Oracle Fusion Cloud ERP and HubSpot integration

What moves between them

The main flow is Oracle Fusion into HubSpot. ml-connector polls Oracle Fusion every 10-15 minutes for new and updated invoices, payments, and customer records. Invoices are transformed into HubSpot's required shape (contact association, line items), then posted as draft records until finance confirms them, at which point ml-connector updates the status. Payment records from Oracle Fusion flow into HubSpot payment objects, always linked to the matching contact. Customer master data (company name, billing address, account number, industry) syncs from Oracle Fusion into HubSpot company accounts whenever a customer is created or updated in Fusion.

How ml-connector handles it

ml-connector stores the Oracle Fusion pod URL and OAuth2 credentials encrypted and refreshes the bearer token when an API call returns 401, so token expiry does not interrupt polling. On the HubSpot side, it uses the Private App access token on every request. Because Oracle Fusion has no outbound webhooks, ml-connector polls both the invoices endpoint and the payments endpoint using a LastUpdateDate filter to fetch only records changed since the last run, avoiding full-table scans. Every Oracle Fusion invoice is examined for its associated customer record in Oracle Fusion; that customer is matched to the HubSpot contact with the same external ID, or created as a new contact if none exists. Line items are read from the Oracle Fusion invoice and posted to HubSpot as separate line-item objects linked to the invoice. Payments are similar: the payment amount and date are extracted from Oracle Fusion, the associated customer is matched in HubSpot, and the payment is posted to HubSpot's payment object. HubSpot rate limits vary by tier, so ml-connector uses exponential backoff on HTTP 429 responses. Every transaction is logged with full audit trail and can be replayed if a write fails.

A real-world example

A mid-market B2B software company runs Oracle Fusion Cloud ERP for Accounts Receivable and GL, and uses HubSpot for sales, account management, and customer communications. Before the integration, the finance team manually exported invoices from Oracle Fusion into a spreadsheet each week and uploaded them into HubSpot so the sales team could see outstanding invoices and payment status. Customer account numbers and names were maintained in both systems, creating a frequent source of misalignment. With Oracle Fusion and HubSpot connected, every new invoice posted in Fusion automatically appears in HubSpot, linked to the correct customer, so sales can see in real time which invoices are outstanding. When a customer pays in Oracle Fusion, the payment posts to HubSpot at the same moment, eliminating manual upload and reducing account reconciliation errors.

What you can do

• •Poll Oracle Fusion invoices and payments on your defined schedule and post them to HubSpot with automatic contact and line-item associations.
• •Sync Oracle Fusion customer records into HubSpot company accounts so sales and finance see the same account data.
• •Authenticate Oracle Fusion with OAuth2 and HubSpot with a Private App access token, managing token refresh and rate-limit backoff automatically.
• •Track every invoice and payment record with a full audit trail and allow replays if a write fails in HubSpot.
• •Filter Oracle Fusion records by LastUpdateDate to poll efficiently without scanning the entire dataset on each run.

Questions

How does ml-connector handle the association of invoices to contacts in HubSpot?

ml-connector reads each Oracle Fusion invoice and identifies the customer account. It then matches that customer to the HubSpot contact with the same external ID, or creates a new contact if no match exists. Line items from the Fusion invoice are posted separately to HubSpot and linked to the invoice, so HubSpot records show the complete invoice structure.

Does Oracle Fusion invoicing directly push to HubSpot, or does ml-connector poll?

Oracle Fusion has no outbound webhook system for cloud connectors, so ml-connector polls the invoices and payments endpoints on your defined schedule (typically 10-15 minutes) using a LastUpdateDate filter to fetch only changed records since the last run. This approach is efficient and does not require additional Oracle Integration Cloud middleware.

How are rate limits and token expiry handled in this integration?

ml-connector refreshes the OAuth2 bearer token when an API call returns 401, so token expiry does not interrupt polling. On HubSpot, it detects HTTP 429 rate-limit responses and backs off with exponential retry, respecting your tier-based limits (100 requests per 10 seconds for Free/Starter, 190 for Professional/Enterprise).