ML-Based Payment Date Prediction Adds Read-Only Fields to Invoice Records

What changed

A new Payment Date Prediction feature (disabled by default) adds machine-learning-driven fields to the invoice transaction record. When enabled, NetSuite analyzes historical payment data and populates predicted payment timing directly on each open invoice.

New system-generated fields (read-only)

• Predicted Payment Date — Primary Information section. Forecasted date payment will be received.
• Predicted Overdue Days — Primary Information section. Predicted days payment may be overdue.
• Predicted Payment Date Availability — Billing subtab. Whether a prediction exists for this invoice (may be blank during initial processing).
• Predicted Payment Date Last Updated — Billing subtab. Timestamp of the most recent prediction update.

User-editable fields (always available, no feature flag required)

• User-Estimated Payment Date — Billing subtab. Optional manual estimate.
• User-Estimated Payment Date Source — Billing subtab. Required when the estimated date is populated; dropdown values are Company Estimate and Customer Promised Date.

Prerequisites

• Minimum 12 weeks of payment history in the account.
• At least 50 paid invoices dated before the start of that 12-week window.
• Initial predictions may take up to 24 hours to appear after enablement.
• Not supported on Release Preview accounts.

What is unclear

Oracle does not publish the internal field IDs (e.g., custbody_* or native IDs like predictedpaymentdate) for the new fields. Without confirmed IDs, you cannot reliably reference them in SuiteScript, N/query, N/search, or workflow action scripts until you inspect the record in your own account. The source also does not state whether these fields are exposed via REST (/services/rest/record/v1/invoice) or SuiteTalk SOAP web services, or whether they appear in the Records Catalog.

The release version this feature ships in is not specified in the documentation page.

What to do

1. Enable the feature: Administrator role → Setup > Company > Setup Tasks > Enable Features → Accounting subtab → check Payment Date Prediction → Save.
2. Identify field IDs: After enablement, open an eligible invoice, inspect the record via SuiteScript Debugger, record.getFields(), or the Records Catalog to find the internal IDs of the four new system fields. Document them for your team.
3. Audit saved searches and scripts: If you have saved searches, SuiteScript customizations, or SuiteAnalytics workbooks that SELECT * or iterate all body fields on invoices, the four new read-only fields will now appear. Verify that downstream consumers (dashboards, integrations, CSV exports) handle the additional columns gracefully.
4. Review custom invoice forms: The new fields will appear on the standard invoice form. If you use custom transaction forms, decide whether to show or hide the prediction fields. Hide them if your users should not see ML-generated estimates.
5. Check integrations: If you push invoice data to external systems via SuiteTalk, RESTlet, or REST API, test whether the new fields are included in the payload. Add mapping or filtering as needed.
6. Validate data requirements: Confirm your account meets the 50-invoice / 12-week history threshold before expecting predictions. Accounts with sparse payment history will see blank prediction fields.