"Therapy isn't working" is one sentence covering at least eight different situations, and they do not have the same answer. Not enough therapy yet is one. The wrong kind of therapy for your particular problem is another. So is needing more care than one weekly hour can carry, and so is liking your therapist a great deal and getting nowhere with them. So is getting somewhere and being unable to feel it, which is the one people rule out first and should not. Pick the wrong explanation and you can lose six months to it.
So sort before you decide. This article covers how common a stall is, how long therapy takes before "nothing yet" starts to mean something, the eight things that usually stall it, how to tell hard-because-it-is-working apart from not-working, what to say to your therapist, and how to leave if leaving is the answer.
If you are already thinking about stopping care altogether, or the thought of stopping arrives with relief or with a plan attached, go straight to additional support, and how to reach it today near the end of this article. It has crisis numbers in it. The rest of this will keep.
One thing first. Wondering whether the money and the Tuesday afternoons are buying you anything is a reasonable question about a service you are paying for. Clinicians ask it about their own work often enough that there is a research literature on it.
Is it normal for therapy to not work?
Yes, and the numbers are sobering.
In 2002 Hansen, Lambert and Forman pooled outcome data on 6,072 patients across six US treatment settings: a national employee assistance program, two HMOs, a university counseling center, a training clinic and a state community mental health service. Everyone completed the same 45-item symptom questionnaire before every session. By the end, 14.1% had recovered, 20.9% had reliably improved, 56.8% had not moved reliably in either direction, and 8.2% had reliably got worse.
Before you read that as a verdict on therapy, look at what sits underneath it. The median patient in the sample attended three sessions. The researchers had also dropped a further 3,101 people, a third of the original sample, from the analysis because they came once and never came back.
Swift and Greenberg later put a number on leaving early across the whole field. Pooling 669 studies and 83,834 clients, they found a weighted dropout rate of 19.7%. About one client in five stops before the therapist considers the work finished.
Carry two cautions with those figures. The 2002 dataset leans on short-term, managed-care and EAP settings where somebody other than the patient caps the number of sessions, and the authors say so themselves. And "no reliable change" does not mean "no benefit"; it means a questionnaire score did not move far enough to clear a statistical threshold.
Tens of thousands of people are in the same position you are, asking the same question, and most of them are not asking it because something is wrong with them.
How long does therapy take to work?
Reviewing the dose-response research, Hansen and colleagues report a consensus figure of 13 to 18 sessions for 50% of patients to improve. In tightly controlled trials of manualized treatments, patients averaged 12.7 sessions, and between 57.6% and 67.2% responded depending on how strictly the researchers defined response.
Set that next to a median of three sessions in ordinary practice. Half the people in that naturalistic sample received roughly a quarter of the dose the research associates with the job getting done, which explains a large share of the disappointing outcomes on its own.
Time to improvement also depends on what you brought. In one study summarized in that same review, half of patients with acute distress had responded by about session 5, half of those with chronic distress by about session 14, and half of those working on long-standing patterns of relating by about session 104. If what you are in therapy for is a decades-old way of getting close to people, a nine-month plateau is a different event than it would be for a panic problem.
The table below is orientation, not a rule. It combines those benchmarks with the ordinary shape of a course of therapy, and no study has tested it as a schedule.
| Where you are | What is reasonable to expect |
|---|---|
| Sessions 1 to 5 | Too early to judge the outcome. Judge fit instead: can you say true things in that room, and does your therapist track what you said last time? |
| Sessions 6 to 12 | You should be able to name what you are working on and what would count as better. Symptom change may not have started. |
| Sessions 13 to 20 | The window in which half of patients improve. If nothing has shifted in your symptoms, your behavior, or your understanding of the problem, raise it directly. |
| Session 20 and beyond with nothing moving | Ask for an explicit review: the formulation, the plan, the expected timeline, and what would trigger a change of approach or a referral. |
Eight things that stall therapy
Most stalls come down to one of these. Read the left column for the one that sounds like your week.
