Skip to main content
Disbursement is in beta. Track B (settlement-config batches → bank CSV) is available; the automated rail engine (Track A) is on the roadmap.
klikit Disbursement answers a single question: when a merchant collects money, how is it split between the parties involved and what fees apply — and how do we pay everyone? You define the split and fees once as a reusable template, bind it to a merchant as an instance (with bank details), then run batches that compute each party’s payout and export a bank file a human takes to the bank. There is no automated money movement in beta — the exported CSV is the deliverable. Every batch is reviewed and approved by a second person before it can be exported.

​
The loop

1

Create a template

Define the split parties (as integer percentages that must total 100%) and the fees (MDR, service charge, tax). See Templates.
2

Instance it to a merchant

Bind the template to a business (and optionally a brand), add each payable party’s bank details, and verify them. See Instances.
3

Run a batch

Feed gross amounts (manual grid or CSV), compute the per-party split, review the lines and any flags, then get a second-tier approval. See Batches.
4

Export the bank file

Download the bank CSV for the approved batch and submit it to the bank.

​
Vocabulary

The engine produces batches — reviewed CSV files a human takes to a bank. This is deliberately distinct from rail runs (automated pushes), which are a future capability. Keep the two terms separate.

Template

A reusable, versioned settlement config: split parties + fees. One template can be instanced for many merchants.

Instance

A template bound to a specific merchant (business/brand), with per-party bank details and optional overrides.

Batch

A period’s disbursement run for one template: gross inputs → computed per-party lines → approval → bank file.

Party

A recipient in a split — the merchant, klikit, or any other counterparty (e.g. a landlord).
See Concepts for the money model (basis points, minor units, the split algorithm) and the permission model.