Deltek and Amazon Seller Central integration

What moves between them

The main flow runs from Amazon Seller Central into Deltek. After each settlement closes, ml-connector reads the Amazon financial events and the settlement report, groups the sales, fees, refunds, and adjustments, and posts them into Deltek Vantagepoint as journal entries or AR invoices mapped to the matching GL accounts and project. Order and transaction detail is read from Amazon so each posting ties back to the underlying payout. Postings move one direction only, since Amazon settlement data is read-only and Deltek is the system of record for the ledger. Reference data such as GL accounts, fee categories, and the target project is aligned up front so every line lands on a valid Deltek account.

How ml-connector handles it

ml-connector stores both credential sets encrypted. On the Amazon side it exchanges the seller refresh token for a short-lived access token, caches it, and refreshes before the one-hour expiry, sending the marketplace ID on each call and following NextToken paging until it is absent. On the Deltek side it accepts the full tenant URL per customer, requests a bearer token with the password grant, and refreshes when a call returns 401. Because Amazon settlement is async, ml-connector waits for a settlement to release, pulls the settlement report through the create-report, poll, and download steps, and reconciles it against the Finances API events so disbursements are not posted twice. Amazon sales, fee types, and refunds are mapped to Deltek GL accounts before any posting, and because Deltek has no idempotency header, ml-connector queries Deltek for an existing entry keyed to the settlement before it writes. Amazon rate limits return HTTP 429 under a token-bucket scheme, so ml-connector backs off with jitter and retries, and every record carries a full audit trail and can be replayed if a downstream call fails.

A real-world example

A consulting firm with about 80 staff runs Deltek Vantagepoint for project accounting and also sells its training workbooks and templates as a side line through Amazon Seller Central. Before the integration, the finance team downloaded each Amazon settlement report, added up gross sales, referral fees, FBA fees, and refunds in a spreadsheet, and hand-keyed a journal entry into Vantagepoint every two weeks, often after the payout had already cleared the bank. With Deltek and Amazon Seller Central connected, each settlement posts into Vantagepoint automatically, split across the right revenue and fee accounts and tagged to the marketplace project. Bank reconciliation matches the Amazon payout to the ledger on the first try, and the manual spreadsheet step is gone.

What you can do

• •Post Amazon settlement revenue, fees, refunds, and adjustments into Deltek Vantagepoint as journal entries or AR invoices after each payout.
• •Map Amazon sales and fee categories to the correct Deltek GL accounts and project.
• •Reconcile the Amazon settlement report against Finances API events so a payout is posted only once.
• •Bridge Login with Amazon refresh-token auth and the Deltek tenant password grant, refreshing both token types automatically.
• •Poll Amazon on a schedule, follow NextToken paging and the async settlement report flow, and retry with backoff on HTTP 429.

Questions

Which direction does data move between Deltek and Amazon Seller Central?

The flow runs from Amazon Seller Central into Deltek. Settlement revenue, fees, refunds, and the underlying transaction detail are read from Amazon and posted into Deltek Vantagepoint as journal entries or AR invoices. Amazon settlement data is read-only, so ml-connector never writes financial entries back to Amazon.

How does the integration handle Amazon's async settlement and lack of plain webhooks?

Amazon delivers push events only through AWS EventBridge or SQS, not HTTP webhooks, and settlement is async, so ml-connector polls instead. It waits for a settlement to release, pulls the settlement report through the create-report, poll, and download steps, and reconciles it against the Finances API events before posting to Deltek.

How are duplicate postings into Deltek prevented?

Deltek Vantagepoint has no native idempotency header, so duplicate prevention is the connector's job. ml-connector queries Deltek for an existing entry keyed to the Amazon settlement before it writes, and every record carries an audit trail so a replay after a failed call does not double-book the ledger.