On this page
Handling a difficult client is a five step process: name the actual behaviour rather than the personality, decide whether it is a scope problem or a respect problem, respond in writing within one working day, put the fix and its cost in the same message, and set a review point. Most difficult clients are not difficult people. They are people reacting to an unclear scope, an unanswered message, or a decision made without them, and all three are things you control. The scripts below are the exact wording for the six situations that actually come up: scope creep, late payment, endless revisions, a hostile message, going quiet, and ending the relationship. The one case worth ending early is disrespect, because unlike scope, it does not respond to a process fix.
Most advice on this topic sorts clients into personality types and tells you to stay calm. That is not much use at nine in the morning when a message has arrived that starts "I have to say I am extremely disappointed" and you have to reply to it.
So this is organised around what to do and what to say, not around what kind of person they are.
The Five Step Process
- Name the behaviour, not the person. "Difficult" is not actionable. "Asks for changes after sign-off" is. Write down the specific thing that happened. If you cannot describe it in one sentence, you are reacting to a tone rather than a problem.
- Decide which of the two problems it is. A scope problem means they want something the agreement did not cover. A respect problem means how they are treating you. Scope problems are fixable with a process. Respect problems are not, and the rest of the process changes depending on the answer.
- Respond in writing within one working day. Not immediately, because the first draft is always worse, and not next week, because silence is what turns an annoyed client into a hostile one. A few hours to draft it properly is the right gap.
- Put the fix and its cost in the same message. Never separate them. "Yes, I can do that" followed two days later by an invoice reads as a trap. "Yes, I can do that, it adds two days and $400, shall I proceed" is a normal business exchange.
- Set a review point. End with a specific next checkpoint and a date. Difficult situations get worse in the gaps, and a named date closes the gap.
Steps three and four are where most of the damage is done or avoided. The delay and the separated invoice are what convert a solvable disagreement into a dispute.
Six Situations, and What to Say
These are the six that actually recur. Each script assumes you have already done steps one and two, so you know what happened and which kind of problem it is. Adjust the wording to sound like you; keep the structure, because the structure is what does the work.
1. They keep adding things that were not in the scope
This is the most common one and the least personal. It is usually not an attempt to get free work. It is that the boundary was never visible to them.
Happy to take that on. It sits outside what we agreed in the original scope, so it would add [TIME] and [AMOUNT] to the project. If you would like to go ahead, say yes here and I will send a short amendment and get started. If you would rather keep the current budget, we can park it for a second phase.
Three things are doing the work there. It says yes rather than no. It names the cost in the same breath as the agreement. And it offers a second option that is not a refusal, which is what keeps the relationship intact.
2. The revisions never end
Endless revision is almost always a symptom of an undefined "finished". If the agreement never said how many rounds, then from their side there is no reason to stop asking.
We are now at round [N]. The original scope covered [N] rounds of revisions, which we have used. I want to get this right, so here are two options. I can make this round of changes as a final pass and we close the project, or we move to an hourly rate of [AMOUNT] for further changes. Which would you prefer?
If the agreement genuinely did not specify a number, say so plainly and set one going forward rather than pretending it was there. Inventing a limit retrospectively is how a scope problem turns into a trust problem.
3. The invoice is late
Chase earlier than feels comfortable. The uncomfortable feeling is yours, not theirs, and an unpaid invoice does not improve with age.
Invoice [NUMBER] for [AMOUNT] was due on [DATE] and is now [N] days overdue. Could you confirm when it will be paid, or let me know if there is a problem with it that I can sort out? I am holding further work on [PROJECT] until it clears.
The last sentence is the one that matters and the one people leave out. A chaser with no consequence is a reminder to ignore. Only include it if you will actually do it.
4. The message is hostile
The instinct is to answer every accusation in order. Do not. A long defensive reply reads as an argument and gives them more surface to attack.
Thanks for being direct, I would rather know. Taking the points in turn: [POINT ONE] is fair and here is what I am doing about it. On [POINT TWO], my understanding was [FACT], from [WHERE IT WAS AGREED]. Can we get fifteen minutes on a call [DAY] to settle the rest? I would rather sort it than trade emails.
Concede the fair point first, and concede it fully. It costs you nothing and it removes most of the heat. Then correct the factual one by pointing at where it was agreed rather than at their memory. Then move to a call, because hostility survives in writing and rarely survives a voice.
5. They have gone silent and the project is stuck
A client who stops replying is a difficult client, even though it feels like the opposite. The project ages, your other work gets rescheduled around a ghost, and the eventual restart is always urgent.
