publisher inventory, supply path, audience, household and device signals, creative, frequency, quality, bidding, commercial terms, and measurement. The DV360 Agent must diagnose that complete system before prescribing changes. When evidence cannot distinguish household exposure, individual exposure, attribution, and causal impact, the agent must state the limitation and design a defensible test.