You can do the best 53 minutes of your clinical career, watch a client land on something real, walk them to the door — and have it gone by the following Tuesday. Most of us stop noticing how often that happens because it's just the texture of the work. But it's worth sitting with, because it points at where the real work is, and it isn't inside the room.
This is a clinician-to-clinician piece on the part of the work nobody trains you for: what happens to the client in the time you're not together, and what — honestly — holds them there.
The 53-minute problem (why this is the real work)
Here's the frame I keep coming back to, and I'd encourage you to say it out loud in your own caseload terms. A client sees you for one session a week. Call it ~53 minutes. There are 168 hours in a week. So the relationship you're building, the insight you're co-creating, the regulation you're modeling — all of it has to survive about 167 hours on its own.
53 against 167. The 167 almost always wins, and not because the client is resistant or unmotivated. It wins because the pattern lives out there. The reactive text to the parent, the third drink, the dissociation at the dinner table, the part that takes over at 11pm — none of that shows up in your office. You get the report of it, the residue of it, sometimes a live flicker of it. But the pattern's home field is the 167 hours, and that's exactly the stretch the client walks into alone.
When we say a client "isn't making progress," what we often mean is that the work isn't surviving the gap. The session is fine. The 167 hours are where it dies.
What "engagement between sessions" actually means (and doesn't)
Let's be precise, because the term has been flattened by software marketing into "did the client open the app."
Between-session engagement is not:
- Compliance with assigned tasks (a client can complete every worksheet and change nothing)
- Logging mood for your dashboard
- Reminder pings and streaks
- A measure of how much the client likes therapy
Between-session engagement is: the degree to which the therapeutic work stays available to the client when the pattern shows up — when the nervous system is activated, the part is blended, the old story is loud. It's whether the client can locate, in the moment, something of what the two of you built. That's a relational and a state-regulation question, not an adherence question. Mistaking one for the other is why so much "engagement tech" fails clinically even when the usage numbers look great.
Why insight doesn't survive the week
If you've ever wondered why a genuine insight evaporates, the cleanest read is state-dependent learning. The understanding a client reaches in your office is encoded in a particular state — co-regulated, ventral, witnessed by you, safe enough to look at the hard thing. That state is part of the memory. When the client is back in the activated state where the pattern actually fires, the insight is filed somewhere they can't reach. It's not that they forgot. It's that the version of them who knew it isn't online.
This is also why the work has to meet the moment, not the calendar. Insight reached in a regulated state and only revisited a week later, in another regulated state, never gets rehearsed in the state where it's needed. The gap isn't a time problem. It's a state-access problem.
And the pattern, meanwhile, is doing what patterns do. The attachment strategy that kept someone safe at seven is still running the show at thirty-seven, faster than thought, before the prefrontal cortex gets a vote. The body sees the pattern coming before the mind names it. In the 167 hours, with no witness and no co-regulation, the client is the least equipped to interrupt the very loop you spent the session illuminating.
Why generic homework fails (the adherence problem)
So we assign homework. And we all know the quiet truth: most of it doesn't get done, and a fair amount of what does get done doesn't help. There are real reasons, and none of them are "the client is lazy."
- A worksheet is a clinician's artifact, not the client's. It's in your language, organized by your model. When the client is activated, your CBT thought record is the last thing that feels like theirs.
- It asks for the wrong state. Homework usually requires the regulated, reflective state — the one that's offline exactly when the pattern hits.
- It points away from you. A form the client fills out alone, to bring back for grading, subtly re-enacts the disconnection. It's work done at the client, not with them.
- It's scheduled, not situational. "Do this Wednesday" rarely lines up with Thursday at 11pm when the part actually takes over.
This is the adherence problem, and it's a clinical design problem, not a motivation problem. The fix isn't more compliance pressure. It's redesigning what we offer the client to hold.
What actually holds a client between sessions
After enough years of watching what survives the gap and what doesn't, three principles hold up. They're modality-agnostic — they're as true for an EMDR clinician as an IFS or somatic one.
Principle 1: Meet the moment the pattern shows up, not the worksheet you assigned. The useful intervention is the one available at 11pm, in the activated state, not the one scheduled for the calm of Wednesday afternoon. Anything that only works when the client is already regulated is solving the wrong problem.
