Introduction

In countries with financial services advertising regulations, Meta may require advertisers to verify their identity and financial license. Regulators will provide Meta with a dataset of financial licenses, which Meta uses to validate licenses submitted by advertisers. This document covers how regulators securely share license data with Meta and the technical specifications for integration.

Overview

The integration works as follows:

1. Meta provides SFTP credentials- Meta sends login details to the regulator's confirmed email address.
2. Regulator uploads two files- A config.json (metadata) and a CSV file(license data) are uploaded to Meta's secure SFTP server. Instructions for uploading files.
3. Meta ingests the data- Files are processed periodically.
4. Regulator regularly uploads- Regulators are responsible for routinely uploading updated files weekly.

What is config.json?

The config.json file is a manifest containing metadata required for processing. It must be included with every upload.

Fields

Example

{
  "version": "v1",
  "load_type": "full",
  "mode": "ingestion",
  "support_email_addresses": [
    "support@domain.com",
    "support2@domain.com"
  ]
}

License Data (CSV)

License data is provided as a CSV file. The file can have any name. There are two load types:

• Full Load- Contains the complete set of all license data.
• Incremental Load- Contains only the changes (deltas) since the last upload.

General Rules

• Column headers must be on the first row.
• All column headers must be unique.
• All required columnsmust be present.
• Each row must have a unique Credential_Id .
• Row-level errors do notblock the entire file. Valid rows are still ingested. For example, if 50 out of 100,000 rows have errors, the remaining 99,950 rows are processed normally.

Full Load Schema (V1)

To prevent impersonation, fraud, and abuse on Meta's advertising platforms, a connection check is essential. The uploaded financial license data must provide at least one of the following for each license: phone numbers, email addresses, or website URLs to perform this check. This ensures that the advertiser attempting to advertise is using a license connected to the entity in the verification.

Incremental Schema (V1)

The incremental schema uses the same columnsas the full load schema, plus one additional required column:

Operation Rules

• UPSERT- Adds a new license or updates an existing one. Follows the same required/optional field rules as the full load schema.
• DELETE- Removes a license. Only Credential_Id is required; all other fields are optional.

Example Incremental Load Scenario

An initial data set contains three licenses: A, B, C.

The following changes occur afterward:

• License B is updated.
• License C is deleted.
• License D is added.

The incremental CSV would contain:

License A is not includedbecause it had no changes.

The timestamp of your last ingested dataset is available via the ingestion_history/ folder on SFTP (see below).

SFTP Upload & Ingestion Process

How to Upload

1. Log into the SFTP server using the credentials Meta provided.
2. Upload both config.json and your CSV data file to your default directory.
3. Wait for processing. Files can be validated several times a day using validation mode in config.json but data can only be ingested once daily.

What Happens After Upload

• After processing, both files are automatically movedto either a validation_history/ or ingestion_history/ directory, depending on the mode set in config.json . If the uploaded CSV file contains any row-level errors, such as missing required values or invalid values, a separate CSV file detailing those errors will be generated. Before performing an ingestion, use validation mode to confirm your files will be processed correctly.
• Each run creates a timestamped folder (e.g., 2026-03-11T12-23-56Z/ ) containing the exact files used in the history directory.
• If your files are still in the default directory (not in ingestion_history/ or validation_history/ ), processing has not occurred yet.
• Retention:The SFTP server retains files for 30 days only.

Example: Before & After Ingestion

Below illustrates the file structure before and after a new ingestion on 3/11/26 involving data1.csv and config.json with mode ingestion .

Before Ingestion

config.json
data1.csv
ingestion_history/
  └── 2026-03-06T04-20-17Z/
       ├── config.json
       ├── license_data.csv
       ├── csv_errors.csv
validation_history/
  └── 2026-03-11T05-18-14Z/
       ├── config.json
       ├── data1.csv
       ├── csv_errors.csv
  └── 2026-03-06T01-15-24Z/
       ├── config.json
       ├── license_data.csv
       ├── csv_errors.csv

After Ingestion

ingestion_history/
  └── 2026-03-11T12-23-56Z/
       ├── config.json
       ├── data1.csv
  └── 2026-03-06T04-20-17Z/
       ├── config.json
       ├── license_data.csv
       ├── csv_errors.csv
validation_history/
  └── 2026-03-11T05-18-14Z/
       ├── config.json
       ├── data1.csv
       ├── csv_errors.csv
  └── 2026-03-06T01-15-24Z/
       ├── config.json
       ├── license_data.csv
       ├── csv_errors.csv

To trigger another ingestion, upload new files to your default directory again.

Important Notes

• Processing time:Data may take up to 1 day to appear in Meta's system after ingestion.