The engagements that damage a solo practice are rarely the ones that go wrong loudly. They are the ones that stay comfortable for a year while the practice quietly stops growing.
A dispatch on the least-written-about skill in independent practice: getting out.
TL;DR — the answer first: there are three exits and they are different jobs. The pre-exit is declining a bad fit on the first call, and it turns on four signals — no influence over the outcome, wanting reassurance rather than change, no authority to decide, and money the person does not have. The term exit is ending a live engagement, and it should happen on the renewal date, using the two-term test: if nothing is operating in their business that was not operating on day one, across two consecutive terms, the engagement is not working no matter how good the calls feel. The emergency exit is for dishonesty, unsafety, or a broken agreement — end it in writing, refund the unused term, do not negotiate. In all three, the rule is the same: exit on a date, not on a feeling, and refund the unused portion whenever you are the one leaving.
Everything written for independent experts points one direction: get found, get the call, close it, structure it, keep them. The literature assumes every engagement should continue and treats an ending as a failure of retention.
That assumption is what does the damage. A practice with one or two working slots does not fail because a client leaves. It fails because a slot stays occupied for four terms by an engagement that is pleasant, well-delivered, and going nowhere — while the person who would have used you properly never got in the door. The cost is invisible on your invoice and it is the largest number in your business.
It is worse for the senior end of the market than the junior end. When you are twenty-five, a client who is not working is obviously a problem. When you are fifty-two and the client is a decent person who says the sessions are valuable, ending it feels like something closer to abandonment. So it gets renewed, and renewed again, and the practice you meant to build stays theoretical.
This dispatch assumes a specific shape of person, so check whether it is yours before spending the time.
You have fifteen to thirty years inside one operating domain — running clinics, running plants, running a book of business, running engineering teams, running a trade — and you are now somewhere in your first two years of charging for that judgment directly. Your income is mixed: some salary or hourly project work, plus one, two, maybe three mentoring or fractional arrangements that arrived through people who already knew you rather than through anything you built. On AI you are between "I use it for email" and "I have a couple of workflows running." And the thing in front of you is not finding clients and not pricing — it is that one of your current arrangements is quietly not working, you have known it for a while, and you have no script for saying so to someone you like. If that is close, the rest of this is the script.
If your engagement has never actually been given a shape — no term, no cadence, no defined outcome — read how to structure a mentorship engagement first. Half of what looks like a bad-fit client is an undesigned engagement, and you cannot fairly end something you never defined.
The cheapest exit is the one that happens before there is anything to exit. Four signals, all visible in the first conversation if you ask directly:
None of these four improve after the deposit clears. Declining costs you one fee. Accepting costs you a term of your only real inventory — attention — plus a finished engagement you cannot point anyone at.
Say it straight: "I do not think I am the right person for this, and here is why." Then, when you can, name someone who is. A well-handled decline is one of the few things that reliably produces a referral, because it demonstrates the one thing prospects cannot otherwise verify — that you will tell them something against your own interest.
The written exercise the book gives for this is deliberately short. From Chapter Four of The Mentor Economy, in the fifteen-minute action at the end of the chapter:
"Take a fresh page. Title it Your Core. Underneath, write three sentences. One: The three values you will not violate to win a client. The lines you would walk away from money to protect."
Read that as an operating instrument rather than a sentiment. Values written before the offer is on the table are a decision procedure; values recalled after it are a negotiation with yourself, and you will lose. Three sentences on a page beside your desk is the entire mechanism — the point is that the lines are decided in a quiet week rather than in the call where someone is offering you the largest fee you have been offered.
For a live engagement, the question is not "is this enjoyable." It is structural, and it has two parts:
Is anything running that was not running on day one? An intake process, a pricing model, a hiring rubric, a weekly review that happens without you. If you cannot name the object after a full term, the engagement is producing clarity rather than change.
Do commitments get done? Track the one thing they agreed to at the end of each session. Done, not done, or done differently. If the ratio is below about half across a term, the sessions are being enjoyed rather than used.
One term failing both is a signal to redesign — smaller commitments, tighter cadence, a single outcome instead of three. Two consecutive terms failing both is the answer. That is not a slow start; that is the diagnostic, and continuing past it is a decision to be paid for something other than the work.
