Homework in therapy: designing the work between sessions

The most visible part of therapy is the consulting room; most of the change, though, happens outside it. One hour a week is a very small slice of a client’s waking time. The remaining days are the territory where what was discussed in session is either tested or never recalled at all. Between-session work — what we usually just call homework — is an attempt to connect those two territories.

The idea was born in the behavioural tradition, and there it counts not as a by-product of therapy but as a direct instrument of it. In CBT a session is largely the place where next week’s trial is designed: a behavioural experiment, a thought record, a step of graded exposure, an activity schedule. But between-session work is not only a behavioural tool. Journalling in schema therapy, values-based action in acceptance and commitment therapy, noting dreams or associations in psychodynamic work, trying out a change to household arrangements in family work — all of them share the same logic: carrying an understanding built in the room to the place where that understanding is actually needed.

Why homework makes a difference

Three mechanisms make it easier to think about.

Generalisation. Something the client can do in session, with the therapist present, does not count as part of their repertoire until it has been repeated in the therapist’s absence. If a sentence that comes easily in the room cannot be said in front of a manager on Monday morning, that sentence has not yet been learned. Homework makes visible the point where the learning comes unstuck from its context — and that point is usually somewhere neither of us expected.

Data. What we know about a client’s week is mostly a retrospective account given in the first few minutes of the session, and that account is shaped by the mood of the last two days. “The whole week was awful” systematically erases three good hours on Tuesday. A simple record — which days, at what times, at what intensity — gives back the detail the retrospective account has smoothed over. Case formulation feeds on that detail.

A sense of owning one’s own action. A client who sees that a step they took on their own worked starts to attribute the improvement to themselves rather than to the therapy. That attribution is closely tied to holding on to gains after termination. A process in which everything is managed by the therapist leaves behind, when it ends, not a method but a longing.

The word “homework” carries the wrong associations for most clients

The schoolroom association is strong: homework is set, marks are deducted if it is not done, and there is an authority checking up. That is precisely the kind of relationship we want to avoid. There are useful alternatives — “between-session work”, “this week’s experiment”, “practice”, “an observation task”, sometimes just “shall we try something?”. Which word you pick matters far less than making the choice together with the client. For some clients “homework” is perfectly neutral and even gives a sense of structure; for others it is a quiet source of resistance from the first session onwards, and they will not tell you.

The same goes for how you put a task forward. The difference between “I’d like you to do this over the week” and “would it help to try this over the week — shall we look at it together?” is not a difference in politeness but one that bears directly on adherence.

The expectation is set in the first session

Bringing between-session work in halfway through therapy looks, from the client’s side, like an unexpected change of rules. Said in a single sentence at the first meeting, while you are explaining how the process works, the same thing becomes an ordinary piece of information: “We’ll be trying out some of what we talk about here during the week in between; that’s where most of the real work happens.”

That sentence does two things. First, it takes homework out of being a burden bolted on later and puts it inside the definition of the process. Second, it lets you hear the client’s objection early — and that objection usually brings with it something worth hearing about a previous therapy they left halfway through, or about their school years.

There is one more thing worth saying in the same meeting: that homework left undone is not a problem but something to talk about. When a client does not do the task in week three, the tendency to miss the session, or to gloss over it if they come, usually grows out of that sentence never having been said.

Tasks differ according to their purpose

In practice, quite different pieces of work travel under the heading of “homework”, and knowing which one you are doing makes it much easier to scale the task correctly.

Observation tasks gather data; they do not aim at change. Sleep times, when panic attacks occur and in what context, how mood moves across the day. This is the most useful type in the early phase of therapy and the lowest-risk one for the client — nothing is being asked of them but to look. A common mistake in presenting these tasks is failing to say that observation is itself an intervention: many clients notice the pattern shifting simply because they started recording it, and take that to mean “I didn’t do the homework properly”.

