Why Clients Don't Do Therapy Homework (And How to Assign Work They'll Actually Complete)

Clients don't skip between-session work because they're unmotivated. In most cases the task failed before they left the room: it was assigned rather than designed with them, it was never made concrete enough to do, its rationale was never made explicit, or it was never reviewed in the following session — which quietly taught the client it didn't matter. Fix those four things and completion rates move without any change to the client.

The reflex when work doesn't get done is to attribute it to readiness or resistance. Sometimes that's accurate. More often the assignment itself was the weak link, and that part is entirely within our control.

Why This Is Worth Getting Right

Kazantzis and colleagues published a meta-analysis in Behavior Therapy in 2016 covering 17 studies and 2,312 CBT clients. Homework quantity — how much got done — related to post-treatment outcome at Hedges' g = 0.79 (k = 15, n = 1,537). Homework quality — how well the client engaged with the task — came in at g = 0.78 post-treatment (k = 3, n = 417) and rose to g = 1.07 at follow-up. Quantity at follow-up dropped to g = 0.51. All differed significantly from zero.

The interesting part is the crossover. At follow-up, quality outperformed quantity — how well a client engaged with a task predicted durable change better than how many boxes got ticked. That cuts against the reflex to assign more when progress stalls. Two caveats: the quality analyses rest on only three studies, so that finding is suggestive rather than settled, and all of this is correlational.

The Four Ways an Assignment Fails

1. It was assigned, not designed together

The task arrived as a directive rather than something the two of you built. Clients comply with directives about as well as anyone complies with an instruction handed down without input — unevenly, and worse when the week gets hard. Collaborative design isn't a courtesy. It's what gives the client a stake in the outcome.

2. It was too vague to be a task

"Notice your thoughts this week" is not an assignment. Neither is "try to be kinder to yourself." A task has a what, a when, and a where — "one thought record on Tuesday evening after the kids are down, at the kitchen table." The vague version can't be completed or failed, which means it can't be reviewed either.

3. The rationale was never made explicit

If a client can't say why they're doing something, it reads as busywork, and busywork loses to every other demand in a week. Ask them to state the reason back in their own words — a rationale they generate holds better than one you supply, and a thin answer is worth hearing before they leave.

4. It was never reviewed

This is the one I'd change first if you only change one. If the client does the work and you open the next session with "so how was your week," you've taught them the task was optional. Review it in the first five minutes, every time, whether or not it got done. What never gets checked stops existing.

5. It was sized for a better week than the one they had

A task calibrated to the client sitting calmly in your office on Thursday is often miscalibrated for that same client on Monday at 10pm with a sick kid. Size the work for their worst realistic week, not their best.

A Better Sequence for Setting It

1. Start with ten minutes left, not thirty seconds. Work assigned while the client is putting their coat on gets the attention it was given.
2. Have them say it back. "Tell me what you're going to do and when." Any gap between what you meant and what they heard shows up here, while it's still fixable.
3. Use a scaling question. "One to ten, how likely is it you'll do this?" Under seven means the task is too big. Shrink it and ask again, until the number goes up — a smaller task completed beats a better task abandoned.
4. Name the obstacle out loud. "What's most likely to get in the way?" Then solve for that specific thing in the room. Anticipated obstacles get planned around; unanticipated ones end the practice.
5. Write it down where they'll see it. Their phone, their notes app, a card in their wallet. Not your progress note.

It helps to keep a few ready-made tasks that carry their own scaffolding. The CBT Thought Record and Labeling Emotions Workbook does the specifying for you — the columns tell the client what a completed task looks like, which removes the ambiguity that sinks open-ended assignments. For what a client carries into a session and takes back out of it, the Pre and Post Therapy Workbook gives them a consistent structure for both ends of the hour. If you're assigning a thought record specifically, send them to the step-by-step guide to using one so the instructions don't have to survive the trip home in memory alone.

Reviewing It Without Shaming

When the work didn't get done, the non-completion is data about the task or about the ambivalence. It is not data about the client's character, and if your opening line implies otherwise you'll get compliance theater for the rest of treatment. Language that opens the conversation rather than closing it:

• "You didn't get to it — that's useful. What got in the way?"
• "I think I made that too big. What would a version you'd actually do have looked like?"
• "Was there a moment you thought about doing it and decided not to? Tell me about that moment."

That last one is the most productive, because what surfaces there is often more useful than the completed worksheet would have been. If the client isn't in an action stage at all, the assignment was never the problem — see working with ambivalence using the stages of change for what to do instead of pushing.

One more thing: consider dropping the word "homework." Many clients carry school associations with it — being graded, being behind, being in trouble for not doing it. "Practice," "experiment," or "between-session work" costs you nothing and lands better. "Experiment" is particularly useful, because an experiment can't be failed, only run.

For clients: to hand a client something that explains the between-session piece from their side, three things to do between therapy sessions covers the same ground without the clinical framing.

The Honest Limitation

Sometimes non-completion isn't an assignment problem at all. It's a signal that the alliance is off, or that you're working toward a goal the client didn't pick, and no amount of scaling questions will fix that. If work stops getting done across several weeks and different task types, stop refining the task and ask about the relationship and the goal directly. Guided discovery suits that conversation better than problem-solving does.

Some clients also improve without between-session tasks at all. The evidence here is correlational, and treating completion as a proxy for engagement will lead you to misread the client whose circumstances leave no room for it — the one working doubles, or caring for a parent, or living somewhere without a private ten minutes. That's a formulation issue, not a compliance one.

The Bottom Line

Between-session work predicts outcome, and how well a client engages with it appears to matter more over time than how much of it they do. Most non-completion traces back to four fixable design errors: the task wasn't built collaboratively, wasn't specific, wasn't justified, or wasn't reviewed. Assign smaller, assign together, and check it in the first five minutes of the next session.

This article is written for clinicians and is offered as general professional education. It doesn't replace your own clinical judgment, your supervision, or the limits of your scope of practice with any particular client.