Skip to content

Partner feed onboarding

Purpose

This page defines the process for onboarding a new partner product feed into the platform.

Scope

This process applies to the following scenarios:

  • New partner integrations
  • Initial product feed ingestion
  • Validation and activation of partner data

It is intended for use by operations teams, integration engineers, and support personnel.

Preconditions

Before starting, confirm the following:

  • Partner has provided a valid CSV product feed
  • Required fields are present (minimum: sku, product_name)
  • API access is available (API key configured)
  • Target environment is identified (test or production)

What happens

  • Upload a partner feed
  • Create submission and validation jobs
  • Validate and process data through ETL
  • Store products in the database
  • Make products available via the API

Procedure

Step 1: Upload partner feed

Submit the product feed using the upload endpoint.

curl -X POST "http://api.example.com/feeds/upload" \
  -H "x-api-key: demo-secret-key" \
  -F "partner_name=Acme Corp" \
  -F "file=@sample_catalog.csv"

What happens

  • A feed_id is generated
  • A submission job (JSxxxxx) is created
  • A validation job (JVxxxxx) is queued

Example response:

{
  "feed_id": "FD00010",
  "status": "processing",
  "job_id": "JS00010"
}

Step 2: Verify feed registration

Retrieve the feed to confirm it was successfully registered.

curl -H "x-api-key: <api-key>" \
  "http://api.example.com/feeds/FD00010"

Validation checks

  • status is uploaded or validating
  • validation_job_id is present
  • partner_name and file_name should match input

Step 3: Monitor validation job

Check the status of the validation job.

curl -H "x-api-key: <api-key>" \
  "http://api.example.com/jobs/JV00010"

Expected status flow

  • queued → running → completed
  • If errors occur: failed

Step 4: Review validation results

Once the job is completed, review the ETL summary.

Example:

{
  "status": "completed",
  "message": "Products processed: 100. Inserted: 80. Updated: 10. Unchanged: 5. Skipped: 5."
}

What the results mean

  • Inserted: New products added
  • Updated: Existing products modified (data changes detected)
  • Unchanged: No changes from existing records
  • Skipped: Invalid rows (missing required fields or malformed data)

Step 5: Validate product availability

Confirm that products are accessible via the API.

curl -H "x-api-key: <api-key>" \
  "http://api.example.com/products?partner_name=<partner_name>"

Validation checks

  • Products are returned for the correct partner
  • Key fields (SKU, name, price, availability) are populated
  • Record counts align with ETL summary

Decision points

If validation fails

  • Review the error message from the job response
  • Check CSV format and required fields
  • Correct issues and re-upload feed

If high skip count

  • Inspect skipped rows for missing or invalid data
  • Confirm required fields (sku, product_name) are present
  • Coordinate with partner to correct source data

If products not returned

  • Confirm ETL job completed successfully
  • Verify filters in product query
  • Check database connectivity or ingestion logs

Results

After successful onboarding:

  • Partner products are available via /products endpoint
  • Feed status is validated
  • Data is persisted in the system
  • Partner can perform subsequent updates via feed re-submission

Additional details

  • Re-running validation on the same feed should not create duplicates
  • Product uniqueness is enforced by (partner_name, sku)
  • Updates are only counted when actual data changes occur
  • Raw files are stored for traceability and reprocessing