Checking in on [PROJECT]. I need [THE SPECIFIC THING] to move to the next stage. If I do not hear back by [DATE], I will pause the project and pick it back up when you are ready, though I will need to reschedule it against other work at that point. No problem either way, I just do not want it sitting half finished.
Name one thing you need, not four. A list is easy to postpone; a single item is easy to answer.
6. It is time to end it
Do this in writing, without a list of grievances. The temptation to explain everything they did wrong is strong and serves nothing, because you are not trying to win the argument, you are trying to leave.
I have been thinking about how [PROJECT] is going, and I do not think I am the right fit for what you need. I would like to finish [DEFINED CHUNK] by [DATE] and close things there. I will send the final invoice for work completed to that point, and hand over [FILES AND ASSETS] so you can pass it to someone else easily. I am sorry it did not work out.
Finish something and hand over cleanly. A client you exited well says nothing about you; a client you abandoned mid-project talks.
Aziz's take: The single line that has saved me the most trouble is putting the price in the same message as the yes. When you say "sure, no problem" and invoice for it later, you have not been generous, you have set a trap that springs two weeks on. Naming the cost in the same sentence feels awkward for about four seconds and then it is just how you do business, and clients respect it more, not less.
What a Difficult Client Actually Costs You
Founders tolerate bad clients far longer than the numbers justify, usually because the revenue is visible and the cost is not. Work it out once and the decision gets much easier.
Take a client paying $4,000 for a project you scoped at 40 hours, which is $100 an hour. Scope creep adds 8 unbilled hours, the extra back and forth adds 5, and one round of unplanned revisions adds 6. That is 59 hours for $4,000, or $68 an hour. The client did not become 32% less valuable, but your rate did.
The part that does not show up in that arithmetic is the displacement. The hours went somewhere, and it is normally the work that has no deadline: the marketing, the case study, the follow up with a better prospect. A difficult client is rarely expensive because of the hours themselves. It is expensive because those hours come out of the only time you had for finding the next client. This is the same mechanism that makes scaling a service business so difficult: the constraint is never the work, it is what the work crowds out.
Run the same calculation on your own last difficult project before deciding whether to keep the next one. If the effective rate has dropped below what you would accept from a new enquiry, you have your answer.
Preventing Most of Them Before They Start
Four of the six situations above are preventable at the start of the relationship, not in the middle of it.
Write the scope as exclusions, not just inclusions. Most scopes list what is included, which quietly implies everything else might be too. A short "not included" list prevents more disputes than a long "included" one.
Put a number on revision rounds. Two is normal. The number matters less than its existence, because "finished" needs a definition that is not a feeling.
Take a deposit. A client who will not pay anything up front is telling you something before the work starts. This is also the cheapest filter available to you, and it happens at onboarding, before you have invested anything.
Set the response expectation both ways. Say when you reply and when you need replies. Most "unresponsive client" problems are actually undefined-response-time problems, and staging the payments against defined milestones gives both sides a reason to keep moving.
When to Stop Trying
The five step process is worth running once, sometimes twice. There are four situations where it is not.
Disrespect that continues after you have named it. Everyone has a bad week. Once you have raised it directly and it happens again, you are not dealing with stress, you are dealing with how this person treats people they pay.
Asking you to do something dishonest. Not a negotiation.
Repeated late payment after a clear conversation. Once is a cashflow problem. Three times is a policy, and it is theirs.
The effective hourly rate has collapsed and will not recover. If the work left to do costs more than the money left to collect, finishing is the expensive option.
Everything else is worth one honest attempt at fixing. Most difficult clients are recoverable, and a client who was difficult and then was handled well is often more loyal afterwards than one who never had a problem, because they have seen what happens when something goes wrong.
Aziz's take: I used to think ending a client relationship needed a justification they would accept. It does not. "I do not think I am the right fit" is a complete sentence and nobody has ever successfully argued me out of it. The long explanation is for your own comfort, and it usually just gives them a list of things to dispute.
The Checklist Before You Reply
- Can I describe what happened in one sentence, without adjectives about them?
- Is this a scope problem or a respect problem?
- Have I checked what the agreement actually says, rather than what I remember?
- Has enough time passed that I am not writing this angry?
- Does my reply contain both the fix and its cost?
- Have I conceded the one point where they are right?
- Is there a specific date in the message?
- Would I be comfortable if this email were read out later?
- Am I proposing a call for anything that has gone past two rounds of email?
- If this is the third incident, am I fixing a process or avoiding a decision?
Point eight is the useful filter. Almost everything that fails it is a sentence written to feel good rather than to solve anything, and it can be deleted without loss.