| What it looks like from the inside | What is usually going on | The first thing to try |
|---|---|---|
| Nothing feels different, and if someone asked you to say what has changed you could not | You may be judging three months on how today feels | Check a record, not a memory (why that matters) |
| Nothing has changed and you are on session four | Not enough therapy yet | Set a review date at session 12 and hold off on deciding |
| Sessions feel good and by Wednesday nothing is different | The work is not surviving the week | Carry something out of the room and back into it (see below) |
| The first fifteen minutes go on catching up | You spend sessions re-establishing context | Arrive with two or three things already chosen |
| You circle the same topic every week and it never resolves | You and your therapist may not have agreed a goal, or the alliance has frayed | Ask to name the goal out loud, in one sentence |
| Your problem has a named first-line treatment and you are not getting it | Modality mismatch | Ask what modality you are in, and why that one for this |
| You are sleeping badly, drinking more, or something outside the room is doing damage | You are asking therapy to outrun something it cannot outrun alone | Put the outside thing on the agenda, and involve a physician or prescriber if relevant |
| You get through the week and lose most of the ground again before the next session, or sessions keep going on stabilizing you | Weekly outpatient therapy may be too low an intensity for what you are carrying | Ask whether weekly is the right level for this, and what more would look like: twice-weekly sessions, a group, an intensive outpatient program, or a psychiatric review |
Note the direction of that last row. Some of these point toward less therapy or different therapy, and one of them points toward more. A reader working down this table who recognizes the last row should read it as a reason to ask for more support, not as a reason to conclude that therapy does not work for them.
None of these is a diagnosis of your therapist. Several are shared problems, and a good therapist will be relieved you raised it.
What happens between sessions is often where it stalls
A weekly fifty-minute session, set against sixteen waking hours a day, is about three quarters of 1% of your week. The pattern you are in therapy for runs during the other 99%, in rooms your therapist will never see.
You can miss this failure from inside the session. You have a good hour, you feel understood, you leave. Nothing about Thursday changes. Six months of good hours can add up to no change at all, and from the chair it still looks like therapy that is going well.
That ratio is also why the fourth row of the table above says what it does. If the first fifteen minutes go on catching up, you spend close to a third of the session reconstructing a week you already lived, from memory, in whatever order it comes back to you. Choosing two or three things before you walk in is the cheapest fix on the table: it takes a few minutes, it needs nobody's agreement but yours, and you can do it before your next appointment instead of after a conversation about whether to.
Both of those rows ask for the same move: you carry something out of the room and back into it. The mechanics are in our guide to journaling between therapy sessions, which is about selecting three things to bring, not about writing more.
How do you know if therapy is working?
Good therapy is often uncomfortable, so discomfort tells you nothing on its own. Four questions separate hard-because-it-is-working from not-working better than a feeling does.
- Where does the discomfort sit? Rough in the session and steadier in the days after is a different pattern from rough in the session and rough all week with nothing new in it.
- Are the words changing? Describing the same problem with more precision than you could three months ago is movement. Producing the same sentences in the same order is not.
- Has any behavior changed? Behavior usually moves before feelings do. Something you did in the last month that you could not have done six months ago counts, even if you felt no better while doing it.
- Is there a stated goal? If neither of you can say in one sentence what you are working on, fix that before you decide anything else.
You do not have to do this by intuition alone. Lambert and Shimokawa note that 5% to 14% of clients worsen while in treatment, and that therapists fail to identify a substantial proportion of those cases. Feedback systems that collect a short client-rated score before each session, flag the people who are not on track, and hand that back to the clinician have been found in meta-analysis to roughly halve the number of clients who deteriorate. Asking your therapist whether they use a routine outcome measure is a fair question, and how they answer it tells you something either way.
Why you cannot answer that question from memory
Every one of those four questions asks you to compare now with then. You run that comparison on an instrument your current mood has already adjusted.
