China Partner and Launch Handoff Checklist

1. 01One responsibility matrix covers the full launch lifecycleCritical gate

   Include product decisions, localization, builds, SDKs, storefronts, community, moderation, customer support, payments, reporting, patches, incidents, and exit.

   Completion evidence

   Every row has exactly one accountable owner plus responsible, consulted, informed, and backup contacts with current details.

2. 02Approval rights have limits and response deadlines

   Define who may approve or reject copy, build changes, promotions, creator activity, moderation, emergency patches, refunds, and public statements.

   Completion evidence

   A missed response has a documented escalation or default outcome; no critical approval depends on an unavailable individual.

3. 03All operational accounts and assets are in a controlled registerCritical gate

   List legal owner, administrator, recovery method, access level, MFA owner, export path, billing party, and handover condition for every account.

   Completion evidence

   Two authorized people can recover each critical account without sharing passwords or using a departing employee's private identity.

4. 04The escalation roster has been tested across time zones

   Send a drill message through primary and fallback paths for studio, partner, platform, localization, build, security, community, and executive contacts.

   Completion evidence

   The drill records delivery, acknowledgement time, unavailable periods, and the next reachable decision-maker for each severity level.

4 checks

1. 01The release candidate can be reproduced from controlled inputsCritical gate

   Record repository revision, dependencies, build configuration, signing party, secrets source, symbols, content bundles, and artifact checksum.

   Completion evidence

   An authorized backup operator produces the same candidate without copying files from a developer's workstation.

2. 02Every partner or platform SDK has a written decision recordCritical gate

   For each SDK, state purpose, version, owner, permissions, data touched, network endpoints, failure behavior, update policy, removal plan, and alternatives.

   Completion evidence

   Security, privacy, engineering, and product owners approve the inventory; no unlisted binary or endpoint appears in the candidate.

3. 03Observed data flows match the approved diagramCritical gate

   Test first launch, account creation, telemetry, crash reporting, support attachments, payments, community links, deletion, and offline behavior.

   Completion evidence

   A network and log review accounts for destinations, fields, purpose, retention, access, deletion, and failure mode without unnecessary collection.

4. 04Patch, content update, configuration rollback, and SDK removal are rehearsed

   Use a harmless change to exercise upload, staged deployment, smoke test, rollback, player messaging, version compatibility, and ownership handoff.

   Completion evidence

   The drill records duration, permissions, artifacts, affected data, verification devices, rollback point, and unresolved dependencies.

4 checks

1. 01A versioned manifest covers every player-facing assetCritical gate

   Include game text, fonts, images with text, trailers, captions, store copy, tags, announcements, creator kits, FAQ, macros, policies, and patch notes.

   Completion evidence

   Each asset has source, Chinese version, owner, review state, target channel, freeze date, and checksum or version identifier.

2. 02Native reviewers accept language in the channel where it appearsCritical gate

   Review store copy on the store, UI in the build, subtitles against video, and support macros in the actual tool; preserve screenshots for defects.

   Completion evidence

   Critical paths pass native-language review at target resolution and no unresolved issue can misstate price, availability, gameplay, or support.

3. 03Post-freeze source changes trigger visible localization impact

   Require changed-string detection, owner approval, affected-language notification, re-import, targeted LQA, and an explicit decision when synchronization is impossible.

   Completion evidence

   A test source edit reaches the correct reviewer and candidate build without relying on chat history or manual memory.

4. 04Store, build, marketing, and support make the same promise

   Cross-check edition, features, languages, price timing, online requirements, release state, regions, events, and known limitations.

   Completion evidence

   The manifest identifies one approved statement for each claim and all public channels match it at freeze.

4 checks

1. 01Every official and monitored channel has coverage rules

   Name owner, hours, response target, moderation scope, prohibited actions, archive or export path, and escalation for each store, social, community, and support channel.

   Completion evidence

   Test posts and tickets reach the scheduled responder; unofficial or unmonitored channels are not presented as support destinations.

2. 02A shared issue taxonomy routes feedback to action

   Tag purchase and access, crash, performance, localization, gameplay confusion, account, payment, moderation, safety, privacy, refund, and content issues consistently.

   Completion evidence

   Studio and partner independently classify the same sample tickets to the same owner and severity with acceptable agreement.

3. 03Response guidance is accurate, bounded, and versioned

   Prepare native-language macros for known issues and common questions, with build applicability, prohibited promises, update date, escalation, and personalization guidance.

   Completion evidence

   Responders can answer the rehearsal queue without inventing dates, requesting excessive personal data, or contradicting the storefront.

4. 04Recurring Chinese-player feedback reaches a product owner

   Define clustering threshold, translation format, evidence links, decision owner, review cadence, status visibility, and the response back to players.

   Completion evidence

   A simulated recurring issue becomes a tracked product or copy decision and its disposition returns to the player-facing team.

4 checks

1. 01One time-zone-aware runbook names every launch actionCritical gate

   List freeze, backup, publish, entitlement check, download, smoke test, SDK verification, store update, announcement, monitoring, handoff, and review in exact order.

   Completion evidence

   Every action has UTC and local time, owner, backup, prerequisite, evidence, communication channel, and abort condition.

2. 02Severity levels connect player impact to decision authority

   Define examples, detection signals, acknowledgement target, technical owner, business authority, partner escalation, update cadence, and closure evidence.

   Completion evidence

   The same simulated incident receives the same severity and authority path from both studio and partner responders.

3. 03A cross-company incident drill reaches recoveryCritical gate

   Exercise at least one build blocker and one unsafe or unexplained data-flow scenario, including pause, investigation, player message, rollback, verification, and record keeping.

   Completion evidence

   The drill meets acknowledgement and recovery targets, and every failure becomes an owned action due before launch.

4. 04The first review has frozen inputs and decision outputs

   Set the observation window and capture availability, acquisition, activation, stability, localization, support, community, partner delivery, data, and cost signals.

   Completion evidence

   The calendar invite names the data snapshot, qualitative sample, decision owner, and required Keep / Change / Stop outputs.

4 checks