Selling on Credit in Retail: Accounts, Limits & Getting Paid
Every Kenyan retailer ends up selling on credit, usually by accident and usually to the customers who ask most confidently. What a credit limit is actually for, why exposure is not the same as the invoice in front of you, and the collection rhythm that keeps a good customer good.
Nobody decides to become a lender. It happens one favour at a time — a good customer is short this week, a contractor needs materials before their client pays, a school buys on an LPO with payment "at the end of term". Each one is reasonable. Two years later a retailer with thin margins is financing a substantial part of their customers' working capital and cannot say precisely how much.
The uncomfortable arithmetic: at a 15% gross margin, one unpaid 100,000 shilling account requires roughly 667,000 shillings of additional sales to recover. Credit losses in retail are not proportional to their size — they are proportional to your margin, which is why a business that could absorb a bad debt at 40% margin cannot at 15%.
A limit is a decision made in advance
The purpose of a credit limit is not to insult customers. It is to move the decision from the counter — where a supervisor is under social pressure, in front of a queue, with an insistent buyer — to a calm conversation held earlier, by someone with the information.
In AWRA OpsHub a customer carries a credit limit, and an invoice can be evaluated against their exposure rather than against the invoice alone. That distinction is the whole point: what matters is not whether this 80,000 invoice is large, but whether this customer already owes 240,000 against a 250,000 limit.
Why the invoice in front of you is the wrong number
Illustrative, in KES. Judged on its own the 80,000 order is unremarkable and would be approved by anybody. Judged against exposure it breaks a limit the business agreed for a reason. Nobody at a counter can hold the first three lines in their head — which is precisely why the check has to be a system behaviour rather than a diligence expectation.
The order at the counter is never the number that matters. What matters is what this customer already owes — and that is the one figure the person deciding cannot see.
Set limits from behaviour, not from hope
-
Start every account at zero
New customers pay on delivery until they have a payment history with you. Trade references describe how someone pays other people; your own record describes how they pay you, and only the second one predicts anything.
-
Set the first limit at what you could afford to lose
Not at what they ask for. The first limit is an experiment, and an experiment should be sized so that a bad result is survivable rather than memorable.
-
Raise it on evidence, in steps
Three or four cycles of paying on time earns an increase. This is also the moment to say the terms out loud again, because a limit raised silently is heard as a limit removed.
-
Reduce it the moment behaviour changes
The hardest step and the one that saves the money. Deteriorating payment behaviour precedes a default by months, and by the time an account has stopped paying entirely the exposure is already at its maximum.
Terms and limit are two separate promises
A limit caps how much they can owe. Terms say how long they may take. A customer can be well inside their limit and 90 days late, or perfectly punctual and over-exposed. Track both, because they fail independently and each hides the other.
The collection rhythm
Collections in a small retail business are almost always reactive: nothing happens until an account is old enough to hurt, and then it is a difficult phone call from an owner who is now annoyed. The alternative is unglamorous and works considerably better, mainly because it is early and unemotional.
| When | What happens | Tone |
|---|---|---|
| 3 days before due | A short reminder that payment falls due | Administrative — most late payment is simply forgetting |
| On the due date | A statement showing what is outstanding | Factual, no comment |
| 7 days late | A phone call, not a message | Asking whether there is a problem, which is a genuinely useful question |
| 21 days late | New orders require the account to be brought current | Firm, and applied consistently or not at all |
| 45 days late | Escalate to whoever owns the relationship | A commercial conversation about whether to continue |
The 21-day step is where most retailers fail, and it fails for the same reason every time: it is applied to small customers and waived for large ones. The large ones are precisely where the exposure is. A rule applied inconsistently teaches customers which category they are in, and they will act accordingly. The full receivables treatment is in receivables and collections in Kenya.
Credit and the counter
One practical boundary worth understanding early: the credit evaluation described here belongs to the invoiced sales flow rather than to the till. A till sale is a completed transaction with a tender behind it — cash, card, mobile money — not an account posting.
So a retailer running both walk-in and account trade should route credit customers through invoiced sales rather than trying to run an account through the counter. In practice most already do this, because account customers order rather than shop. What matters is that the split is deliberate, so that account exposure is visible in one place instead of half at the till and half in the ledger.
What we do and do not do
What AWRA OpsHub does today
- A credit limit per customer, held on the customer record.
- Exposure evaluated against the limit when an invoice is raised — outstanding balance plus the new invoice, not the invoice alone.
- A hold indication with the reason and the amount over, so the decision reaching a person is specific rather than a vague warning.
- Receivables ageing, so late accounts are visible in bands rather than as one total.
- Customer statements, so a reminder is a document rather than an assertion.
- Payment history per customer, which is the evidence a limit increase should rest on.
What it does not do
- No credit scoring. Nothing predicts whether a customer will pay; the limit is your judgement from their history with you.
- No automatic dunning sequence. The reminder rhythm above is a process you run, not a workflow that runs itself.
- No credit control at the till. The evaluation applies to invoiced sales; a POS transaction settles with a tender.
- No interest or late-payment charges. If your terms include them, they are raised as a separate charge, not computed for you.
The absence of automatic dunning is the one to plan around. Everything the rhythm needs — ageing, statements, history — exists; the discipline of who sends what on which day belongs to a named person.
Our take
Give every account a limit sized to what you could afford to lose, judge orders on exposure rather than on the invoice in front of you, and apply the 21-day rule to your biggest customer first. In a 15%-margin business, one recovered bad debt is worth more than a very good month.
See customer 360 and credit
Credit limits with exposure evaluated at invoicing, receivables ageing, statements and full payment history per customer.
Explore customer recordsFrequently asked questions
How do we set a credit limit for a new customer?
Start at zero and sell on delivery until there is a payment history with you. When you do extend credit, set the first limit at an amount you could afford to lose rather than at what the customer asks for — the first limit is an experiment and should be sized so a bad outcome is survivable. Trade references tell you how someone pays other people, which is far less predictive than how they pay you.
Does the system block a sale to a customer over their limit?
It evaluates the invoice against the customer's exposure — what they already owe plus this invoice — and indicates that a hold is required, with the reason and the amount over. That puts a specific fact in front of the person deciding rather than a vague warning. The commercial decision to proceed anyway remains a human one, which is right: sometimes you extend credit to a customer over their limit, and that should be a deliberate, visible choice.
Can we take credit sales at the till?
Credit evaluation belongs to the invoiced sales flow rather than the point of sale, where a transaction settles with a tender. In practice this suits most retailers, because account customers order rather than shop. What matters is making the split deliberate — running some account trade through the counter and some through invoicing means your total exposure to a customer is not visible in one place.
What is the most effective collection step?
The reminder three days before the due date, because most late payment is administrative rather than financial — an invoice sitting unapproved on somebody's desk. It is also the cheapest and least awkward contact you will ever make. After that, the step that actually changes behaviour is requiring an overdue account to be brought current before new orders, applied consistently. Waiving it for large customers is how retailers lose real money, since large customers are where the exposure is.
Should we charge interest on overdue accounts?
Commercially that is your decision and it depends on your market and your contracts. Practically, note that nothing computes interest or late-payment charges for you — such a charge would be raised as a separate item. In most Kenyan retail relationships the credible sanction is not interest, which customers rarely pay, but the withholding of further supply, which they notice immediately.