Skills practice runs something learned in session under cold conditions: a breathing exercise, a mindfulness practice, reading a coping card. The critical point here is that the skill should never be tried for the first time as homework. An exercise never done in session gets done wrongly at home and comes back with the conclusion “it doesn’t work for me”.

Behavioural experiments test a prediction. The client’s expectation is written down in session — “if I raise my hand everyone will look at me and laugh” — and compared with what actually happened during the week. In this type it matters less that the task was done than that the prediction was recorded beforehand; without the record, the client describes the outcome as they remember it, which is to say as it fits their expectation.

Exposure steps work a single rung of the hierarchy, and the duration and number of repetitions have to be set explicitly. “Going out into a bit of a crowd” is not an exposure task.

Values-based action tasks aim not at an outcome but at a direction; the criterion is whether the person acted in line with their own values. Here, instead of “did it work?”, the question is “did you act like the person you want to be?”.

Choosing the type of task is a decision that comes before calibrating difficulty. Setting up a behavioural experiment while the client still cannot see their own pattern means skipping the observation stage, and usually you have to go back and start there.

Not a prescription but a joint design

Adherence drops on tasks the therapist sets alone; the reason is not the client’s stubbornness but the fact that the task does not belong to their week. You do not know their Tuesday evenings, how crowded the house is, how much energy is left at the end of a shift. They do.

This does not mean leaving the task entirely to the client. The division that works goes like this: you bring the direction and the set of options, you make the choice and work out the detail together, and the final version goes into the record in the client’s own words.

• “I’m thinking about approaching one of the situations you avoid this week” — the direction is yours.
• “Which ones come to mind?” — the list is the client’s.
• “Which of these is genuinely possible this week?” — the choice is joint.
• “What exactly will you do, when, and where?” — the detail is the client’s.

Skip that last step and what is left is an intention, not a task. “I’ll go out more” is not a task. “A walk round the neighbourhood after dinner on Wednesday and Saturday, about twenty minutes” is a task.

What makes a task a good one

Running a task past the following headings before the session ends removes most of the next week’s surprises:

• Observable. What is to be done is defined as a behaviour: not “worry less” but “get into the lift”.
• Fixed in time and place. An unspecified day means, in practice, no day at all.
• A single step. Trying two new things in the same week raises the odds that neither gets done.
• Tied to the action, not the outcome. If the criterion for success is “if my anxiety drops”, then when it does not drop the client counts both the task and themselves as failures. The criterion should be “I got into the lift”; what the anxiety did is data to be collected.
• Quick to record. Any record-keeping format that takes longer than thirty seconds is abandoned by day three.
• Justifiable in the client’s own words. Asking, before the session ends, “why are we doing this — can you put it in your own words?” catches most misunderstood tasks right there.

Calibrating difficulty to readiness

A task that is too easy is boring and devalues the therapy; the client thinks “what good is this supposed to do?”, and they are right. A task that is too hard simply does not get done, and not doing it usually feeds the inadequacy schema the client was already carrying. The second error is the more expensive of the two: once a task is experienced as a failure, it leaves a reluctance towards the tasks that follow.

A practical dial: once the task is defined, ask “how confident are you that you can do this, from zero to ten?” If the answer is below seven, make the task smaller. Trust that number, not the polite “I’ll do it” the client gives you.

The first piece of homework in particular should be all but certain to be doable. The function of the first week is not to produce clinical progress but to establish the arrangement that something gets done between sessions and that the something gets discussed in session. In depression that might be the smallest step of behavioural activation — opening the curtains, sending one person a one-sentence message. In anxiety, the bottom rung of the hierarchy. Making the first step smaller than it needed to be costs almost nothing; making it bigger than it needed to be costs several weeks.

A rehearsal inside the session

Mentioning the task in the last two minutes, while the client is putting their coat on, is a widespread habit and one of the things that most damages adherence. If you take between-session work seriously, it has to show up in the schedule too: the last eight to ten minutes belong to design and rehearsal.

