Therapy Homework Clients Actually Do: A Guide for Therapists

You know the pattern. You recommend something for the client to try between sessions. They agree—but it does not happen, or it happens once and disappears. The next week, you are both back in the room talking about the same pattern.

The arithmetic is stark. You get 50 minutes a week. The client gets the other 10,030.

The problem is rarely the assignment itself. It is the handoff: the fifteen seconds when you explain it, whether the client leaves with it written down, whether it is tied to a moment in their actual life, and whether you return to it next session.

If you landed here from the client side rather than the practitioner's, journaling between therapy sessions covers the same week from your seat, and what to talk about in therapy covers choosing what to bring back in.

What makes a between-session assignment come back done?

Five rules, in the order they break.

1. Assign one thing, and let it be the small version. Conklin and colleagues tracked homework across five CBT protocols in 179 clients with generalized anxiety, panic, social anxiety, or OCD. The Unified Protocol carried a lighter homework load than most of the single-diagnosis protocols, and it produced no difference in average compliance or quality. Assigning less cost nothing on either measure. That is not the same as heavier protocols getting less done, which this study does not test. In the same sample, homework quality was significantly related to every symptom outcome measured, while compliance reached significance only on clinician-rated anxiety.

2. Attach it to a cue, not to a day. "Journal this week" asks the client to make a fresh decision every evening about when. With an if-then plan they decide once. Toli, Webb and Hardy at Sheffield meta-analyzed 29 experimental studies of if-then planning in people with a DSM-IV or ICD-10 diagnosis or scores above a clinical cutoff. Excluding one very large outlier, forming implementation intentions produced a large effect on goal attainment (d+ = 0.99, k = 28, N = 1,636). Those were goal-attainment studies rather than trials of therapy homework, so treat the jump as a plausible mechanism and not a demonstrated result. It is still the cheapest change on this list. "After you close the laptop on Thursday" beats "sometime this week."

3. Send it out of the room in writing, in the client's words. Helbig and Fehm asked 77 CBT therapists to recall two clients each and describe both the tasks and the trouble. Problems during assignment and during completion occurred regularly. Compliance was positively associated with three things: the client's motivation for therapy, outcome later in therapy, and the provision of a written note or homework sheet. An older crossover study of 30 participants found written prescriptions produced better recall of assignments and better adherence than verbal ones. Neither study shows the note caused the compliance. Both point at the same fifteen seconds.

4. Say out loud what you will do with it. A client who does not know why they are writing writes for you, and then hands in the polished version. Tell them the use: "I want to see whether the tight chest shows up before the phone calls or after."

5. Open with it next session, including the weeks it did not happen. Cronin and colleagues describe designing, planning, and reviewing between-session work as one integrated in-session process tied to the case conceptualization, rather than a task bolted to the end of the hour. That paper is clinical guidance built on case material, not a trial. Its practical claim is easy to test in your own room: if you never ask, the client learns the assignment was optional, and they are right.

Two screens before you recommend writing at all

Start with brooding. In a randomized trial of 90 adults who completed a writing task after a recent marital separation, participants scoring one standard deviation above the mean on rumination reported worse intrusive distress after expressive writing than after control writing, at three months (b = .59, SE = .17, t(84) = 3.36, p = .001) and again at eight (b = .64, SE = .20, t(74) = 3.16, p = .002). Participants scoring one standard deviation below the mean showed no difference across any condition at either follow-up (p = .97 and p = .69). One population, and one protocol of 20 minutes a day by hand across three days, so read it narrowly. For a client who broods, it is a reason to keep the writing short and structured, to read entries together rather than letting six weeks pile up unread, and to say the stop rule out loud.

The stop rule is the second screen, and the client has to be able to run it in ten seconds while distressed: read the last thing you wrote. Same three sentences, same charge? Stop, and bring it in.

Writing between sessions is not a containment plan. For a client at risk, it sits behind the safety plan, not in front of it.

What does the research actually say about therapy homework?

Homework survives meta-analysis, with the usual caveats about what kind of evidence you are holding.

The causal end is the strongest and the oldest. Kazantzis, Whittington and Dattilio analyzed 46 studies (N = 1,072) and reported a pre-post effect of d = 0.63 for control conditions against d = 1.08 for therapy conditions with homework. Restricting the analysis to controlled studies that contrasted the same therapy with and without homework gave a pooled d = 0.48 favoring homework. They found no evidence of outlier or publication bias effects.

The figure clinicians quote more often is correlational and needs handling. Mausbach and colleagues pooled 23 studies and 2,183 clients published since 2000 and found compliance related to outcome at r = .26 (95% CI .19 to .33), robust across target symptoms. A correlation between compliance and outcome runs in both directions at once. Clients who are getting better have more capacity to do the homework. That same paper found effect sizes varied by who rated compliance and when, which makes part of it a measurement story rather than a clinical one.

Quality is where the interesting result sits. Kazantzis and colleagues meta-analyzed 17 CBT studies (N = 2,312) and found homework relations with outcome at posttreatment for both quantity (g = 0.79, 95% CI 0.57 to 1.02, k = 15) and quality (g = 0.78, 95% CI 0.03 to 1.53, k = 3). At follow-up the quality estimate rose to g = 1.07 while quantity fell to g = 0.51. Read the k values before you get excited: three studies, and a confidence interval that nearly touches zero. The 2021 anxiety and OCD sample points the same way with a tighter design, which is why the direction is worth acting on while the numbers are not worth quoting.

None of this literature tested handing a client an app. Kazantzis and colleagues also reviewed adherence assessment across 2011 to 2016 and concluded that a large component of the adherence process still goes unmeasured, with client effort, engagement, and treatment beliefs sitting outside most instruments. The honest summary: assign homework, watch how it gets done more than how much gets done, and treat the size of the effect as unsettled.

