Skip to content
allanninal.dev/build
Collaborate

Part 4 of 7 · Credit limit reviewer series ~6 min read

What makes a limit worth revisiting

An annual review means that for eleven months of the year, nothing is watching, and companies do not fail on the anniversary of their credit application.

Key takeaways

  • Days beyond terms, trended, is the strongest signal you own.
  • Review on events: a drift, a large order, a filing, an ownership change.
  • A limit that is never approached should come down, quietly.
  • Reductions are a conversation, not an email at 2am.
  • The trigger list is short and it should stay short.

Why the annual review fails

Average days to pay across the year before a customer failureA bar chart with four bars showing average days to pay. Months one to nine: twenty-nine days. Month ten: thirty-eight days. Month eleven: fifty-two days. Month twelve: seventy-one days. A note says the review was scheduled for month twelve, by which time the question had answered itself.0255075100~29Month 1-9~38Month 10~52Month 11~71Month 12Average days to payThe review was scheduled for month 12. By then the question had answered itself.
Fig 1. One customer’s payment behaviour in the year before they failed. Every bar after the first is a trigger; the calendar review arrives after the last one.

Nothing in that chart is subtle, and yet it is invisible to a business that looks at an account once a year and otherwise only notices whether today’s payment arrived. The individual invoices were all eventually paid, which is exactly why nobody escalated.

Days beyond terms is the metric

Not whether an invoice is overdue today, which is noise, but the trend in how long the customer takes relative to the terms they agreed. A customer on thirty-day terms paying consistently at thirty-four is fine. The same customer moving from thirty-four to forty-eight over a quarter is the signal.

It is also a metric you own outright. It needs no subscription, it is computed from your own ledger, and it describes how the customer treats you specifically rather than how they treat their largest supplier.

The trigger list

Three triggers that queue a credit limit reviewThree boxes stacked on the left. Behaviour drift, days beyond terms up ten days over a quarter, labelled the important one. Unusual order, three times their normal, labelled size. And Public record: late filing, charge or director change, labelled external. All three converge on A review task with the reason attached, which leads down to Raise, hold or reduce, where a person decides. A note says three triggers deliberately, because a longer list produces a queue nobody works.Behaviour driftDBT up 10 daysover a quarterthe important oneUnusual order3x their normalsizePublic recordlate filing, charge,director changeexternalA review taskwith the reasonattachedRaise, hold, reducea person decidesThree triggers, deliberately. A longer list produces a queue nobody works.
Fig 2. What causes a limit to be looked at again. The review arrives with its reason already stated, which is the difference between a task and a notification.
  • Analytics
  • Front-end & mobile
  • People

Keep the list short

It is tempting to add triggers: a returned direct debit, a change of address, a new email domain, a sudden gap in ordering. Each is individually defensible and together they produce forty tasks a week, which produces a queue that gets closed in bulk on a Friday afternoon without being read.

Three triggers producing two or three real reviews a month is a system somebody actually works. That is worth more than a comprehensive one that gets ignored.

Reviewing downwards

Three honest outcomes, and how each is delivered

  • Raise. Evidence-based, on request or on a large order, and the easiest conversation you will ever have.
  • Hold. The most common outcome, and it needs no conversation at all — but the reason still gets recorded.
  • Reduce. Never by email, never automatically, and never as a surprise on the next order. Somebody rings them.
  • The unused limit. A customer with a twenty thousand limit who has never exceeded two thousand should come down at the next natural moment. It costs nothing to them and removes a risk you were carrying for no return.
  • The recorded reason is the deliverable in all four cases. The number without it is unusable at the next review.

Reducing a limit is the only genuinely difficult action in credit control, which is why most businesses never do it and simply carry limits that were set years ago against trading volumes that no longer exist.

A quiet reduction of an unused limit is almost never noticed and almost never disputed. A reduction on a customer who is actively trading is a phone call, made before the system starts refusing things, by somebody who can explain it.

What automation must not do here

It must not send the reduction. A limit cut arriving as an automated email at two in the morning turns a manageable conversation into a relationship ending, and it will reach the customer’s whole team before it reaches yours.

The system’s job is to notice, to assemble the evidence, and to put a task in front of a human with everything they need to make the call in four minutes rather than forty.

Next: getting the answer to the person taking the order.

All posts