The short answer
Decide the most any 1 client may owe you before you have a client who owes you too much. The number is useless invented in the moment, because in the moment you will be arguing with yourself about a relationship rather than setting a policy.
Work out your true exposure, which is larger than the overdue figure.
Set a ceiling you could lose without it mattering.
Use different ceilings for different kinds of client.
Decide in advance what crossing it triggers.
Review the numbers quarterly, and raise them deliberately rather than by drift.
This is the cheapest control in payment operations. It costs an afternoon, it requires no conversation with anybody, and it converts the hardest judgment in the business into arithmetic you did when you were calm.
Your exposure is bigger than the overdue number
Most agencies look at what is overdue. That is the smallest of 4 numbers and the last to move.
Component | Example | Why it counts |
|---|---|---|
Overdue invoices | $12,000 past the due date | The obvious part |
Issued, not yet due | $9,000 on net 30, day 12 | Already delivered, still unpaid |
Delivered, not yet invoiced | 2 weeks of work since the last stage | Earned, and entirely unprotected |
Costs committed for them | A contractor booked for next month | Money you owe because of this client |
Add the 4 and the figure is usually 2 to 3 times what the aged receivables report shows. The third row is the one that surprises people, because work delivered but not invoiced appears in no system at all and is the easiest thing to lose if the client fails tomorrow.
Run the sum per client rather than in total. The total tells you about your business; the per-client figure tells you where the risk actually sits, which is the distinction at the heart of when your biggest client is your latest payer.
3 ways to set the number
Pick 1 and stay with it. The method matters less than having one you apply consistently.
What you could absorb. The amount you could write off entirely without missing payroll or borrowing. The FDIC's small business guidance defines working capital as the money a business has available to meet its day-to-day expenses, current assets less current liabilities, which is the figure this test is really asking about. For most small agencies it produces a smaller number than instinct suggests.
A share of monthly costs. A ceiling of 1 month of operating costs per client is a common rule of thumb and easy to explain to a team.
A share of annual revenue. Where no client should represent more than a set share of income, the same logic applied to what they can owe at any moment.
Whichever you choose, write the number down somewhere other than your head. A limit nobody can look up is a feeling with a number attached, and it will move when you are busy.
Different clients, different ceilings
A single limit across every client is simpler and wrong. The relevant difference is how likely the money is to arrive, which varies enormously by client type:
Client type | Ceiling | Why |
|---|---|---|
Public body or large corporate | High | Slow but near certain, with statutory deadlines |
Established private company | Moderate | Checkable record, real balance sheet |
New client, any size | Low until invoice 1 clears | No history with you yet |
Another agency | Low | Their cash depends on a client you cannot see |
Startup | Lowest | Risk is existence rather than willingness |
The bottom 2 rows have their own articles, since the reasoning differs: agency clients and startups. The third row is the useful habit. Keep every new client on a low ceiling until 1 invoice has travelled their whole process and been paid, because that single event tells you more than any check beforehand.
What checks are available before signing is covered in how to check a client's payment record, and the answer is more for large companies than for small ones.
What crossing it triggers
A limit with no consequence is a note. Decide the consequence now, in writing, while nothing is at stake:
Stop starting new work on that client until the balance comes down. Finishing what is in progress is usually right; beginning the next stage is not.
Invoice everything deliverable immediately, including work completed but not yet billed, which removes the invisible part of the exposure.
Raise it with the client factually, as a process rather than an accusation. The wording in pausing work applies.
Require the balance cleared before the next phase, which is easier to state when it was agreed in advance than invented in response.
The point of deciding in advance is that crossing the limit stops being a decision. It becomes a thing that happens, which is far easier to say to a client you like.
Reviewing, and raising deliberately
Check the per-client exposure weekly alongside the other numbers in the weekly review, and review the ceilings themselves quarterly.
Raise a limit when the evidence supports it: several invoices paid on time, a growing engagement, a client whose finance process you now understand. Raising it because the work is exciting or the client is charming is how agencies end up with a single client owing them a quarter of annual revenue.
And put the ceiling into the cash flow picture. A limit of $40,000 across 5 clients is $200,000 of theoretical exposure, which is a different business from $40,000 across 1, and the shape of that shows up in a 13-week forecast long before it shows up anywhere else.
Why a number beats a feeling
Every agency already has an informal limit. It is the point at which the founder starts feeling uneasy, and it has 3 problems: it arrives late, it moves with mood, and it cannot be delegated.
A written number fixes all 3. It triggers at a consistent point, anybody can apply it, and nobody has to decide whether this is the week to be worried. It also makes the conversation with the client easier, because a policy is less personal than a judgment about them.
The clients who matter least to this are the ones who pay on time, who will never come close to the ceiling and will never know it exists. The ones who do approach it are precisely the ones you wanted a rule for.
FAQ
How much should I let a client owe me?
No more than you could write off without it mattering. For most small agencies that is a smaller figure than instinct suggests, and 1 month of operating costs per client is a workable starting rule.
What counts towards the limit?
Overdue invoices, invoices issued but not yet due, work delivered but not yet invoiced, and costs you have committed for that client. The last 2 are usually missing from the aged receivables report and often double the real figure.
Should every client have the same limit?
No. A public body with a statutory payment deadline and a 6-month-old startup carry very different risk. Keep new clients on a low ceiling until 1 invoice has been through their full process and been paid.
What do I do when a client goes over the limit?
Stop starting new work, invoice everything deliverable including unbilled work, and raise it as process rather than accusation. Deciding the consequence in advance is what makes it easy to say.
How often should I review the limits?
Check exposure weekly with the rest of the numbers and review the ceilings quarterly. Raise them on evidence such as a record of prompt payment, rather than on enthusiasm for the work.
Is this worth doing for a small agency?
It is most worth doing for a small agency, because a single bad debt is a larger share of the year. It costs an afternoon once and then runs itself.
Not financial advice. Credit policy depends on your own cost base, client mix and risk tolerance. Discuss the numbers with your accountant before relying on them.