In this article
A credit control policy is a set of decisions made in advance, so that they don't have to be made in the middle of a busy week: the terms you offer, how you'll be paid, who follows up and when, and where you escalate. One page that people actually follow is worth more than ten pages that sit unread.
The test isn't whether the document is thorough. It's whether anyone can tell you what happens on day 8 without opening it.
The short version
- A policy is a set of pre-made decisions, not a statement of intent.
- Decide five things: terms, payment method, who owns follow-up, the schedule, and the escalation point.
- Write the schedule as dates and actions, not principles.
- Put the collection mechanism in the policy. A policy that relies on somebody remembering to chase tends to be followed until the week gets busy.
- Review it against your own numbers, not against how reasonable it sounds.
The short answer
Write down what happens, when, and who does it, for the ordinary case. Then write down where your exceptions get decided and by whom. Best practice is to keep it to a page.
What a policy has to decide
| Decision | The question it answers | The failure if it's missing |
|---|---|---|
| Terms | What do we offer, and to whom? | Terms get negotiated per deal and nobody knows the baseline |
| Payment method | How will the money actually arrive? | Every invoice becomes a fresh request |
| Ownership | Who follows up? | Everyone assumes someone else |
| Schedule | What happens on which day? | Chasing depends on remembering |
| Escalation | At what point does this stop being admin? | Escalation happens in frustration, or never |
A structure that works
Seven things, in this order, will cover the ordinary case on one page.
- Standard terms. Your default payment period, and who can agree something different. If you agree nothing, the statutory default gives you 30 days from the later of performance and the customer having notice of the amount.
- How you get paid. The methods you offer, and which of them you'd prefer. Your policy earns its keep here: for recurring work, say that you capture a direct debit mandate at onboarding instead of treating it as an upgrade.
- What you need before the invoice goes out. The named contact, the approver, and any reference your customer's system needs. All three are cheap to collect at the start and expensive to discover on day 30.
- Your follow-up schedule, written as days and actions. For example: confirmation three days before due, reminder day 1 overdue, second reminder day 8, call day 21, formal notice day 30. The specific numbers matter less than the fact you've written them down.
- Interest and compensation. Whether you charge statutory interest, and from when. The right applies automatically on business debts at base rate plus 8%, currently 11.75%, along with fixed compensation of £40, £70 or £100 by debt size. Decide it in advance and you can use it later.
- Escalation and stopping. Who authorises a formal notice, a collection agency or a claim. And when you write a debt off, because a debt that costs you more to pursue than it recovers is a decision and not a failure.
- Exceptions. Who can vary the policy, and where you record that. A policy with no exception route gets ignored instead of varied.
Making it stick
Automate the schedule, because a policy that depends on someone remembering it tends to slip once the week gets busy. Adfin sends your follow-ups on the schedule you set, over email, WhatsApp and SMS from your own domain, and can apply statutory late fees automatically.
Put the mechanism ahead of the reminder. The measurable difference isn't in the chasing: invoices set to collect automatically are paid 98.0% of the time against 90.8% collected on demand (Adfin platform data, requests created in the seven months to the end of January 2026), and a mandate is usually what's behind the first figure.
Give the policy one owner, not a committee and not "the finance team".
Make your first step a confirmation and not a chase. Checking an invoice is approved before it's due prevents more delay than any reminder you send afterwards.
Then review the policy against your own numbers: your on-time rate, your overdue rate and your debtor days. A policy that reads well and doesn't move those needs changing, not defending.
Why most policies fail
Policies tend to fail in the same handful of ways, and you can check yours against them.
The first three are about what the document says. It describes intentions instead of actions, and "We will chase promptly" isn't a schedule, because it doesn't say when. It's too long to consult, so when somebody needs to know what happens on day 8 they'll guess instead of opening it. And it assumes the invoice is the problem, when most of your delay comes from agreement and routing, both of which happen before the invoice exists.
The other three are about mechanism and maintenance. A policy with no automatic collection and no automated follow-up competes with everything else in your week. A policy nobody reviews carries terms you agreed for a client three years ago, and those often outlive the reason you agreed them. And a policy that treats all your invoices alike misses that a £90 invoice and a £9,000 invoice fail differently: mandate coverage drops from 74.1% of paid invoices under £100 to 37.4% at £2,500 and over (Adfin platform data), so your largest invoices are usually the ones with no collection arrangement behind them. Say in the policy which invoices get a mandate and which get a link.
A policy an agent can run
Read those five decisions again and notice what you've actually written down: standard terms, how payment gets collected, who owns follow-up, the schedule, and the escalation point. That's a configuration, and it's the same list you'd hand a new credit controller on their first morning, as well as the list an agent needs before it can act on your behalf.
Which answers the problem this article opened with. Your policy fails because it depends on people remembering it in a busy week, and a policy that doubles as the rules of engagement for something that executes it doesn't fail that way. The schedule runs, your escalation point becomes a setting instead of a memory, and the review step happens where you asked for one.
The parts that stay with you are the ones your policy already flags as judgement calls: who can vary the terms, when you write a debt off, and what to do about a client you'd rather keep than press. Those were always worth writing down. What changes is that the rest of the document stops being a description of good intentions and becomes something that actually happens.
Common questions
What should a credit control policy include? Five decisions: your standard terms, how payment will be collected, who owns follow-up, the day-by-day schedule, and the escalation point. Plus who can vary it and how a write-off is decided.
How long should a credit control policy be? Short enough to consult. One page for the ordinary case, with the escalation and exception routes named. The longer it gets, the less anyone consults it.
Should the policy say we charge interest? Decide it in advance either way. The statutory right applies to business debts whether the contract mentions it or not, at base rate plus 8%, so the useful decision is whether you'll use it and from what point.
Who should own credit control? One named person. Shared ownership is the most reliable way for follow-up to stop happening.
How often should we review it? Annually, and against your own on-time rate, overdue rate and debtor days rather than against how sensible the document reads.
Does a policy actually change anything? Only where it includes a mechanism. Automatic collection and an automated schedule change what actually happens, and a written intention on its own leaves the week as it was.
Sources
- GOV.UK — charging interest on a commercial debt (accurate as of August 2026)
This article explains how credit control works and is not legal advice. The statutory interest rate moves with the Bank of England base rate, so check the current position before relying on a figure. Last updated August 2026.
