| name | ad-monetization-review |
| description | Design or audit one complete in-product advertising system for apps, games, web, and content products: rewarded ads, interstitials, banners/native placements, offerwalls, mediation, payer suppression, reward authority, consent, age/territory modes, frequency and fatigue budgets, IAP/IAA cannibalization, fraud, experiments, and shutdown. Use when deciding ad formats, placements, rewards, provider portfolio, or ad economics; combine with an App or Game Blueprint only when whole-product coherence is also unresolved. |
Ad Monetization Review
Produce an Ad Monetization Contract that maximizes incremental retained
contribution without turning attention, privacy, gameplay, or core utility into
an interruption tax.
Atomic boundary
Own ad-side placement semantics, eligible audiences, reward authority,
frequency/fatigue, provider/mediation portfolio, economics, experiments, and
shutdown. Own the ad domain's required measurement meanings, but let
product-analytics-instrumentation-review own the shared event identity,
schema/versioning, collection, and product-metric contract. Do not own
whole-app/game design, buyer payment ledgers, the broader game economy, one
marketing campaign, or provider SDK implementation details.
Use a draft artifact ID and consume sibling decisions by owner and explicit
contract. Let deterministic delivery tooling seal serialized versions and
digests later; never fabricate them in a design response.
Agent-first invariant
Construct the complete production-shaped system now: provider adapters,
consent/age/territory modes, reward ledger, caps, payer states, mediation,
fraud, observability, experiments, kill switches, and reconciliation. Separate
construction from exposure. A disabled placement or provider performs zero SDK
initialization, request, tracking, asset download, background work, or public
call until its exact authority and exposure gate pass.
Workflow
- Label inputs
given, observed, assumed, hypothesis, or decision.
- Define product promise, audience/age modes, session loop, business model,
IAP/subscription mix, platforms, minimum devices, consent regimes, and ruin
boundaries.
- Read
references/ad-monetization-systems.md. Inventory every potential
placement by user intent, natural break, frustration/safety state, format,
expected latency, reward, cap, and degradation.
- Segment new users, engaged non-payers, payers, subscribers/ad-free owners,
lapsed users, child/unknown-age modes, high-frustration states, and
accessibility/data-saver/low-end tiers. Define explainable eligibility and
suppression reason codes.
- Model reward flow as
eligible -> requested -> loaded -> started -> completed -> verified -> granted
with no-fill, timeout, abandonment, duplicate callback, provider dispute,
rollback, and support correction branches.
- Define global, format, placement, session, lifecycle-event, and user+campaign
caps; cross-provider dedupe; cooldown; quiet/sensitive states; and an
emergency kill switch independent of the client release.
- Design mediation and provider adapters with consent-aware initialization,
data maps, latency/no-fill budgets, revenue/impression reconciliation,
provider leases, fraud controls, brand safety, and zero-cost dormant state.
- Evaluate incremental retained contribution after provider/platform fees,
IAP/subscription cannibalization, reward cost/inflation, churn, support,
fraud, privacy risk, latency, battery, and low-end performance.
- Specify candidate-specific tests, capped canaries, scale/hold/pause/withdraw
predicates, automatic shutdown, and live readback.
Source verification
Retrieve current platform/store advertising, consent, child-directed treatment,
rewarded-ad, offerwall, attribution, privacy, and provider SDK authority for the
exact platform, territory, audience mode, and release. Static reference text is
never a current compliance verdict. Unknown/stale authority disables exposure.
When not to use
- Use
game-economy-review when the primary artifact is currency/reward
inflation, sources/sinks, progression, gacha, or payer fairness.
- Use
marketing-automation-blueprint for external paid acquisition, creative
spend, attribution, and lifecycle channel automation.
- Use
app-design-blueprint or game-design-blueprint when ads are only one
component of a whole-product design.
- Use
payment-platform-readiness for purchases, receipts, entitlements,
settlement, refunds, and finance-close authority.
Guardrails
- Never place interstitials inside input/mastery, onboarding trust, payment,
safety, accessibility, error recovery, or irreversible work.
- State the reward, duration/format expectation, qualification, cap, and failure
behavior before start. Never falsely confirm an unverified reward.
- No targeted/personalized ads in child or unknown-age modes unless exact
current authority permits; default to the safer mode.
- Do not punish consent refusal, create fake close buttons, force clicks, hide
ad identity, or make ads the only practical path through core utility.
- Optimize neither impressions nor eCPM alone. Retention, IAP conversion,
complaints, opt-out, latency, crashes, battery, fairness, support, and
long-term value are mandatory countermetrics.
- An autonomous optimizer cannot change consent, child mode, reward value,
placement class, payer suppression, spend/provider caps, or promotion gates.
Output contract
Return one typed Ad Monetization Contract containing:
- objective, audience/platform/age modes, monetization mix, assumptions, and
ruin boundaries;
- placement × format × state matrix with value, interruption, latency,
eligibility, cap, degradation, and owner;
- reward/verification/reversal state machine and ledger/idempotency contract;
- consent, privacy, age/territory, provider, payer/ad-free, and support matrix;
- mediation/provider adapter, dormant-state, no-fill, timeout, and shutdown
design;
- economics model with IAP/IAA cannibalization and long-term countermetrics;
- ad measurement requirements and the product-analytics handoff for event
schema, plus reconciliation, fraud/brand-safety, and support evidence;
- experiment/canary plan with scale, hold, pause, withdraw, and live-readback
predicates;
- specialist handoffs by draft ID, owner, required input/output, and acceptance
questions, without fabricated proof.
Complete only when another agent can implement every state, prove reward
authority and low-end behavior, explain every impression/suppression/grant, and
shut the system down without a client release.