What an experienced clinician assigns, and what comes back undone

G. Scott Graham is a therapist and substance abuse counselor with 40 years of experience, and a member of the Rosebud Practitioner Program who earns from referrals. He works with a caseload of roughly 20 remote clients, and he recommends journaling with Rosebud to essentially all of them. His method is the part worth stealing regardless of tool: he and each client co-build a persona for them to write to, translating that client's goals into the prompts the journal will ask. The client is not handed a generic exercise. They are handed one that already knows what they are working on, because they helped write it.

Forty years in, he does not decide in advance which clients are the writing type. Everybody gets the offer, and he tailors it inside the persona. You lose most between-session work at that earlier step, before you write a single prompt: you decide in the room that this particular client will not do it, and you stop offering without ever saying so out loud.

What comes back undone is more consistent than what comes back done. Multi-part worksheets. Anything daily that runs to a whole set of questions rather than one or two. Assignments that require a fresh decision about timing. Anything the client reads as a test they can fail, which is most things that arrive on a form with your letterhead on it.

Where a journaling app fits

Rosebud can be one option practitioners offer clients to help carry the work of a session into daily life. Instead of sending someone home with a static worksheet or one-time exercise, it responds to what they write, remembers relevant context, and personalizes its follow-up over time.

Rosebud aligns with the principles above: keep between-session work small and specific, put it in the client's own words, connect it to what matters in their daily life, and bring what they notice back into the next session. Clients can ask Rosebud to draw from the frameworks or modalities they use, while its personalized responses help turn a one-time assignment into an ongoing practice of noticing patterns and applying the work of therapy.

You get no view into any of it. Rosebud has no clinician dashboard and sends you no entry or summary, so what reaches your room is whatever the client chooses to bring. Treat that as a treatment-planning question rather than a settings one.

Rosebud isn't designed to replace therapy; it's a complementary tool that extends the impact of your work beyond the session, with guided journals grounded in IFS, CBT, and motivational interviewing rather than a general-purpose chatbot.

If you're interested in evaluating Rosebud before recommending to clients, you can apply to the Rosebud Practitioner Program to try it for yourself free for 60 days.

References

1. Kazantzis, N., Whittington, C., & Dattilio, F. (2010). Meta-Analysis of Homework Effects in Cognitive and Behavioral Therapy: A Replication and Extension. Clinical Psychology: Science and Practice, 17(2), 144-156. https://research.monash.edu/en/publications/meta-analysis-of-homework-effects-in-cognitive-and-behavioral-the/
2. Mausbach, B. T., Moore, R., Roesch, S., Cardenas, V., & Patterson, T. L. (2010). The Relationship Between Homework Compliance and Therapy Outcomes: An Updated Meta-Analysis. Cognitive Therapy and Research, 34(5), 429-438. https://pubmed.ncbi.nlm.nih.gov/20930925/
3. Kazantzis, N., Whittington, C., Zelencich, L., Kyrios, M., Norton, P. J., & Hofmann, S. G. (2016). Quantity and Quality of Homework Compliance: A Meta-Analysis of Relations With Outcome in Cognitive Behavior Therapy. Behavior Therapy, 47(5), 755-772. https://pubmed.ncbi.nlm.nih.gov/27816086/
4. Conklin, L. R., Curreri, A. J., Farchione, T. J., & Barlow, D. H. (2021). Homework Compliance and Quality in Cognitive Behavioral Therapies for Anxiety Disorders and Obsessive-Compulsive Disorder. Behavior Therapy, 52(4), 1008-1018. https://pubmed.ncbi.nlm.nih.gov/34134818/
5. Helbig, S., & Fehm, L. (2004). Problems With Homework in CBT: Rare Exception or Rather Frequent? Behavioural and Cognitive Psychotherapy, 32(3), 291-301. https://www.cambridge.org/core/journals/behavioural-and-cognitive-psychotherapy/article/abs/problems-with-homework-in-cbt-rare-exception-or-rather-frequent/0C47B89190ED53495A5223D38216B45A
6. Cox, D. J., Tisdelle, D. A., & Culbert, J. P. (1988). Increasing adherence to behavioral homework assignments. Journal of Behavioral Medicine, 11(5), 519-522. https://pubmed.ncbi.nlm.nih.gov/3236383/
7. Toli, A., Webb, T. L., & Hardy, G. E. (2016). Does forming implementation intentions help people with mental health problems to achieve goals? A meta-analysis of experimental studies with clinical and analogue samples. British Journal of Clinical Psychology, 55(1), 69-90. https://pubmed.ncbi.nlm.nih.gov/25965276/
8. Cronin, T. J., Lawrence, K. A., Taylor, K., Norton, P. J., & Kazantzis, N. (2015). Integrating between-session interventions (homework) in therapy: the importance of the therapeutic relationship and cognitive case conceptualization. Journal of Clinical Psychology, 71(5), 439-450. https://pubmed.ncbi.nlm.nih.gov/25809713/
9. Kazantzis, N., Brownfield, N. R., Mosely, L., Usatoff, A. S., & Flighty, A. J. (2017). Homework in Cognitive Behavioral Therapy: A Systematic Review of Adherence Assessment in Anxiety and Depression (2011-2016). Psychiatric Clinics of North America, 40(4), 625-639. https://pubmed.ncbi.nlm.nih.gov/29080590/
10. Sbarra, D. A., Boals, A., Mason, A. E., Larson, G. M., & Mehl, M. R. (2013). Expressive Writing Can Impede Emotional Recovery Following Marital Separation. Clinical Psychological Science, 1(2), 120-134. https://pubmed.ncbi.nlm.nih.gov/25606351/