New: the Kontor regulatory map, with UK, EU and US obligations mapped to controls and evidence. Read it

[ Blog / Platform ]

What a loan management system actually has to do

The Kontor team·6 October 2026·5 min read

A payment arrives three days late. The borrower has a fixed-rate agreement, but your system has already accrued interest at the wrong daily rate because a rate-change event was applied to the wrong effective date. Your operations team corrects the balance manually. Six months later, a complaint lands, and your audit trail shows a number that was overwritten, not a record of what happened and why.

That sequence is not unusual. It happens when a loan management system is built around a mutable balance rather than an immutable record of events. The fix is not a new UI or a faster API. It is a different approach to the ledger itself.

What a loan management system must contain

The term covers a wide surface area. In practice, a system that can support a regulated consumer credit book needs to handle all of the following without any one component undermining another.

  • An event-sourced ledger. Every financial movement, payment, fee, interest accrual, rate change or write-off, is stored as an immutable event with an effective date and a processing date. The current balance is derived, not stored. This is what makes retrospective correction possible without destroying the audit trail. It is also what separates business time from system time in a way that satisfies regulators.
  • A decision engine. Origination and ongoing credit decisions need to be reproducible. A decision that cannot be replayed with the inputs that existed at the moment it was made cannot be explained under Consumer Duty or justified in a Subject Access Request.
  • Servicing and collections logic. Payment allocation, arrears staging, forbearance treatment and default triggers all need to follow documented rules, not ad hoc operational judgement. Under CONC 7, a lender must be able to show that arrears treatment was consistent and appropriate.
  • Jurisdiction-aware compliance controls. A rate cap that applies in one market, a cooling-off period required by the Consumer Credit Directive, a TILA disclosure requirement under Regulation Z: these are not configuration options. They are binding obligations that must be encoded into the system's behaviour, not left to individual users to remember.
  • A complete audit trail. Every state change, every agent action, every automated decision must be recoverable in sequence. When a complaint is upheld and a remediation payment is required, the system must be able to reconstruct the correct balance from first principles.

How the ledger controls everything downstream

The quality of every other function, servicing, collections, reporting, complaints, depends on the integrity of the ledger. A ledger that allows direct balance edits will eventually produce a book that cannot be reconciled. Getting the ledger right follows a specific sequence.

  1. Separate effective date from processing date. When a payment is received on a Tuesday but covers a Monday obligation, the ledger must record both dates. Accrual, interest and arrears calculations all depend on the effective date. Reporting and reconciliation depend on the processing date. Conflating them is the root cause of most balance disputes.
  2. Make every event immutable. A correction is a new event that references the original, not an overwrite. If a rate was applied incorrectly, the correction event records the error, the corrected value and the timestamp of the correction. The original event remains. This is the only way to produce a reliable audit trail for a regulatory review or an upheld complaint.
  3. Derive balances, never store them. The outstanding principal, accrued interest and total amount payable are computed from the event stream on demand. This means a change to a past event, applied as a correction, flows through to every subsequent calculation automatically. Recomputing a balance after a disputed payment allocation takes seconds, not a manual rebuild.
  4. Bind compliance rules to the jurisdiction, not the user. A system that requires a user to remember which markets require a 14-day withdrawal right, or which rate cap applies to a short-term credit product, will eventually produce a breach. Jurisdiction packs encode those rules into the platform so that the system refuses a non-compliant action rather than permitting it and hoping someone notices.
  5. Make agent actions as auditable as ledger events. Servicing decisions, payment deferrals, arrears escalations, are not metadata. They are part of the loan's history and must be recorded with the same rigour as financial events. Auditable servicing agents apply documented procedures and leave a recoverable record of every action taken.

A lender operating under Consumer Duty needs to demonstrate, at any point in a loan's life, that the customer was treated fairly and that the system's behaviour was consistent with the documented product terms. A ledger built on this sequence can produce that demonstration. One built on mutable balances and manual corrections cannot.

Why replacing an existing system feels riskier than it is

The most common objection is that a live credit book cannot be moved without a period of parallel running that is expensive, slow and operationally hazardous. That is true of a migration designed around a big-bang cutover. It is not true of a migration designed around event replay. If the new system can ingest historical events and derive the same current balances as the legacy system, you can validate correctness account by account before any live traffic moves. Migrating a live credit book without a big bang describes how that works in practice.

The real risk is staying on a system whose ledger cannot be trusted. A balance that was manually corrected three times cannot be reconstructed. A complaint that arrives two years after the account was settled cannot be answered. Those risks compound over time. The migration risk is bounded and finite.

A loan management system earns its keep not in origination or reporting, but in the moment someone asks why a balance is what it is, and the system can answer without anyone opening a spreadsheet. Kontor is built on that principle: an event-sourced ledger, a decision engine and auditable servicing, bound to the rules of each market the lender operates in.

[ More from the blog ]

Run your receivables on Kontor.

Bring a product spec or a portfolio extract. We'll show you how the ledger, decisions and agents would run it, and what your regulator would see.