Dana spent that night looking for the message where it turned. There wasn’t one.
She read the whole thread twice, October through February. Every message was warm. Every message answered the question she’d asked, and answered it inside a day. The January one was five words shorter than the October one, and that was the only real difference between them. Nobody notices five words at the time.
She’d spent a year learning how to reach into a silence. She had a reminder for the stranger who goes quiet before a first call. She had a message for the two weeks after a client signs and hears nothing. She had a check-in somebody could answer even if they’d done none of the work. Wes was never silent once. For six sessions running he produced exactly the behavior all that machinery was built to produce. Then he took another job in January and mentioned it in the middle of February.
This is the version of losing a client that none of it catches. It looks like an easy one, all the way to the end.
Every answer was a yes
Wes did the work. Hold onto that. It’s what makes this hard to see, and it’s what makes it a different problem from the last one.
By the fall, Dana’s system asked one question about every gap between sessions. Did the thing they agreed to happen. It’s a good question and it caught a great deal. Marisol not speaking to Ray was exactly the failure it was built for, and the commitment that loses to the rest of the week works that whole problem through. Wes produced not a single instance of it. Six sessions, six commitments, six completions. Each one reported inside a day.
One thing sat outside that question: whether any of it still pointed at something he cared about.
Here’s the definition, and it’s worth being exact, because the word gets used loosely. A drifting client is one where the engagement is still running and the client’s stake in it has ended. They attend. They comply. They’re pleasant. And everything that happens in the hour sits apart from what they’re actually deciding about their life.
Three things keep that invisible from your side. All three are built into the shape of the work.
Every instrument you have measures compliance, and a drifting client is unusually compliant. They agree with you. They leave the session where it sits on the calendar. The difficult message at ten at night stops arriving. In the plainest sense they’re easy, and Dana called Wes the easiest client she’d ever had, which she meant as praise. Difficulty is a form of participation. Its absence reads as good manners.
Everybody is telling the truth. That’s what separates this from most of the things a coach gets warned about. Wes told the truth every time. He had done the thing. The missing information was never withheld from Dana. It simply had nowhere to go, because nothing either of them sent had room in it for a sentence like “and none of this matters to me now.”
Continuing is cheaper than stopping. Six months are agreed and very possibly paid. The sessions already sit on the calendar. Turning up twice a month for a pleasant hour is cheap. The conversation about ending is expensive. It asks the client to have a reason, deliver it to somebody they like, and manage how it lands. Most people take the cheaper option. That’s arithmetic, and it’s why drift runs all the way to the end of an engagement with neither person naming it.
You’ll want a rate at this point. Here’s why there isn’t one. The client who stops writing altogether has been counted, most carefully in psychotherapy, and the coaching collection’s guide to the client who has gone dark sets out that count and what to send somebody who has stopped answering. Wes produced none of that behavior. He attended, he replied, and he missed nothing. Every measurement of clients leaving counts the ones who stopped coming. The people this chapter is about sit outside all of them, including your own records. Your drift rate stays out of reach. What you can have is a way of noticing one client at a time, and that’s the rest of this.
How urgently to take any of it is a calendar question. Drift is often reversible in month three and almost never in month six, for a plain reason. In month three there are still sessions left to spend on whatever the real subject turned out to be. Everything below is arranged around getting the question asked while there’s still something to do with the answer.
The client who drifts quietly to the end, meanwhile, stays polite all the way out. They ask for nothing back. They refer nobody. Their honest account of six months with you is that it was fine. Fine doesn’t get said at a dinner table.
What a client stops doing before they stop answering
Six signals, roughly in the order they appear. The first is worth more than the other five together.
They stop bringing you anything you didn’t ask for. In a live engagement the client supplies material. They arrive with something that happened on Tuesday. Or a question that occurred to them in the car. Or a piece of news with no connection at all to what was agreed last time. In a drifting engagement, every item on the table came from you or from the previous session. The loop is closed. Nothing new enters from their life.
