The Amazon data API, without the SP-API plumbing

Create a key under Integrations → REST API in DataDoe, give it a name and an expiry date, and send it in the datadoe-api-key header on every request. The MCP server takes its own key in the datadoe-mcp-key header, while Claude and ChatGPT sign in with OAuth instead. The API reference and its OpenAPI JSON are public.

Data comes out as exports. List your accounts with GET /api/v1/util/sellers-and-vendors, find a table with GET /api/v1/exports/sources, then send POST /api/v1/exports with the account IDs, table, columns, date range and output type. Exports are asynchronous: poll the export, usually ready in under 30 seconds, and download the file from GET /api/v1/exports/{exportId}/raw. Filters, grouping and aggregations run on DataDoe's side.

Each Seller Central, Vendor Central or Amazon Ads account is connected once per marketplace, and its owner authorizes DataDoe on Amazon's own consent screen. From your own product, create a single-use invitation with POST /api/v1/util/sellers-and-vendors/invite, choose read-only or read and write, and your redirect URL receives the new account ID. If Amazon stops accepting an authorization, a reconnect invitation keeps the same ID and data.

The daily fetch starts at 2am in each account's local time, and most data is in by 5am. Order items, listings, Vendor purchase orders, ad performance and inventory health also update during the day. Search Query Performance, repeat purchase and market basket refresh weekly. On connect, DataDoe backfills about two years, recent data first: 730 days of order items and reimbursements, 720 days of Vendor sales, traffic and inventory, 365 days of settlements, and around 60 days for most Amazon Ads reports. Notification and Marketing Stream tables are off by default.

Yes. DataDoe shares your data through a private Google Analytics Hub listing; subscribing creates a read-only linked dataset in your own Google Cloud project, in us-east4. Query it from BigQuery, dbt, Looker, notebooks or an agent, and join it with your own tables. On the Base Plan each query uses 0.2 AI Tokens for integrated data or 0.5 for raw data, and Google bills the BigQuery compute to your project.

Yes, with webhooks. DataDoe sends an HTTPS POST with a JSON body to your endpoint for the message types you choose, such as AMAZON_ORDER_UPDATE, AMAZON_ANY_OFFER_CHANGED or AMAZON_LISTINGS_ITEM_ISSUES_CHANGE. Each message has a messageId for deduplication, you can add a secret header to verify requests, and a failed delivery pauses the webhook and emails your team. Each delivered message costs 0.05 AI Tokens.

Yes, through Actions: POST /api/v1/actions updates listings, prices, orders, shipments, A+ Content, Amazon Ads and Vendor purchase orders. Writes are off by default, each Action type is switched on separately, you can preview any change as a dry run first, and every change is logged with Amazon's answer. See Actions for the full list.

The Base Plan is $97 a month with 2,000 AI Tokens and 10 million data rows included, and you can try it free for 14 days. Calls that list accounts or read the spec are free; an export from a standard table uses 2 AI Tokens and from a premium table 5. Extra tokens are only charged if you opt in to a spending limit or buy a bundle.

The data is plain tables with documented columns, not a proprietary format. Export any table as CSV, JSON, TSV, XML or Excel through the API, or copy what you need from the BigQuery linked dataset into a dataset you own with a regular query. Account connections are authorized on Amazon, so you can revoke DataDoe's access there at any time.