A borrower misses a payment. The system should apply the correct late charge, update the amortisation schedule, trigger the right arrears communication under CONC 7.9, and write a ledger entry that can be replayed in audit three years later. Most servicing platforms do some of that. Few do all of it, and fewer still do it in a way a compliance officer can actually inspect.
The question lenders should be asking is not which platform has the longest feature list. It is which platform handles the cases that go wrong, in a way that stands up to regulatory scrutiny, without requiring a developer every time a rule changes. What has to happen after drawdown is a useful reference for what that set of obligations actually looks like in practice.
The capabilities that separate serviceable platforms from good ones
Loan servicing software tends to look similar at the top level: repayment schedules, payment processing, customer communications. The differences surface in how a platform handles the things that do not go to plan. The following capabilities are where platforms diverge most sharply.
- Event-sourced ledger. Every state change is written as an immutable event, not an overwrite. When a regulator asks why a balance changed on a given date, the ledger produces a complete replay. An event-sourced ledger needs two clocks: the time the event occurred and the time it was recorded, which is critical for back-dated corrections and complaint resolution.
- Rule-based decision engine. Rate changes, arrears triggers, fee caps and forbearance eligibility are rules, not code. A platform that externalises these rules lets a compliance team update them without a deployment. One that hardcodes them creates a change-management problem every time Consumer Duty guidance shifts.
- Jurisdiction-aware configuration. A lender operating in the UK and Germany faces CONC and BaFin oversight simultaneously. Jurisdiction packs bind the platform to each market's consumer credit rules, so the same product can run under different regulatory regimes without bespoke builds.
- Automated but auditable servicing agents. Arrears outreach, payment plan offers and hardship assessments can be handled by agents, but only if every decision the agent makes is logged against the rule it applied. Agent Operating Procedures are the mechanism that makes that possible.
- Past-due and collections handling. The ledger needs to distinguish between a payment that is late, a payment that is in a formal arrangement, and a balance that has been charged off. Each state carries different regulatory obligations. How a ledger handles past-due credit determines whether your collections process holds up in audit.
How to evaluate a servicing platform before you commit
Most platform evaluations focus on integration timelines and pricing. The questions that matter more are about what the system produces when something goes wrong. A structured approach looks like this.
- Test the ledger under a realistic edge case. Take an actual scenario from your current book: a borrower who overpays, then disputes a charge, then enters a payment arrangement. Walk the scenario through the candidate platform and ask to see the ledger entries it produces. If the output cannot be explained line by line, the audit trail will not satisfy a regulator either.
- Ask where the rules live. Request a demonstration of how a fee cap change is applied. If the answer involves a developer and a release cycle, that is a governance risk. Rules that live in configuration, not code, are rules that compliance can own.
- Probe the arrears workflow against CONC. CONC 7 sets out what a lender must do before and during collections: contact obligations, breathing space under the Debt Respite Scheme, fair treatment requirements. Ask the vendor to map their arrears workflow to those obligations explicitly. Vague answers about flexibility are not the same as a compliant process. AI-driven dunning can accelerate collections, but only within a workflow that is already correctly structured.
- Understand the migration path. Switching servicers with a live book is one of the highest-risk operations in lending. Ask for a worked example of how a mid-cycle loan transfers without losing payment history or accrued interest. Migrating a live credit book without a big bang explains what a low-risk migration actually requires from the receiving platform.
- Check what the platform does not cover. No single platform handles every adjacent need. Understand whether banking-as-a-service connectivity, credit decisioning and product configuration are native or require third-party stitching. What a loan management system actually has to do sets out the full scope of obligations a platform needs to meet.
A platform that answers these questions with evidence rather than assurances is one that has been built for lenders who operate under regulatory scrutiny, not one that has been built for a demo.
Why "configurable" is not the same as "compliant"
Vendors use configurability as a selling point, and it is a legitimate one: a platform that can model a revolving credit facility, a fixed-rate personal loan and a buy-now-pay-later product without bespoke development is genuinely valuable. But configurability describes what a platform can represent, not what it will enforce. A system that lets you configure a 30-day grace period does not automatically apply that grace period correctly under the Consumer Credit Act 1974, or produce the Section 86 notice at the right moment, or suspend interest accrual when the Debt Respite Scheme applies.
Compliance is not a configuration option. It is the set of rules that govern what the configuration is allowed to do. What your platform must handle across credit products is a useful reference for the range of obligations a genuinely compliant platform needs to enforce by default, not by hope. Kontor builds those obligations into the platform itself, so that configurability operates within a compliant envelope rather than outside it.
The best loan servicing software is the one that handles a missed payment, a rate change and an upheld complaint correctly, produces a ledger entry for each that a regulator can read, and does it without a developer on call. That is the standard worth evaluating against.