You can test that in about a minute. Over the last three sessions, ask who put each thing on the table. If the answer is you and the last session, both times, the loop is closed. A closed loop is drift even when everybody is enjoying it.
It’s checkable in writing too, and this is where the door the last chapter argues for opening in the first session earns its keep. The messages that arrive because you asked for them carry nothing at all. The one that arrives because something happened is the entire signal. Wes sent none of the second kind after November.
The replies shrink against their own baseline. Nine words in October, six in November, four in January. Measure the direction of travel against that person’s own first month. Plenty of capable people write four words in week one and mean nothing by it, so your other clients make a poor yardstick. On its own this signal is weak and people over-read it. In company with the first one it’s strong.
The reschedules only ever move later. Direction tells you more than frequency here. A client who moves Thursday to Tuesday is rearranging a week they’re still inside. A client who moves it to the following week, and then to the week after that, is buying distance from it. Somebody who reschedules twice a month and always earlier is busy. Somebody who reschedules once a quarter and always later is going. Catching the request at all is its own small problem. It usually arrives as a line in the middle of a friendly paragraph: a reschedule buried in a reply covers that half.
They take the easy commitment. The previous chapter argues for leaving a session with one commitment rather than three, partly so that which one they pick tells you something. Here’s what it tells you. A client still in it picks the one that will cost them something, because that’s the one they came for. A drifting client picks the one they can complete without changing anything. Then they complete it, on time, and report it. The choice is the signal. The completion is just the receipt.
News arrives late. Wes took the job in January and mentioned it in February. This one is definitive, and as a warning it’s worthless, because by the time it reaches you they’ve been gone for months. Treat it as confirmation of something you’d have wanted earlier. Then go and look at your other clients.
The sessions have become pleasant. Nothing difficult happens in them. You come off the call feeling good about your work. It’s the hardest of the six to notice, because you’ve no reason to want to. It’s also the one available to you today, about every client on your list, with no record at all. Try to remember the last time a particular client said something that surprised you, or something uncomfortable for either of you to hear. If nothing comes, that’s information.
Take these in company. Two of them across two consecutive sessions is worth a question. One of them is worth writing down and looking at again next time.
The quarter that looks exactly the same
Three states look identical from your chair, and only one of them is drift. Being wrong here in the confident direction costs more than missing it. A client who’s merely swamped, asked whether they’re still committed, hears an accusation, and they’re right to.
A heavy season. Somebody in the middle of a launch or a reorganization is short with everybody and brings nothing to anyone, and that’s the last chapter’s territory. Tell the two apart by what happens when you ask about the season. A client who’s buried will tell you about the thing burying them at length, and that telling is itself them bringing you something. A drifted client says it’s been busy and moves on to whatever you had planned.
A client who’s actually finished. Real progress looks a lot like drift. This one gets misread most often, because the misreading runs in your favor. Somebody who got what they came for writes shorter messages, because there’s less to say. Ask them what has changed and they answer with something specific and recent. That’s what separates them. And it’s worth naming plainly that finishing early is a good outcome your calendar has an interest in missing. If a client is done in month four of six, the useful conversation is about ending well.
A person who writes like that. Some people sent you eight words in week one. Baseline is the whole of this one, and it’s why the shrinking-replies signal is measured against the client’s own first month.
All three wear the same clothes. So a drift signal earns a question, never a conclusion. You’re not diagnosing anybody, and the question is worth asking even when the answer turns out to be a launch, because a client who has just explained their launch to you has stopped drifting for the length of the explanation.
A date in the middle, put there on day one
The single most useful thing you can build for this is a named session at the midpoint, with the engagement itself as its subject. The work waits an hour.
Almost all of its value is in when you schedule it, and the answer is before the first session. A review you invent in month four, because you’ve started to worry, arrives carrying your worry. It reads as a complaint or as the opening of a sales conversation, and the client handles it as one of those. A review that’s been on the calendar since the beginning is furniture, and nobody interprets furniture. It also costs you nothing to hold when everything is fine, and everything is fine in most of them. That’s the point. An instrument you reach for only when you’re suspicious announces suspicion.