Principle 2: It has to be in the client's own language and parts, not your form. The thing a client can reach when they're flooded is their word for the part, their image, their felt sense — not your intake vocabulary. Whatever you give them to hold has to be built out of what's already theirs.
Principle 3: It carries material back into the room — it doesn't replace the room. This is the line that separates a clinical tool from a substitute. The point of between-session work is not to give the client somewhere else to go. It's to deepen what they bring back to you. If a between-session practice pulls a client away from the relationship, it's working against the therapy, however slick it looks.
Hold those three and the modality-specific applications fall into place.
IFS parts work between sessions
IFS is the modality I see clinicians reaching for hardest right now, and it has a real between-session problem and a real between-session gift. The gift: parts language is portable. A client who can say "I notice the manager is online right now" in the moment is doing genuine work, alone, in their own vocabulary — Principle 2, exactly. The problem: blending. When a part takes over in the 167 hours, the client is the part; there's no Self present to unblend, and no therapist to help them find it.
What holds here is anything that helps the client notice the blend as it's happening — to catch "a part is driving" before the whole afternoon is gone — and to do it in the parts language they already trust. Not a Self-led meditation they'll never open when flooded. A way to register, in real time, that this is a part and not all of them. That noticing is the entire ballgame, and it's what they then bring back to the next session.
Somatic and nervous-system regulation
For the somatic and polyvagal-trained among us, the between-session question is even more concrete: the body is where the pattern fires first. The client doesn't think their way into a shutdown; they're in it before the story arrives. So the between-session work has to be bottom-up and available pre-cognitively — an orienting cue, a breath the body already knows, a felt anchor you built together that they can find without having to be calm first.
The trap is handing a dysregulated client a regulation exercise that assumes regulation. What holds is something that meets the activated body where it is and gives it one small, real degree of choice.
EMDR and attachment containment between sessions
EMDR clinicians already think rigorously about between-session safety — the container exercise, the calm place, the resourcing you do precisely so processing doesn't run wild outside the room. The between-session question is whether those resources are actually reachable when material surfaces midweek. Containment that only exists as a memory of a session isn't containment. The attachment frame says the same thing from the other side: what you're really extending into the 167 hours is the holding — the felt sense that someone has the client in mind. The aim isn't to do EMDR alone at home. It's to make the resourcing portable enough that the client can steady themselves until they're back with you.
The line between a clinical mirror and an AI chatbot
I'll name the fear most of us carry, because the clinicians I work with carry it loudest: we are watching consumer AI chatbots market themselves as therapy, and we are right to be alarmed. So let me draw the line clearly, because it's a category line, not a feature line.
The contrast that matters: licensed clinicians respond appropriately to clients in distress around 93% of the time. The appropriateness rate for general-purpose AI chatbots in the same situations falls below 60%. That gap is not closing because someone tunes a prompt. A chatbot that pretends to be the therapist is doing a job it cannot do safely, and every clinician's instinct to recoil from that is sound.
But "a chatbot pretending to be the therapist" and "a tool that helps the client see their own pattern between sessions" are different categories. One tries to replace the relationship. The other is a mirror — it reflects the client's pattern back to them, in their own language, so they can notice it while they're inside it, and bring that noticing to you. A mirror doesn't interpret, doesn't treat, doesn't hold the clinical responsibility. It extends your holding into the hours you can't be there, and then routes the client back to the room. The clinician stays the clinician.
Two non-negotiables follow. First, anything touching a client's clinical material has to be HIPAA-compliant by design, not as an afterthought — a consumer app is not that. Second, the clinician has to stay in the loop; a tool that goes around you, straight to the client, has crossed into the consumer-app lane that sells "do this without a therapist," which is the opposite of what we want.
A quick self-audit — is your between-session work actually working?
Run this on your own caseload. It's ten questions, and it's diagnostic, not a grade. The pattern of your answers tells you whether your between-session work is meeting the 167 hours or just generating activity.
- Can your client name the pattern in their own words — without your handout in front of them?
- Does what you've given them work in an activated state, or only a calm one?
- When the pattern fires midweek, is there anything reachable in the moment — or only the memory of last session?
- Is the between-session practice in the client's language, or in your clinical vocabulary?
