Agency paid-media planning questions
Status: qualitative, unscored planning aid.
Required gates
- Context gate: identify the client, business model, objective, conversion definition, geography, currency, data window, budget authority, and active platforms separately.
- Evidence gate: use only client-approved inputs and current official source IDs; keep assumptions, stale evidence, and cross-client comparisons explicit.
- Privacy gate: isolate every client's credentials, exports, audiences, creatives, reports, run directory, and retention policy.
- Regulatory gate: identify the client's category and jurisdictions, then obtain the required platform, privacy, and specialist review before proposing regulated activity.
- Capability gate: distinguish export analysis, authorized reads, drafts, and disabled writes for each client and platform.
- Approval gate: record the client owner, agency owner, creative approval, budget ceiling, mutation authority, verification plan, and rollback path.
Planning questions
- Which business outcome does the client own, and how is it reconciled downstream?
- Which client facts, claims, audiences, assets, and account permissions are approved?
- Where do tracking, CRM, consent, feed, and offline outcome definitions disagree?
- Which platforms are eligible and measurable for this client's objective and geography?
- Which responsibilities belong to the client, agency, platform, or another vendor?
- How will findings, approvals, incidents, and changes remain client-specific and auditable?
- What would cause the plan to stop for missing evidence, privacy, policy, or authority?
Candidate considerations
- Compare platform roles from current client evidence rather than a standard agency mix.
- Consider a small, reversible learning plan with client-defined guardrails.
- Separate internal quality review from client-facing reporting and account execution.
- Preserve unsuccessful hypotheses and contradictory evidence for the next planning cycle.
Output guardrails
- Return options, tradeoffs, owners, assumptions, source IDs, and verification steps.
- Do not provide universal allocations, unsupported targets, automatic account-action rules, or cross-client performance promises.