Put it halfway. Session six of twelve, month three of six. Anywhere near the end it turns into a renewal conversation whatever you decide to call it, everybody in the room knows it, and the answers come out shaped by that. Whether to continue is a real conversation with its own difficulties, and it belongs to the last chapter of this series. The midpoint review is its own instrument, used at the one point where there’s still time to act on what it tells you.
Twenty minutes at the start of a session you were having anyway is enough. Three questions.
What’s different since we started? Ask for different, specific, and dated from a named month. A live client answers in about four seconds, usually with something you didn’t know. A drifting client comes up empty, and the silence while they try is the most useful twenty seconds of the engagement so far. Let it run. The urge to fill that pause with an example of your own is strong, and it’s the most common way this question gets wasted.
What have we spent time on that hasn’t been worth it? A client keeps criticism to themselves when they like you and are paying you, and asking whether there’s anything they’d change gets a no. This version takes for granted that there’s an answer. That’s the whole of the difference, and it routinely produces the most useful sentence of the engagement.
If the next three months went as well as they could, what would be true in March? Name the actual month. This is the one that separates the finished client from the drifted one. A finished client describes something modest and real. A live client describes something specific, usually a little different from what they said in session one, and that difference is the thing you came for. A drifted client produces something general that anybody could have said about anything.
One sentence when you book it and one when you open it, both saying the same thing. This is a look at whether the arrangement is aimed at the right thing. Their homework stays out of it. Skip that sentence and the second question gets a no and the first gets a performance.
Then write down what they said in their own words. You’ll want the words in month five, and the sentence a client gives you in a midpoint review is very often one they’ve said out loud to nobody else.
The question a yes cannot answer
Between reviews there’s one written instrument, and it’s built the opposite way round from the last chapter’s. That check-in was engineered so a client who had done nothing could still answer it in one line, and the engineering was correct. It’s also the exact message Wes answered six times, every word of it true.
So the requirement inverts. A question with no yes in it. Anything answerable in a word gets a word back from somebody who has gone. Still on for the 20th. How did it go. All good. Those are fine messages doing other jobs.
What you want is a question that asks the client for something only they have. It asks them to hold two points in time next to each other. A drifting client can’t do that without noticing that they’re drifting.
Frequency first, because it matters more than the wording. Once in an engagement, twice in a long one. Send it every two weeks and it becomes another message that gets a compliant answer, which is the thing you were trying to escape.
Subject: A question, and no hurry on it
Nothing to do with the 20th, which is on the calendar and is fine.
When we started in October you said you wanted to work out whether you actually liked the job or just liked being good at it. I’ve been wondering where you’ve got to on that, five months in. It’s the one part I can’t see from here.
Two lines whenever it suits, and there’s no right answer here. If the honest version is that it’s stopped being the thing you’re working on, I’d rather have that than a tidy one.
Four things are carrying that.
It quotes them. The sentence from October, in their words, word for word. That quote is the part a yes can’t answer. It asks them to put then and now side by side, and nobody can do that for them.
It carries nothing operational. Put one logistics item in and the reply comes back about logistics. The line about the 20th is there to settle a worry.
No deadline. A drift question with a due date is a task. Tasks get compliance, and compliance is the exact thing you’re trying to see past.
The last line makes the disappointing answer sayable. It’s the same move as the exit in a reminder, and the “either answer is a fine answer” in a check-in. Same work, same reason. If the only acceptable answer is a good one, you get a good one or you get silence, and neither was worth the message.
Read the shape of what comes back as well as the content. Two lines the same day naming something concrete is a live client, whatever the concrete thing turns out to be. Four days later, three warm sentences that could have been written in month one, that’s your answer. It’s a good answer to have. It’s the first true thing you’ve had in a while, and it arrives while you still have sessions left.
