Online Troubleshooting and Retrospective for Ad Creative Review: Conversation Logs and Auditing

The data for this category primarily comes from advertising placement platforms, landing page resources, product qualification documents, and

What the Data for This Category Looks Like

The data for this category primarily comes from advertising placement platforms, landing page resources, product qualification documents, and advertising law compliance clauses. Update frequency is real-time or daily batches. Daily retained documents include ad creative files, landing page screenshots, compliance opinions issued by reviewers, and creative modification history records. Data fields cover creative associated ID, review node timestamp, violation type tag, reviewer identifier, and pre- and post-modification content comparison items. Units are primarily characters, timestamps, and status codes.

Constraints Imposed on Conversation Logs and Auditing Workflows

Real-time or daily batch data updates require conversation logs to support quick retrieval by creative associated ID, to avoid mixing data across batches. Retention of multiple document types requires audit logs to simultaneously associate creative files, review opinions, and modification records, to ensure full-link traceability. The time-consuming landing page consistency check process generates long conversation content. This requires log storage to adapt to large character volume conversation fragments, while supporting filtering by violation type and review node. The severity of error consequences requires audit logs to fully record the context of each review operation, without omitting key modifications or judgment bases. It also requires support for exporting compliance audit reports by time range.

How to Configure the System

Configuration ItemRecommended SettingRationale
Log Storage Quota1000 MB/review batchAdapts to the size of multi-turn conversations and associated documents generated by a single batch of creative reviews
Retrieval Associated FieldsBy chatId, responseChatItemId, creative IDMatches business requirements for quickly locating logs by single review session, single reply, or single creative
Log Retention Period180 daysMeets conventional retention requirements for compliance audits
Full Log SwitchEnable detail full recordingCovers complete context of review opinions and pre- and post-modification content
Session Timeout Threshold600 secondsAdapts to long-process review links such as landing page consistency checks
Audit Export PermissionOnly review supervisor rolesPrevents unauthorized personnel from viewing compliance-sensitive review logs

Three Common Misconfigurations

  • Historical reply content is empty after opening a review session, with no clear error prompt on the interface. Cause: The full log recording for the detail parameter is not enabled, resulting in conversation context not being persistently stored.
  • No matching results are returned when specifying MCP service review logs in the retrieval workflow, with a 404 status code. Cause: The MCP service identifier is not bound to chatId, causing the retrieval dimension to not match the actual stored log fields.
  • The exported compliance audit log lacks associated information for landing page screenshots. Cause: The log configuration only records text-based review content, and the audit switch for external document association is not enabled.

How to Verify Correct Configuration

  • Access the log management page of FastGPT v4.17.0, enter the chatId of the test creative, and confirm that the complete log of the corresponding review session can be retrieved.
  • Trigger a complete landing page consistency check process, and verify that the log list contains complete review opinions and modification records.
  • Export audit logs for a specified time range, and confirm that the exported file includes core fields such as creative ID and review node timestamp.
  • Switch to a non-review supervisor role, and confirm that the audit log export function is not accessible.

Note: The values listed above are common starting points. Adjust them based on your own samples and testing.

Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-10-19.