Two adjacent cases resolve the same way. The outgrown client — work you now do without thinking — should be renewed at a scope that uses your judgment rather than your hours, or handed to someone you would vouch for. And the expanding client, the one whose requests keep growing without the fee moving, is not an exit case at all; that is the leak The Scope You Never Wrote Down covers, and it is repriced, not ended.
Say it live, in a session, never in an email that arrives without warning. Keep it under five minutes and give it four parts:
Then stop talking. The instinct is to keep explaining until the other person seems fine, and that is how a clean exit turns into a renegotiation. If you are ending it, refund the unused portion of any prepaid term without being asked — the sum is nearly always smaller than the cost of the conversation about it.
The exit itself is not a place for automation. But the reason exits get delayed usually is.
People stay in engagements they have outgrown because leaving one means having none, and rebuilding demand from a standing start is slow. The fix is upstream: keep the surface that produces inbound conversations running continuously, so that the renewal date is a choice rather than a cliff. That is the case for a practice that renews on purpose — and the argument for keeping your visibility work automated even in the quarters when you are full.
What AI should not touch: the decision, the call, and the words. A summary drafted by a model is fine. An ending delivered by one is a thing people remember about you for a decade.
Ending an engagement with someone who likes you and says you are helping feels like a small betrayal, and no framing entirely removes that. What it does not feel like — from inside the call — is the alternative: the client who is not changing knows it too, usually earlier than you do, and is often relieved when someone finally says it out loud.
Every renewal is a decision, including the ones made by not deciding. A practice is defined at least as much by the engagements it declines and ends as by the ones it wins, because those are the only two ways a working slot ever opens.
The exercise quoted above — writing the three lines you will not cross to win a client, before anyone offers you money to cross them — is the fifteen-minute action at the end of Chapter Four of The Mentor Economy. Get your copy →
End it on a date rather than on a feeling. Say it in a live session, not in writing: name the term you agreed, say the work is complete or is not the right next step, state what they are leaving with, and offer one concrete handover — a written summary, an introduction, or a single follow-up call in sixty days. Do not offer a discount to soften it and do not invent a reason outside yourself. The exit is short, warm, and specific, and it should take under five minutes of the session.
Turn it down when the person wants an outcome you cannot influence, when they want reassurance rather than change, when they cannot make the decisions the work requires, or when they are paying from money they do not have. Those four are visible on the first call if you ask about them, and none of them improve once the engagement starts. Declining costs you one fee. Taking it costs you a term of your capacity and a result you cannot show anyone.
Usually yes, if you are inside a fixed term and the person is holding up their side. The term is a promise, and honoring it is cheaper than the reputational cost of an unfinished engagement in a small professional world. The exception is when the arrangement is unsafe, dishonest, or the client has broken the agreement themselves — then end it in writing, refund the unused portion of the term, and do not negotiate.
Refund the unused portion of a prepaid term whenever you are the one ending it. It is unambiguous, it removes the argument entirely, and the amount is almost always smaller than the time you would spend defending the alternative. If the client is the one leaving mid-term and you have held up your side, the term stands — which is exactly why the term should be written down before the first session.
Check whether anything is running in their business that was not running on day one, and whether last session’s commitment gets done more often than not. Two consecutive terms of enjoyable sessions with no operating change is not a slow start; it is the diagnostic. Pleasant and productive feel identical from inside the call, which is why the check has to be structural rather than emotional.
Then do not end it today — but stop treating the concentration as normal. One client at more than about half your revenue is a client who sets your terms whether or not either of you says so. Finish the current term, put the equivalent hours into replacement demand before you renew, and use the renewal date as the decision point. The exit gets easy the moment it is no longer the only option.
Not at the same shape. Work you can do without thinking is work someone earlier in the ladder should be doing, and holding it keeps the slot closed for the engagement you actually want. Offer to renew at a scope that uses your judgment rather than your hours, or hand it over cleanly to someone you would vouch for. Both are better than a term you resent.
Author of The Mentor Economy and co-founder of MentorMe. He writes about turning hard-won expertise into AI-leveraged one-person businesses.
The Mentor Economy is the full system — free, you just cover $9.95 shipping.
Get Your Free Copy