Most consultants don't fire a bad client too early — they fire one too late, months after the relationship stopped making sense, because the decision felt too personal to make on principle alone. Firing a client isn't a reaction to a bad week. It's a decision that should follow from a small number of concrete, repeated conditions — not from how frustrated you feel on any given Tuesday.
It's a pattern question, not a moment question
A single missed boundary, one unreasonable request, or one rough call is a Tuesday, not a referendum on the relationship. The decision to fire a client should be based on whether a problem has repeated after you've already tried to fix it — not on the emotional weight of the most recent incident. If you haven't yet had the direct conversation about the issue, you don't have enough information to fire the client. You have a scope conversation to have first.
Criterion 1: Chronic scope-boundary violations
Every engagement has an occasional scope slip — that's normal and not a red flag on its own. The red flag is a client who, after you've explicitly restated the boundary, keeps pushing past it anyway: emailing outside agreed hours and expecting same-day replies, adding "quick asks" that aren't quick, or treating a defined monthly scope as a starting point for negotiation every single month. One restated boundary that gets crossed again is the signal — not the first crossing, the second one after you named it.
Criterion 2: Below-profitable rate despite a real renegotiation attempt
If you've already run the profit-per-hour numbers on your client list — the framework in Which of Your Clients Are Actually Profitable? covers how — you may already know a client is underwater. That alone isn't a firing decision. The firing decision comes after you've had the renegotiation conversation — a scope cut or a rate increase — and the client either declined it or agreed to it and then quietly drifted back to the old behavior within a month or two. A client who won't accept a fair, well-argued fix to an unprofitable arrangement has told you the arrangement isn't going to improve on its own.
Criterion 3: The relationship is actively damaging your other work
This is the criterion consultants most often underweight, because it doesn't show up on an invoice. A client who puts you in a defensive, anxious state before every call, who you find yourself avoiding or delaying, or who's costing you sleep or focus on days you're not even working with them — that cost is real even though nothing prices it. If a client's effect on you is measurably degrading the quality of work you deliver to other clients, the relationship has moved from "difficult but worth it" to actively costing you more than the invoice covers.
A consultant bills a client $2,400/month for 12 hours of defined scope — a fine rate on paper. Over two months, the client repeatedly messages outside the agreed hours expecting immediate replies, despite two direct conversations resetting the boundary. The consultant starts checking email at 9pm out of dread and shows up distracted to a call with their best client the next morning. The $2,400 hasn't changed. What it's actually costing — in a damaged relationship with a better client and in hours the consultant can't bill anyone for — has. That's all three criteria stacking, and it's a clear signal, not a borderline case.
Two out of three is a strong signal. One is a conversation.
None of these three criteria on its own, in isolation, automatically means fire the client — a single scope violation gets a boundary reset, a single low-margin month gets watched, a single bad week gets shrugged off. But a client who's hit two of the three, after you've already tried the direct fix for each, isn't a maybe. The data and the direct conversation have both already happened; what's left is deciding whether to keep re-running them indefinitely.
If you haven't already run the profitability numbers that usually surface criterion two, start there — the free Client Profitability Calculator turns fee and hours into a profit-per-hour figure in seconds, so the decision is grounded in a number instead of a feeling.