
The credit limit set at onboarding and never revisited
The limit was set from a credit report and an optimistic sales forecast. Four years of actual payment behaviour has never been used to revise it.
- Credit limits are usually set at onboarding, when the least information about the customer exists.
- Payment behaviour accumulated afterwards is a better predictor than the external data used at the outset, and it is rarely used.
- Limits that are never revised block good customers and leave exposure with deteriorating ones.
- A periodic review driven by behaviour and exposure, rather than by a calendar, is more effective and less work.
- The same behavioural data feeds the credit loss estimate, so the analysis serves two purposes.
A credit limit is set at the moment you know least about a customer. They have no payment history with you, the forecast of what they will buy is a sales estimate, and the external data is generic.
From that point onwards you accumulate exactly the information that would improve the decision. How they actually pay, how often they dispute, whether they take deductions, how their volume has changed.
In most companies none of that ever feeds back into the limit. It sits at the number set at onboarding until an order is blocked or a bad debt occurs.
The result is a control that is simultaneously too tight for good customers and too loose for deteriorating ones, which is the worst of both. It also produces friction at exactly the wrong moment, when an order is placed and a limit blocks it, so the exception is resolved under commercial pressure by someone who wants the order shipped. Decisions made that way are rarely documented, which is the familiar problem in controls that leave no record, and it is why the review should happen before the order rather than at it, as reporting that reaches people early allows.
What the original limit was based on
Typically a credit reference agency score, a published financial statement that may be over a year old, and an expected trading volume supplied by the salesperson who won the account.
Each of those is reasonable at the time and each has a short useful life. Scores change, statements age and expected volumes are optimistic more often than not.
What replaces them is your own experience of the customer, which is more specific and more current than any external source.
That experience sits in the receivables ledger, unused, in the way that most useful data sits unused when nobody owns it, as unowned data describes.

Four behavioural signals worth using
Each is derivable from your own ledger and each is more predictive than an external score for an established customer.
Why calendar reviews fail
An annual review of every account is a large exercise performed by a small team, so it becomes a formality or it slips.
It also allocates attention evenly across a population where risk is highly concentrated, which means the accounts that matter get the same treatment as the ones that do not.
A trigger based review works better. Accounts are reviewed when something changes: a deterioration in payment behaviour, exposure approaching the limit, a change in trading pattern or an external alert.
That concentrates effort where the information has changed, which is both less work and more effective, and it fits the pattern in thresholds that should respond to conditions.
An annual review treats every account alike. Risk does not distribute itself that way.
Triggers worth defining
Four triggers cover most of the risk, and each is calculable from data you already hold.
| Trigger | Threshold | Response |
|---|---|---|
| Days beyond terms deteriorating | A sustained increase over three consecutive periods. | Review the limit and the terms before the next large order. |
| Exposure near the limit | Utilisation above a defined percentage. | Decide deliberately whether to raise it rather than blocking an order at the counter. |
| Dispute frequency rising | More than a defined number in a rolling period. | Investigate the cause, since it may be your billing rather than their solvency. |
Credit policy, limit authority and any use of external data should follow your own governance and applicable data protection requirements.
The limit was set on the day you knew least, and it has not moved since.
ChatFinBuilding the review process
Four steps, replacing an annual exercise with something continuous and considerably smaller.
Score the book on behaviour
Using your own ledger data, so every account has a current behavioural picture rather than an onboarding one.
Define the triggers and thresholds
Written down, so a review is prompted by data rather than by someone remembering.
Set review authority by size
Small changes handled by credit control, large ones escalated, so the process does not bottleneck on one person.
Feed the same data to the provision
Behavioural deterioration is an input to the credit loss estimate as well, which is the connection made in a provision that recalculates monthly.

Two costs of leaving limits alone
Both are real and only one of them is usually noticed.
Blocked good customers
A growing customer hits a limit set years ago, an order is held, and the commercial relationship suffers over an administrative number nobody reviewed.
Exposure with deteriorating ones
A customer whose payment behaviour has been slipping for a year continues to trade at a limit set when they were healthy, which is where bad debts come from.
How ChatFin supports credit review
Behavioural scoring across a customer book is repetitive analysis over ledger data, prepared for a credit professional to act on.
Credit policy, limit authority and any decision to extend or withdraw credit remain entirely with your team.
AI Cost Estimator for the CFO Office
Four inputs. A first year range, a timeline, and three delivery models compared. No login required.
Frequently Asked Questions
How often should credit limits be reviewed?
Continuously by trigger rather than annually by calendar. An annual review of an entire book allocates equal attention to accounts with very different risk, and it tends to become a formality. Triggers based on behaviour and exposure concentrate the work where something has changed.
Is internal payment behaviour better than an external score?
For an established customer, generally yes, because it is specific to your relationship, your terms and your products, and it is current. External data remains valuable at onboarding and as an independent check, particularly for early warning of events outside your trading relationship.
What is the most useful single signal?
The trend in days beyond terms rather than the level. A customer who has always paid fifteen days late is predictable. One who has moved from two days to fifteen over six months has changed something, and that change is worth understanding before the next large order.
Who should approve limit changes?
Set authority by size so that routine adjustments do not queue behind one person. The point of a trigger based process is to make small, timely adjustments possible, which does not work if every change requires the same level of approval.
You have four years of evidence the original decision lacked.
Scoring the book on your own payment data, defining triggers and setting authority by size replaces an annual formality with something continuous and smaller.
If you want a first year range for behavioural credit analysis on your ERP, the estimator will produce one before any scoping.
Open the estimatorBook a Demo
See what this looks like in your ERP.
We will walk through the workflow behind this article on the system you already run, using your own data, and show what a first phase would actually cover.
This article is general information for finance and accounting teams. It is not legal, tax, accounting or investment advice. Dates and thresholds are given as published at the time of writing and several of these timelines have moved before, so confirm the current position with your advisers and with the relevant authority before acting.