Ben-Zeev and Young had hospitalized depressed patients and non-clinical controls report symptoms in the moment across a week using experience sampling, then asked them to summarize the same week afterwards. The depressed group's recall skewed negative for loss of pleasure, sadness, confusion and suicidality, and stayed close to accurate for helplessness, detachment and self-control. Controls showed the opposite bias on several of those. Both groups recalled more tension, difficulty concentrating, guilt and fear than they had reported while living the week.
Solhan and colleagues found something similar in outpatients with borderline personality disorder or a current depressive episode: agreement between recalled mood changes and momentary measures was poor.
Hold the bounds on that. One week, mostly inpatients, a specific symptom list. None of it proves your memory of March is wrong. It establishes something narrower and more useful: the instrument you are judging three months of therapy with has known error, and that error moves with how you feel while using it.
So on a bad Sunday you will flatten a decent April into the same grey as the last two weeks. People quit therapy that was working that way. People also stay eighteen months in therapy that was not, because a good week fresh in mind covers a lot of ground.
A written record from the weeks in question is the only thing that settles it, and it pays off in six weeks rather than on Tuesday. So start with the thing you can do sooner, which is to say it out loud in the room.
How to tell your therapist that therapy isn't working
Most people do not. In a survey of 547 psychotherapy clients, 93% said they had lied to their therapist at some point, and 72.6% had lied about at least one therapy-related topic. Among the common therapy-related lies was pretending to find therapy effective. Among the lies respondents rated most extreme in their own extent of dishonesty was not admitting to wanting to end therapy. Your instinct to manage the moment is the ordinary one, and the scripts below assume it.
The evidence points the other way. Eubanks, Muran and Safran pooled 11 studies and 1,314 patients and found that episodes in which a rupture in the alliance got repaired went with better patient outcomes. That is correlational, so it does not prove that raising your doubt causes improvement. It does mean the conversation you are dreading is a recognized part of the work with a literature behind it.
Say it early in a session, not in the last two minutes, and give your therapist something to answer:
- "I want to check something. I don't think I'm getting better, and I want to know what you're seeing."
- "Can we name what we're working on in one sentence, so I know what better would look like?"
- "What would you expect to have changed by October, and what would you do differently if it hadn't?"
A useful response looks like curiosity, a reformulation of the problem, an offer to measure something, or a frank discussion of referral. A response that should give you pause treats your doubt only as a symptom of the thing you came in with, offers no goal when you ask for one, or gets defensive.
One honest complication. In psychodynamic and analytic work, your dissatisfaction is legitimate material, and exploring it is no dodge. The question that separates the two cases is whether the exploration also produces a plan. If you have raised it three times and each conversation ends inside the exploration, name that as a pattern of its own.
If the hard part is finding the words at all, our piece on what to talk about in therapy covers how to choose and open a difficult subject.
When it is the wrong therapy rather than the wrong you
Some problems have named first-line treatments, and general talk therapy is not among them.
For OCD, the International OCD Foundation says of traditional talk therapy that "there is no research evidence that it is effective in treating OCD," and that "it is important to try ERP or medication first, as these are the types of treatment that have been shown through extensive research to be the most effective for treating OCD." Note that the superlative covers both, so medication is on the table alongside exposure and response prevention, and that is worth carrying into the conversation. You can spend months in insight-oriented conversation about your intrusive thoughts, feel understood every week, and get no better.
For PTSD, the VA's National Center for PTSD lists cognitive processing therapy, prolonged exposure and EMDR as the trauma-focused therapies with the most research support, and notes that other recommended therapies "do not have as much research support."
Take this as a reason to ask a question, not to diagnose yourself off a web page. "What modality are we working in, and why that one for this problem?" has a reasonable answer available, and your therapist should be able to give it.
Some evidence sits behind that question too. Crawford and colleagues surveyed 14,587 people who received psychological treatment across 184 services in England and Wales; 5.2% reported lasting bad effects from it. People who were unsure what type of therapy they had received were more likely to report those effects, and people who said they had been given enough information before treatment started were less likely to. The survey is cross-sectional and self-reported, so it cannot show that information prevents harm. Knowing what you are in is still worth the question.