Send it once. A second one is a different message, and what it says is that you need something.
Saying the thing you have noticed
When the signals stack and the question comes back thin, the next move happens in a session. It depends on a pause, and writing has no pauses in it.
Two ways it goes wrong, and both are well meant.
The plea. “I feel like we’ve lost some momentum.” That puts your feelings in the room and hands the client the job of managing them. What comes back is reassurance. Reassurance closes the topic and changes nothing. A topic that’s been reassured shut is very hard to reopen.
The diagnosis. “I think you’ve disengaged.” That’s a verdict. People argue with verdicts or apologize to them, and the rest of the hour goes on whichever one they chose.
What works is an observation, bounded as small as you can make it, followed by silence.
“Something I want to put in front of you before we start. The last three times, everything we worked on came from me. In October you were arriving with things. I don’t know what that means and it might be nothing, but I’d rather ask than guess.”
Then stop talking. The pause is the instrument. It runs longer than is comfortable. Whatever fills it is the most useful thing you’ll get that hour.
The second half decides whether the first half was worth saying, and it’s the part most people leave out. Make continuing a choice. Say in a sentence that stopping is available, that pausing is available, and that changing what the two of you work on is available. Three doors, offered plainly.
A client who can’t leave can’t honestly stay. A good number of drifting clients drift for exactly that reason: ending felt like a bigger conversation than continuing. Hand them the smaller version and the third door opens far more often than the fear suggests. They tell you what actually changed, and you have a live engagement again with sessions left in it.
What happens after you ask
Four outcomes, and three of them are better than the thing you were doing before.
It turns back around. More common than people expect, and the reason is almost always the same. The subject changed and nobody updated the agenda. Somebody came in October about a promotion. By January the live question is whether to leave the company, or a parent is ill, or the promotion arrived and turned out to be the wrong job. The coaching held up fine through all of it. That’s the ordinary rate at which a working life moves underneath a six-month arrangement.
They finish, on purpose. They got what they came for. Close it out properly, and offer to hold the unused sessions open for a quarter if that suits how you work. What you buy is a former client whose account of you includes the fact that you told them when they were done.
They stay, and nothing changes. You asked, they said it was fine, and it isn’t. This happens. Ask once. A second attempt is you managing your own discomfort, and it turns the remaining sessions into a negotiation about the engagement. Do the work well, finish, and write down what you saw and when you saw it, because the next one will look the same.
They end it. This is the outcome coaches are afraid of, and it beats four more months of a dead engagement for a reason that’s easy to miss while you’re looking at the invoice. Your practice is ten or eleven slots. A drifted client is occupying one of them. The hour you get back is the next person who writes to you.
The honest cost, said once. Asking will occasionally end an engagement that would have run to term and been renewed out of inertia, and that’s real money you’d otherwise have had. Here’s the trade. An inertia renewal is the one that ends without warning in month two of the next cycle. A practice built on them can forecast nothing, and it can tell you nothing about which parts of the work are doing anything. Whether that trade is worth making in your particular year is yours to decide, and both answers are legitimate.
That leaves one thing to record, and it’s the column that was missing from everything Dana built in the fall. One binary, answered in two seconds at the end of every session: did anything come from them today that I didn’t ask for. Yes or no.
It takes about four minutes a month across a full list. Three nos in a row is the trigger for the question, and that’s the whole of what it is. It’s a great deal more reliable than the feeling that something is slightly off, because the feeling arrives about eight weeks late and only about the clients you happen to be thinking of.
The half of this an inbox can hold
The night Dana understood what had happened, she spent an hour reading five months of one thread in order. That instinct was right, and reading a correspondence back is the one thing in this chapter a mailbox is genuinely good at. Point turns that hour into a few minutes. A long chain carries a plain summary of what it is and where it got to, so the shape of five months reads at a glance, which is catching up on a long chain on its own terms. Finding the sentence from October is the same problem in miniature, and it’s usually why the drift question stays unwritten, since quoting somebody accurately means locating something you half remember from a season ago. You can go after it with the detail you actually retained, which is rarely the phrasing anybody used.
