Most agency owners keep difficult clients too long. The revenue is visible on the invoice and the cost is spread across a dozen places that never get totaled — your coordinator's morale, your Sunday evening, the new business you did not pursue because you were handling a Saturday complaint.
Here is how to recognize the ones worth ending, and how to end them well.
The signals that actually predict it
Their hours are double the account average. This is the most reliable single indicator, and most agencies never measure it. Track hours per client for one month. Any client at more than 1.5x your average for the same package is either mispriced or misaligned.
Approval is chronically late and always your fault. Some clients are slow. That is manageable with default-approve policies. The problem client is slow and treats the resulting delay as your failure. That combination does not improve.
Scope requests arrive as assumptions. The client who asks whether something is in scope is fine. The client who asks where the thing is, having never discussed it, has a different model of the relationship than you do. If two conversations do not correct it, it will not correct.
Your team dreads them. Ask your coordinator which client they would drop. Their answer is usually right and usually earlier than yours, because they absorb the daily friction while you see the monthly invoice.
They will not supply anything. No photos, no product information, no responses to questions, and then dissatisfaction that the content feels generic. You cannot make specific content from nothing, and this becomes a permanent grievance.
Payment is chronically late. Once is life. Every month is a statement about how they value the work.
They contradict themselves across rounds. Approve, publish, then complain about the thing they approved. If there is no consistent standard to hit, you cannot hit it, and every month will contain a version of this.
Do the arithmetic before deciding
Before firing anyone, calculate what they actually contribute.
Monthly revenue, minus hours multiplied by your blended cost, minus a fair share of overhead. Then note what percentage of your total agency hours they consume.
The common finding is a client at 8 percent of revenue consuming 20 percent of hours. Losing them costs $1,400 a month and returns capacity worth roughly $3,500 in new business you have not had time to sell.
Sometimes the arithmetic says keep them. A demanding client at 20 percent of revenue with acceptable margin might be worth managing rather than releasing, at least until you have replaced the revenue.
Try repair first, once, explicitly
Most difficult clients are fixable, and the fix is usually a conversation nobody has had.
Book thirty minutes. Say what is not working in specific, non-accusatory terms: "We are averaging eleven days to approval, which pushes everything else back. What would make that easier on your side?"
Then propose the change concretely — a default-approve deadline, a single named approver, a monthly asset drop, a scope document you both sign.
Give it sixty days. If it holds, you saved a client. If it does not, you have a clean record of having tried, which makes the exit conversation straightforward for both of you.
How to exit properly
Give real notice. Thirty days minimum, sixty if they are large. This is about your reputation, and this industry is smaller than it appears.
Do not itemize grievances. "We are not the right fit for what you need going forward" is complete. A list of complaints invites a rebuttal and turns an ending into an argument.
Deliver everything owed. Finish the notice period at full quality. The temptation to coast is real and it is exactly what generates a bad review.
Hand over cleanly. Remove your access, confirm they retain admin on all their accounts, export their content archive, and send a short document on what was scheduled and where things stand. A client who leaves with a clean handoff sometimes refers people anyway.
Refer them somewhere honest. If you know an agency that would genuinely suit them better — more hands-on, different specialty, different price point — say so. This costs you nothing and often converts a bad ending into a neutral one.
What to do with the capacity
The failure mode after firing a client is refilling the slot immediately with whoever is available, usually a similar client.
Use the space deliberately. Write the case study you have been meaning to write. Fix the onboarding process. Raise prices on the clients you kept. Go get one better client instead of one replacement client.
The capacity is the return on the decision. Spending it on more of the same means you did the hard part and kept none of the benefit.
The signal you are waiting too long
If you have read this and thought of a specific client three times, that is the answer. Agency owners do not deliberate about clients who are working.