Payment ReconcilerDocumentation
QFind anything/
API Reference

Reconciliation

Reconciliation is the core domain—matching incoming transactions (payments received) to expected transactions (payments anticipated). LedgerMatch ranks candidates by confidence; you accept or reject matches.

What is reconciliation

Reconciliation is the process of linking incoming transactions (payments you have actually received) to expected transactions (payments you anticipated). The goal is to confirm that each received payment corresponds to an expected one, or to identify unmatched items that need attention.

Payment Reconciler automates this by:

  1. Letting you define expected transactions (merchant, amount, date, reference)
  2. Ingesting incoming transactions from manual entry, Stripe, CSV, or EBICS
  3. Using LedgerMatch to find candidate matches and rank them by confidence
  4. Letting you accept or reject matches—or auto-accept when confidence meets your threshold

Core concepts

ConceptDescription
Expected transactionA payment you anticipate receiving. Define merchant, amount, date, reference.
Incoming transactionA payment you have received. Comes from manual, Stripe, CSV, or EBICS.
MatchA proposed link between an incoming and an expected transaction. LedgerMatch scores candidates; you resolve.
ConfidenceMatch score (0–1). Higher means better fit. Auto-accept when above your threshold.

The reconciliation flow

  1. Expected — Create expected transactions via API. These are your anticipated payments.
  2. Incoming — Incoming transactions arrive via manual creation, Stripe sync, CSV upload, or EBICS import.
  3. Match — LedgerMatch is invoked (auto, manual, or scheduled) and returns ranked candidates with confidence scores.
  4. Resolve — Accept the correct match or reject. Accepting links incoming to expected and updates both to MATCHED.

See Match Flow for a detailed walkthrough of each step.

LedgerMatch

Matching is powered by LedgerMatch, an external service. It uses:

  • Embedding similarity — Semantic similarity of merchant/description text
  • Amount score — How closely amounts align
  • Date score — Proximity of transaction dates
  • Reference match — Whether reference IDs match

LedgerMatch returns ranked matches with a combined confidence score. Configure when and how it runs in LedgerMatch Integration.

Auto-match vs manual

  • Auto-match — When autoMatchEnabled is true, LedgerMatch runs as soon as a new incoming transaction is created. Matches above your matchingConfidenceThreshold can be auto-accepted.
  • Manual trigger — Call POST /matches/match with an incomingTransactionId to run matching on demand.
  • Scheduled job — A batch matching job can process PENDING incoming transactions on a schedule.

Configure these in Tenant Settings.