The rehearsal can be done mentally. “Thursday evening, eight o’clock. Where are you, who else is at home, what’s going on?” As the client sets the scene, the obstacles surface by themselves: the child is being put to bed at that hour, work finishes late that day, if the notebook is left somewhere visible their partner will ask about it.

Writing the obstacles straight into the plan then helps: “If …, then …” If I’m late on Thursday, I’ll do it at the same time on Friday. If I can’t find a quiet spot at home, I’ll take the record as two lines in my phone’s notes app. A backup plan of this kind pulls the task out of the all-or-nothing binary.

Who writes the task down is not incidental either. A task written in the client’s own handwriting, in their own words, is a different thing from a sheet of paper you hand them.

The format of the record decides more than its content

Where adherence drops is usually not the task itself but the logistics of the record. A seven-column thought record is clinically rich, but it does not get filled in by a tired person at eleven at night. Collecting the same information in three columns beats never collecting the full form at all.

A few practical notes:

• Put it where the client already is. Do not hand a notebook to someone who keeps notes on their phone; do not recommend an app to someone who carries a notebook.
• Think about the frequency of the record, not the quantity. One line a day does more work, and gets forgotten less often, than one page a week.
• Make the reminder part of the task. A phone alarm or a note stuck to the door reduces the work loaded onto willpower.
• Send the form before the session. When the client completes a scale or a record in their own time, that work comes out of the session and it is in your hands before you walk into the room.

Reviewing the record format after a few weeks should also be part of the routine. The format chosen first usually gets simpler on the second attempt.

When the homework is not done

This is the most instructive moment in between-session work, and the one most often used badly. An undone piece of homework is not an adherence problem but a finding. The question to ask is not “why didn’t you do it” — that question generates defence — but “where did it stop?”

The patterns you meet in practice fall under a few headings.

The task, or its rationale, was not understood. The most common reason, and the easiest to fix. The client says “I forgot”, but in fact they were not sure exactly what to do. The fix belongs to the next design: have the client repeat the task back, and write it down.

There is a practical obstacle. Shared housing, no privacy, shift work, childcare, a phone used by the whole family. If these get mistaken for resistance, the client has been interpreted unfairly. The fix is to change the form of the task: a short voice note instead of a written record, lunch break instead of evening, on the way to work instead of at home.

The task was too big. The client puts off starting for two days and then decides it is too late. The fix is to make the task smaller — even at the price of making it smaller again next week.

The task touches exactly what is being avoided. This is a sign that the task is working, and it is the case in which not doing the homework is clinically most meaningful. What to do here is not to abandon the task but to make the avoidance itself the subject of the session. “What happened when you sat down on Thursday evening?” often opens the most productive fifteen minutes of the week.

The client cannot connect the task to their own goal. The link between the formulation and the task is clear in your mind but has not been shared. The fix is to go back and talk the goal through again; sometimes it is to notice that the goal has changed.

Something relational is happening. The task has been experienced as a demand, and the client has taken up a position somewhere on the compliance–defiance axis. For a client whose relationships with authority figures are difficult, homework is a transference object planted in the middle of the therapeutic relationship. Handling it as an adherence problem does nothing but play the pattern through again.

At this point the therapist’s own reaction is data too. The disappointment felt in the face of undone homework, the “I put the work in” resentment, the impatience — these are the voice of countertransference, and they leak easily into the tone of the session. When they leak, the client notices — usually before you do. It is material worth taking to supervision.

And the reverse holds as well: never mentioning the homework is also a reaction. Skipping the subject so as not to upset the client is the clearest possible way of signalling that it did not matter.

The first five minutes of the next session

What settles the fate of homework is less how it is given than how it is reviewed. Homework that goes unmentioned once has in practice been cancelled; the client reads that correctly and does not do it the following week.

A review that works is short but curious:

• “What did you do?” — the facts first.
• “What did you notice?” — then the observation.
• “Did what you expected happen?” — then the comparison with the prediction. In behavioural experiments the real clinical gain is here.
• “What does that tell you?” — the meaning last.