How to switch therapists
Sometimes the answer is a different clinician. Fit is not a soft consideration: Flückiger and colleagues pooled 295 studies covering more than 30,000 patients and found a consistent association between the therapeutic alliance and outcome, at r = .278. Researchers still argue about the direction of that relationship, since early improvement can build the alliance as easily as the reverse. Even so, "I don't click with them" is no trivial complaint about a service whose main instrument is a relationship.
Weigh it against the cost. Switching resets you to session one with a new person, and the timeline above starts again. If the problem is a fixable one, the same therapist plus an explicit change of plan gets you there faster.
If you do leave:
- You can end at any time, without a justification. An email is enough. You do not need a good reason, and you do not need to deliver it in person.
- Ask for one closing session if you can face it. Useful things to request: their formulation of the problem, what has been tried, what has not, and what a next clinician should know.
- Ask for a referral, ideally a named modality. "Who would you send someone to for this?" is a question most therapists can answer well.
- Line up the next thing before you end the current one. A gap of a few weeks is fine. A gap of six months usually was not the plan.
- Tell any prescriber involved. If a psychiatrist or physician is managing medication, they need to know your therapy has changed.
If the last episode of care went badly, tell the next clinician in the first session. A vague bad feeling about therapy is hard for anyone to work with. A specific account of what happened, what you asked for, and what you got instead is something a new therapist can use.
How to keep a record you can actually check
Writing has its own therapeutic case, which we cover in therapeutic journaling. The reason a record settles the question is duller than that: it has dates on it. Four things become checkable that were not before.
- Frequency, not intensity. How many bad days in the last month, against how many in March. What you feel tells you about today, and what you count tells you about the trend.
- Floor and ceiling. Your worst day this month next to your worst day three months ago. From inside a bad week, a plateau and a decline feel identical; on a page they do not.
- What happened. The specific incident, in the words you used at the time, before it got smoothed into a summary.
- Something to hand over. "I don't think this is helping" is hard for a therapist to work with. "Here are four Tuesdays" is not.
You can point the record at the right question with one line. At the end of your next session, write down what you are working on in your therapist's words, not your summary of them, and put that line at the top of wherever you write. Without it you write toward whatever was loudest that week. With it at the top, you read the month back against the goal you agreed on instead of against your mood.
Rosebud is one way to keep that record. It holds every entry and surfaces patterns across weeks, including a weekly report, so the comparison does not depend on you being in the right state to go looking for it. That is the whole contribution, and the edges are worth stating. Rosebud is a self-reflection tool, not a treatment and not a substitute for professional care. It does not give your therapist access to your entries; anything that reaches the room, you bring.
Rosebud only knows what you put into it, so the month it shows you is the month you wrote down. Skip the flat weeks and it reads better than the year did; skip the worst ones and it reads worse. And nothing arrives from the other direction: the goal you agreed on in the room does not turn up inside your week unless you carry it there, which is what the line above is for. That loop is not closed, in our product or in anyone else's.
None of this needs an app. A paper notebook holds dates too, and the only real cost is that you will do the reading back by hand. Six weeks from now, on the day you most doubt everything, you want something with dates on it to check.
Additional support, and how to reach it today
Stopping this therapy and stopping care are two different decisions, and people make them together more often than they should. The stall you are in points at a specific fixable thing far more often than it points at treatment being pointless for you.
If the thought of stopping has a plan attached to it, or arrives with relief, or comes alongside thoughts of ending your life, reach someone today, and do not withdraw. You can contact:
- US: 988 Suicide & Crisis Lifeline (call or text 988)
- UK: Samaritans on 116 123
- International: befrienders.org
And if you are getting worse rather than standing still, tell a clinician: your therapist, or a different one if you would rather not raise it with them. You do not have to be sure, and you do not have to put it well.
Nothing in this article is a reason to change a medication. Stopping some psychiatric medications abruptly carries real risk, and that decision belongs to you and the person who prescribed it.
