Deployment and Upgrade for Brand Agency Operation Revenue Yield

Brand agency operation revenue yield and market data comes from three sources: e-commerce platform background export files provided by brand partners

What the data for this category looks like

Brand agency operation revenue yield and market data comes from three sources: e-commerce platform background export files provided by brand partners, API-synced data from third-party business analysis tools, and manually entered records for offline service settlements. Full service data for the prior day is updated every early morning. Data can be filtered by service store and cycle. Supported document formats are CSV and Excel. A single file can include multi-dimensional data for multiple stores. Fields include: store unique identifier, service cycle, total transaction amount, platform commission, agency operation service fee, actual settlement revenue, and core SKU sales. Units are yuan, date, and quantity respectively. There is no unified fixed header order. Some fields may contain null values.

Constraints for Deployment and Upgrade

Since data sources include multi-platform export files and API-synced data, multi-format parsing rules must be configured during deployment to accommodate field differences across platform exports. The fixed daily update requirement demands precise scheduled synchronization tasks during deployment, to avoid early task triggers before data is generated. The mixed multi-store document structure requires compatible parsing logic for new store fields during upgrades, to prevent old configuration versions from failing to process new daily report files. Data includes sensitive settlement-related amount fields, so data desensitization rules must be configured during deployment to prevent unauthorized access.

Configuration Settings

Configuration ItemRecommended ValueRationale
UPLOAD_FILE_MAX_SIZE2000 MBDaily report files for brand agency operations may include monthly summary data for multiple stores, resulting in large single-file size
PARSE_FILE_TIMEOUT_SECONDS300 secondsParsing multi-store data requires traversing multiple worksheets, which takes significant time
DATA_SYNC_CRON0 2 * * *Daily report data updates T+1. Triggering synchronization tasks at 2 AM daily aligns with business rhythms
RERANK_MODEL_PATH/data/models/bge-reranker-largeBrand agency operation revenue yield data requires precise matching of store and revenue fields. Reranking models improve recall accuracy
FE_DOMAINUse http://localhost:3000 for local deployments, fill in the corresponding domain for online deploymentsSpecify the frontend access domain during local deployments to allow cross-origin requests for file uploads. Match the actual access address for online deployments
MAX_CONTEXT1500 charactersDaily report broadcast content must be concise to avoid exceeding model context limits

The parameter values provided on this page are common starting points for configuration. Actual values are affected by material form, data volume, and business rules. Specific issues require individual analysis. It is recommended to test on your own samples before finalizing settings.

Three Common Misconfigurations

  • An ERRO-prefixed log error appears after running docker-compose up -d. The cause is failing to mount the correct file directory in the WSL2 environment, preventing FastGPT from reading locally uploaded daily report files.
  • Only the IP address is entered for FE_DOMAIN without adding a port number, causing the frontend to fail to call the backend file upload interface and returning a 403 status code.
  • Using the default recall rule without configuring RERANK_MODEL_PATH, resulting in low field matching accuracy for revenue yield data and insufficient returned result counts.

How to Confirm Successful Configuration

  • Upload a sample daily report file for brand agency operations, and check if the parsed fields match expected values to confirm the data parsing rule is active.
  • Manually trigger a scheduled synchronization task, and check if background logs show successful task execution with no timeout or error messages.
  • Call the reranking model interface, pass candidate fields for revenue yield data, and check if the returned results are sorted in line with business logic.
  • Access the file upload module of the frontend page, confirm that local files can be selected and uploaded normally with no cross-domain error prompts.

Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.