The therapist is the last to know the last session was the last

A woman I once heard about kept her last appointment, said "see you next week," and never came back. No call, no cancellation, no "this isn't working for me." Just a chair that stayed empty the following Tuesday, and then the Tuesday after that. Her therapist spent weeks replaying the final hour, looking for the moment she should have caught. There wasn't one clean moment. That is usually the whole problem.

And she was not rare. Of 100 people who first contact a mental health clinic, only about 50 show up for the evaluation, around 33 reach a first treatment session, and fewer than 17 are still there by session 10 (Barrett et al., 2008). Between 20% and 57% of the people who do attend a first session never return for a second (Baekeland & Lundwall, 1975). The empty chair, in other words, is the most ordinary thing in our field. We just keep being surprised by it one patient at a time.

We tell ourselves a comforting story about endings like this: that they arrive out of nowhere, that the client gave no sign, that nobody could have known. But that story is mostly wrong, and the way it's wrong is worth sitting with. The therapist is not the last to know because the signal never came. The therapist is often the last to know because the signal came, was felt, and went unspoken.

The signal that goes unspoken

Here is the finding that turns the usual narrative on its head. When Kullgard and colleagues (2022) surveyed 116 clinicians, 76% said they had sensed at some point that a client was about to drop out. Only 23% raised it with the client. Read those two numbers next to each other. Three in four felt the drift; fewer than one in four said a word about it. The quiet ending isn't a failure of perception. It's a failure of speech.

The same study asked those clinicians to estimate their own dropout rate. The average answer was 8.89%. The real figure for that population runs somewhere between 20% and 26%. So most of us are carrying a private number that is less than half of what is actually happening in our rooms, which is its own quiet kind of not-knowing.

Why does the sensing stay silent? Maybe because naming it out loud feels like inviting the very thing you fear. Maybe because a session that felt fine to you offers no obvious place to say "are we still in this together?" And maybe, this is the hard one, because we trust our own read of the room far more than the evidence says we should, and the gap between how the hour felt to us and how it felt to the person on the other side of the desk is exactly the gap that goes unmeasured and unspoken until the chair is empty. Murphy and colleagues (2022), in one study of CBT for depression, found that the client's session-one rating of the alliance predicted who would leave, and the therapist's rating didn't. Sit with that. The broader picture is more mixed than that single study, but the direction is consistent: how it feels across the desk is the reading that travels home with the person who might not return.

What they take home

So what is it they take home? When researchers ask the people who left, not therapists guessing but the clients themselves, the answer is rarely the technique. Homan (2025) found the most common reason was simpler and harder: they didn't feel heard. Not dismissed dramatically, not argued with, just the slow, sessions-long accumulation of being a case rather than a person, of sitting across from someone who was clearly attentive and clearly skilled and yet somehow tracking a framework instead of tracking you. That is the alliance quietly coming apart. And the alliance is the strongest consistent thread to outcome we have, across 295 studies (Flückiger et al., 2018). It pulls just as hard on who leaves: when therapists and clients rate the bond as weak, the odds of dropout climb at an effect size of d = 0.55, which in our field is not a footnote (Sharf et al., 2010).

Count your own caseload

Now hold the scale of it. About 20% of adults leave before treatment reaches its natural end (Swift & Greenberg, 2012); in high-income countries the cross-national figure runs closer to 26% (Wells et al., 2013). For adolescents and some other presentations it climbs higher still. And the exit isn't scattered evenly: in that same cross-national survey, dropout was most likely early, especially after the second visit, before any relief could anchor the person to the work (Wells et al., 2013). No Dutch national body, not the NZa, not Vektis, not Akwa GGZ, publishes a dropout rate at all. So do the arithmetic the system won't. Take your own caseload, the people you'll see this month, and count one in every five. That is not an abstraction. Those are faces you can name.

What does the silent ending actually cost? Not just the empty chair. It costs the one conversation that might have changed the trajectory, the chance to ask, to adjust, to repair the small tear before it widened, to close well if closing was right. Because leaving isn't always failure: Lopes and colleagues (2017) found 62% of dropouts had reached clinically significant change by long-term follow-up. Some people leave because they're done, not because they're hurt. But you only learn which one you're looking at if someone says it out loud while the chair is still warm.

The question still forming

You probably can't see the drift in real time. Almost none of us can. But you can ask. "Are we working on what matters to you? Do you feel heard in here?" Not as a risk check. As a real question, while there's still a next Tuesday to answer it.

This is the one thing Freudche does here. After a session, Engagement shows you how this patient engaged today, held only against their own usual pattern, never against anyone else. Whether their answers ran shorter than they used to, how often they moved into deeper or more feeling territory, where the longer silences fell. Not a score, not a percentage, not a risk level, no verdict at all. Just soft bands and a gentle trend, the conversational texture that is hard to feel while you are in the room with them. It does not interpret, and it does not predict. It surfaces what shifted, so you might notice it a little sooner. You read it. You decide. You bring it into the room.

Somewhere there is a Tuesday you are still replaying, the hour you went over looking for the moment you should have caught, the "see you next week" that turned out to be goodbye. There was no clean moment. There rarely is. There was only the question you felt forming and let pass, the one that would have kept the chair from going quiet on you. Ask it next time, while it's still next time.

Researched by Freudche.