What to do next, in order
- Ask the question in the room. Early in your next session, not in the last two minutes: "I don't think I'm getting better, and I want to know what you're seeing." It needs no tool and no preparation, and your therapist has most likely seen this stall before and can name it faster than you can.
- Start a record tonight. Three lines: the date, a number out of ten, one sentence about what happened. Write on the days you remember, including the days when nothing happened. In six weeks you will have something to check instead of something to recall, and a pattern to bring into a session instead of a verdict.
- Name the date you will decide on. Pick the session where you and your therapist review this together, and say the number out loud: session 12 if you are early, six weeks from tonight if you are further in. Then leave the decision alone until you get there.
If a notebook is not going to happen, start the record in Rosebud. It needs the same two things from you either way: a date on every entry, and an entry on the days you feel nothing worth recording.
References
- Hansen, N. B., Lambert, M. J., & Forman, E. M. (2002). The Psychotherapy Dose-Response Effect and Its Implications for Treatment Delivery Services. Clinical Psychology: Science and Practice, 9(3), 329-343. https://onlinelibrary.wiley.com/doi/abs/10.1093/clipsy.9.3.329
- Swift, J. K., & Greenberg, R. P. (2012). Premature Discontinuation in Adult Psychotherapy: A Meta-Analysis. Journal of Consulting and Clinical Psychology, 80(4), 547-559. https://pubmed.ncbi.nlm.nih.gov/22506792/
- Lambert, M. J., & Shimokawa, K. (2011). Collecting Client Feedback. Psychotherapy, 48(1), 72-79. https://pubmed.ncbi.nlm.nih.gov/21401277/
- Ben-Zeev, D., & Young, M. A. (2010). Accuracy of Hospitalized Depressed Patients' and Healthy Controls' Retrospective Symptom Reports: An Experience Sampling Study. The Journal of Nervous and Mental Disease, 198(4), 280-285. https://pubmed.ncbi.nlm.nih.gov/20386257/
- Solhan, M. B., Trull, T. J., Jahng, S., & Wood, P. K. (2009). Clinical Assessment of Affective Instability: Comparing EMA Indices, Questionnaire Reports, and Retrospective Recall. Psychological Assessment, 21(3), 425-436. https://pubmed.ncbi.nlm.nih.gov/19719353/
- Eubanks, C. F., Muran, J. C., & Safran, J. D. (2018). Alliance Rupture Repair: A Meta-Analysis. Psychotherapy, 55(4), 508-519. https://pubmed.ncbi.nlm.nih.gov/30335462/
- Flückiger, C., Del Re, A. C., Wampold, B. E., & Horvath, A. O. (2018). The Alliance in Adult Psychotherapy: A Meta-Analytic Synthesis. Psychotherapy, 55(4), 316-340. https://pubmed.ncbi.nlm.nih.gov/29792475/
- Crawford, M. J., Thana, L., Farquharson, L., Palmer, L., Hancock, E., Bassett, P., Clarke, J., & Parry, G. D. (2016). Patient Experience of Negative Effects of Psychological Treatment: Results of a National Survey. The British Journal of Psychiatry, 208(3), 260-265. https://pubmed.ncbi.nlm.nih.gov/26932486/
- Blanchard, M., & Farber, B. A. (2016). Lying in Psychotherapy: Why and What Clients Don't Tell Their Therapist About Therapy and Their Relationship. Counselling Psychology Quarterly, 29(1), 90-112. https://www.tandfonline.com/doi/full/10.1080/09515070.2015.1085365
- International OCD Foundation. Exposure and Response Prevention (ERP). https://iocdf.org/about-ocd/ocd-treatment/erp/
- U.S. Department of Veterans Affairs, National Center for PTSD. Talk Therapy for PTSD. https://www.ptsd.va.gov/understand_tx/talk_therapy.asp
- 988 Suicide & Crisis Lifeline. https://988lifeline.org/