- Does the work carry material back to you, or send the client somewhere else with it?
- Have you ever actually asked what they did — or didn't do — and why, without it landing as a test?
- Are you assigning tasks, or building something the client co-owns?
- Does your between-session approach assume a regulated nervous system that may not be available?
- If a client is blended or shut down at 11pm, is there a path back to one degree of choice?
- Is whatever you use HIPAA-compliant, and does it keep you — the clinician — in the loop?
If most of your answers point to "scheduled, in my language, only works when they're calm, and I'm not really in the loop for it," that's not a you-problem. That's the design problem the whole field is sitting in.
Run the full 10-question self-audit on its own page → — free to use, print, or share with colleagues.
Comparison — between-session tools, honestly
A quick, non-promotional map of the categories, because the words get muddied. None of these is wrong; they're built for different jobs.
| Category | Examples | What it does | What it doesn't do |
|---|---|---|---|
| Homework / worksheet engines | Quenza, TherapyAssist | Assign and track structured exercises and care pathways | Catch the pattern in the moment; work from the client's own language |
| Documentation-first EHR portals | Practice-management suites with client portals | Forms, reminders, scheduling, goal-tracking inside your records system | Hold the between-session relationship clinically |
| Consumer self-help apps | Direct-to-client parts/regulation apps | Give the client tools on their own | Keep the clinician in the loop — they go around you |
| A between-session mirror | VibeCheck | Reflect the client's pattern back, in their language, in the moment — and route it back into the room | Treat, interpret, or replace you |
And the clearest way to say what a mirror is and isn't, straight from how I describe it to colleagues:
| A between-session mirror IS | It is NOT |
|---|---|
| A mirror that helps a client see their own pattern | An AI therapist |
| A clinical tool, built by a clinician, for the space between sessions | A chatbot |
| HIPAA-compliant by design | A wellness app |
| A way to extend the holding into the 167 hours | A translator |
| Grounded in attachment, parts work, and nervous-system regulation | A replacement for you or the work |
It's worth saying out loud that this all connects to the part of the job that's quietly grinding us down. Documentation eats something like 30% of the clinical week, and a disengaged 167 hours means more crisis-management, more re-treading of last session's ground, more notes about why nothing moved. Between-session work that actually holds isn't one more thing on your plate. Done right, it's the thing that takes a few things off it.
FAQ
How do I get clients to do anything between sessions?
Stop assigning and start co-building. Adherence collapses when the task is in your language, scheduled for the wrong moment, and done alone to be graded. What works is something in the client's own words, reachable in the activated state when the pattern fires, that carries back into the room. You're not trying to manufacture compliance — you're trying to make the work available when it's actually needed.
What's the difference between therapy homework and between-session engagement?
Homework is a task you assign and the client completes (or doesn't). Engagement is whether the therapeutic work stays available to the client when the pattern shows up midweek. A client can finish every worksheet and change nothing; another can do no formal homework but catch the pattern in real time and bring it back. The second is engagement. The first is just compliance.
Is it safe to use an AI tool with clients?
It depends entirely on the category. A general-purpose chatbot pretending to be a therapist is not safe — clinicians respond appropriately to distress around 93% of the time; AI chatbots fall below 60%. A HIPAA-compliant tool that acts as a mirror — reflecting the client's own pattern back, keeping the clinician in the loop, and routing the work back to the room rather than replacing it — is a different thing entirely. Safety lives in the design: HIPAA-compliant, clinician-in-the-loop, and clearly not a treatment substitute.
How does between-session work reduce my own burnout?
A disengaged 167 hours shows up in your week as crisis calls, re-treading old ground, and notes about why nothing moved. When between-session work actually holds, sessions go deeper faster because the client arrives with real material instead of starting from zero. Less re-litigating, less crisis triage, less documentation about stalled progress. It's structural relief, not one more task.
Does this replace the therapeutic relationship?
No — it extends the holding into the 167 hours you can't be there. A clinical mirror reflects the client's pattern back so they can notice it in the moment and bring it to you. It doesn't interpret, treat, or hold clinical responsibility; that stays with the clinician. Anything that pulls the client away from the relationship is working against the therapy. The point is to deepen the room, not to substitute for it.