When it has been done, “well done” is tempting, but it turns the homework into something that gets marked. Curiosity works better than approval: what happened, what was surprising, which part was hard.

When it has been done in part, go over the part that was done. A record kept on three days out of seven is three days of data — not nothing. Most clients present a partially completed task as though it had not been done at all; correcting that presentation is an intervention in its own right.

Tracking adherence over time

Patterns that go unnoticed session by session become visible with a simple record. Three pieces of information per session are enough: was homework set, was it done, was it done in part. Keeping this in the client file pays off in several ways.

A run of tasks left undone is a warning about the formulation. If three sessions in a row have gone undone, the thing to do is not to make the task bigger or to press harder, but to review the formulation. Usually either the goal is not the client’s goal, or there is something in the alliance that has not been talked about.

If adherence is high but the scale is flat, the task is wrong. If the client does the homework every week but the PHQ-9 or GAD-7 curve runs level, the problem is not effort but direction. This is the picture in which the decision to change the task shows up most clearly.

Read it alongside the measurement rhythm. The homework record on its own is an indicator of effort; set beside the trend on a scale, it comes closer to an indicator of effect.

The real risk here is that the tracking itself turns therapy back into school. The moment you start calculating an adherence percentage, checking a chart at the start of the session or saying “you didn’t do it last week either”, the relationship slides towards teacher and pupil, and between-session work stops delivering what it should. The record exists for the therapist’s own thinking; it is not a report card shown to the client. What gets shared with the client is not the adherence rate but the curve on the scale — the data on whether it is working, not on how hard they tried.

Constraints that come with the setting

A few constraints that come up often in practice in Turkey bear directly on the design of homework, and are usually assumed rather than discussed.

Privacy comes first. In a home shared with an extended family or with flatmates, the chance of a journal being seen is a real worry, and the client will often not raise it unprompted. A note on the phone, a locked app, or a record kept in abbreviations only the client can decipher are alternatives that do the job.

Working patterns are another. For a client on shifts, working long hours or holding down two jobs, “every evening at the same time” does not work from the outset. Instead of a fixed clock time, a fixed anchor — “when I get on the bus after work”, “after dinner” — proves more durable.

Then there is stigma. For a client who has not told their family they are in therapy, filling in a visible form at home is not a simple logistical matter. Designing the task around that reality is more productive than interpreting the resistance.

In online sessions

Working over a screen makes the concreteness of the task even more important, because the cues the room provides are missing: a sheet of paper pressed into your hand, a board seen on the way out, the place itself. Against that, being able to send a form instantly is an advantage.

A few arrangements work. Filling the task in during the session, with the screen shared, means the first attempt happens right there. Sending the text of the task in writing as the session ends removes the “what exactly was it we said?” question. Sending the scale or record form before the session is particularly valuable in online work — because a minute spent in front of a screen costs more than a minute spent face to face.

Towards termination: handing over ownership

As therapy progresses, who sets the tasks should change. At the start the therapist brings the direction; in the middle they are built jointly; towards the end the client should be able to define their own task. That shift does not happen by itself, it has to be made explicitly: from a certain point on, asking “what are you suggesting to yourself this week?” regularly.

The relapse prevention plan is itself a continuation of the same logic. A text in which the client writes, in their own words, the signs of deterioration, the first steps to take and whom to contact — that is the most meaningful piece of between-session work done in the final sessions. What the client is left holding when therapy ends is precisely a text of that kind and the habit of using it.

---

The summary of between-session work is short, in fact: build the task together, shrink it until it fits the client’s week, rehearse it inside the session, always ask about it in the next session, and read it as information when it has not been done. The rest is a repetition of those five steps.

Keeping the homework assignment, its completion status and the scale trend in the same client file makes that tracking easier; in psisula the three sit in the same record. But the tool is secondary — what really decides the matter is that the task was built in the client’s own words and that it is genuinely asked about the following week.