The midpoint review is a date four months out with nothing arriving to remind you of it, which makes it the most forgettable useful thing in the practice. An ask that arrives inside a client’s message becomes a dated item on its own, which is a task you never wrote down. The review is the other kind, the one you type yourself, and what matters is that it stays fastened to that client’s thread, where you’ll see it. So does the column. “Did anything come from Wes today” is one line after a session, filed where the rest of him is.
A standing watch is worth setting here too, and it’s worth being exact about its reach: it would have missed Wes. Elsewhere in this series a watch holds a specific overdue reply, a stranger still to come back by Thursday. Pointed at a client, it holds an absence over a much longer horizon. Tell me if six weeks go by with nothing at all from Hollis, said once, in those words, and then off your mind. That catches the client who’s drifting and going quiet at the same time, which is the more common combination. For the one who answers everything, it stays silent. Handing the waiting over in one sentence is the day-to-day version, and it’s honest about the kinds of waiting it doesn’t suit. Everything Point does sits behind all of it as a plain list, and email for business and executive coaches is this trade’s version of the whole thing, written for a practice with ten clients rather than a firm with staff.
Three limits, and they’re larger here than anywhere else in this series.
Everything that mattered about Wes happened where a mailbox has no access. The interviews. The weekend he decided. The evening in November when he told somebody at home. None of it produced an email to Dana, and none of it ever would have. A mailbox is a record of what was said to you, and drift is mostly made of what was left unsaid.
It can count days, and reading a person stays with you. Six weeks of quiet from Marisol in November means one thing. The same six weeks from Hollis in the middle of a reorganization means another. Telling those apart is a judgment about two people you know, and that judgment is the job.
And this is the one message in the practice with nothing to hand over. Every other message in this series gets easier the more of them you send, which is exactly what makes a reminder or a confirmation a sensible thing to let run further ahead of you. The drift question happens once. Its entire content is one sentence a particular person said to you in October. Review is the position everything starts in, which means the message written and nothing sent, and that position is held for each kind of action separately rather than once across the board. The autonomy dial walks through each position on it, up to and including the last one, where the work is handled without you. Leave this one where it is. The reason is plain arithmetic: one message, sent once, gives a pattern nothing to learn from.
Five seconds in March
Dana changed two things in February and backdated both onto every engagement already running. The column at the end of each session. A review at the midpoint, on the calendar from the start.
Wes first, since he was the reason. She sent the question in the last week of February, and he answered in two days, at some length, and with obvious relief. The new job started in April, and he’d been circling it since November. They spent his last three sessions on the transition. That had been the live problem since before Christmas, and it dropped off the agenda the moment he made up his mind about it. The final one in April was the best hour the two of them had. She’d got two useful sessions out of six, and she’d been eleven weeks late.
Then Hollis, who started in January. At the review in March, Dana asked what was different since they’d started, and Hollis sat there. Dana let the pause run, which took some doing. Then Hollis said that the promotion wasn’t happening, that she’d known since the middle of February, and that she’d kept it to herself because she hadn’t worked out yet what she wanted to do about it. Everything the two of them had worked on since January had been aimed at a job that no longer existed. The second half of that engagement was the best work Dana did all year, and a conversation she’d scheduled because she was worried would never have got there.
And then Marisol, whose six months ended on a Tuesday in March. She’d done the hard conversations, including one nobody asked her for. She’d answered her midpoint review with something specific and a little different from what she’d said in September, which is what a live one sounds like. The last session ran twenty minutes over and neither of them minded.
At the end of it Marisol said, “So what happens now?” and then waited.
Dana had built something for every silence in her practice by then. The one she had no answer for was five seconds long. She was sitting across from a client who was happy, and it was her turn to speak.