Match Suggestions for Imported Bank Transactions Now Include AI-Assisted Matching via NetSuite Next

What changed

The Match Bank Data page now surfaces up to five match suggestions per imported bank or credit card transaction on the Match Suggestions subtab. Three suggestion types are supported:

• Transaction match — pairs one or more imported bank lines with existing GL transactions based on system and user matching rules.
• Payment application — proposes creating a payment record, applying it to an open invoice or payable, and matching the resulting payment to the imported line.
• Auto-create — proposes creating a transaction from a configured auto-create rule and matching it to the imported line.

Enrichment-based matching

For ambiguous transactions that do not hit a clean rule match, NetSuite now uses enrichment-based matching to improve suggestion quality. Oracle's documentation does not specify which enrichment data points are used; see Enriched Bank Data for Transaction Matching in SuiteAnswers for implementation details.

Transaction Matching Assistant (NetSuite Next)

If your account is enrolled in the NetSuite Next program and the Transaction Matching Assistant feature is enabled, an AI layer evaluates all available suggestions and recommends the most likely match. NetSuite Next is rolling out in phases — accounts receive in-app notifications when eligible. Oracle provides no detail on the model, confidence thresholds, or whether the assistant's recommendation can be overridden per-transaction, so treat this as a black-box ranking aid until more technical documentation appears.

Automatically Submit Exact Matches preference

The accounting preference Automatically Submit Exact Matches is enabled by default. When active, matches generated by system and user matching rules that score as exact are auto-submitted without manual review. Only non-exact suggestions remain on the subtab for human review. This preference existed previously but is worth verifying in any account where auto-submission could conflict with internal controls or approval workflows.

What to do

1. Verify the auto-submit preference. Navigate to Setup > Accounting > Accounting Preferences and confirm whether Automatically Submit Exact Matches is appropriate for your reconciliation process. Disabling it forces all suggestions through manual review.
2. Review matching rules. Because suggestion quality depends on system and user matching rules, audit your existing rules at Setup > Accounting > Reconciliation Rules. Poorly scoped rules may generate false-positive exact matches that auto-submit silently.
3. Check NetSuite Next eligibility. If you want the AI-assisted Transaction Matching Assistant, confirm your account's NetSuite Next enrollment status via in-app notifications or contact your account manager. Enable the feature at Setup > Company > Enable Features once available.
4. SuiteScript / integration impact. Oracle does not document new records, fields, or API endpoints associated with match suggestions in this article. If you have scripts that interact with bank reconciliation records (banktransaction, reconcile) or custom workflows triggered on match/reconciliation events, test them against the new suggestion types — particularly auto-submitted matches — to ensure no downstream breakage.
5. Rejected suggestions are regenerable. Rejecting a suggestion does not permanently discard it. Re-importing bank data or manually re-running reconciliation rules will regenerate all suggestion types, including previously rejected ones. Factor this into any custom rejection-tracking logic.

Gaps in the documentation

This article is light on technical specifics. Notably absent:

• No SuiteScript record type or field IDs for match suggestions.
• No REST or SOAP endpoint for reading or acting on suggestions programmatically.
• No detail on the enrichment data model or how to influence enrichment-based scoring.
• No confidence score or threshold documentation for the Transaction Matching Assistant.

Until Oracle publishes deeper technical references, treat match suggestions as a UI-only workflow and verify any programmatic interaction against a sandbox.