“Therapy Isn’t Working”: What to Try Before You Quit

11 min read · Fact-checked

It’s session eleven. You know the drill now: the walk in, the same chair, the same opening question, the same forty-eight minutes. Your therapist is kind. You like them. And on the drive home you keep doing the same maths, which is eleven times the fee, plus eleven Tuesdays, against a life that feels roughly the same as it did in August.

Before you cancel, it’s worth knowing that “therapy isn’t working” is not one problem. It’s four, and they look almost identical from the inside while needing completely different responses. Here’s how to tell them apart, what the research says about each, and how to know when stopping is the right call rather than a failure.

First, get an actual measurement

Almost nobody does this, and it changes the entire conversation.

Memory is a poor instrument for tracking mood, because the thing you’re trying to measure distorts the thing you’re measuring with. On a bad week, a bad week feels like every week. This is why clinicians increasingly use brief symptom questionnaires between sessions rather than asking “how have things been?”

The evidence for doing so is strong. Lambert and colleagues reviewed the research on routine outcome monitoring, where progress is measured session by session and fed back to the clinician. Two-thirds of studies found monitored therapy outperformed the same practitioners’ usual care, and among clients flagged as heading for a poor outcome, feedback reduced deterioration and nearly doubled rates of clinically significant improvement (Lambert et al., 2018).

You can approximate this yourself. Pick a short standard measure, score it every two weeks, and keep the numbers somewhere you can see them side by side. A therapist may already do this. If not, asking for it is a reasonable request and a good therapist will welcome it.

When to start asking the question at all

Judging too early is its own failure mode. But there’s a defensible window.

Beard and Delgadillo reviewed the research on early response and found that symptom change in the first few sessions predicts how depression and anxiety treatment turns out overall (Beard & Delgadillo, 2019). That finding gets misread constantly, so be careful with it: it means early movement is informative, not that its absence is a verdict. Plenty of people who start slowly finish well.

What it does justify is a review point. Somewhere around sessions four to six, with a couple of data points in hand, is the moment to ask whether anything needs to change. Earlier than that and you’re reading noise. Much later and you’ve spent months finding out.

Version one: the wrong fit

The signs here are relational rather than clinical. You edit yourself before you speak. You perform insight you don’t feel. You’ve stopped bringing the thing you actually think about, because raising it feels like more work than not raising it.

This is the most common version, and it’s also the most fixable, because the fix often isn’t switching. It’s saying the awkward thing out loud in the room. Raising a problem with your therapist tends to improve outcomes rather than damage them, and how they respond tells you almost everything you need to know. Our guide to therapist red flags and the signs of a good one has the script for that conversation, plus the distinction between a fit problem and a genuine ethical one.

If you raise it and nothing changes, that’s your answer. Fit problems that survive being named are usually real.

Version two: the wrong approach

Different problems respond to different methods, and starting in the wrong one costs months.

The clearest evidence that matching matters comes from a cluster randomised trial across four English NHS services. Delgadillo and colleagues compared stratified care, where people were matched to treatment intensity at the start using their own data, against standard stepped care, where everyone begins at low intensity and escalates only after failing. Stratified care produced better depression outcomes at modest additional cost (Delgadillo et al., 2022).

The practical read: starting with whatever was available is not the same as starting with what fits. Signals you may be in the wrong approach include doing insight-oriented work when your problem is a specific behavior you can’t change, or doing structured skills work when what keeps surfacing is old grief. Our breakdown of the major types of therapy and what each is for is the reference for that conversation, and it’s a conversation to have with your therapist rather than a reason to disappear.

Version three: the wrong dose

This one is unglamorous and probably the most common reason therapy quietly fails.

Hansen and colleagues compared what the dose-response research recommends against what happens in ordinary practice. Controlled trials saw 57.6% to 67.2% of patients improve within an average of 12.7 sessions. In a national database of more than 6,000 people in routine care, the average number of sessions attended was fewer than five, and the improvement rate was about 20% (Hansen et al., 2002). Their review puts the consensus estimate at 13 to 18 sessions for half of clients to improve.

So if you’re at session six and fortnightly because of scheduling, you may not have a therapy problem. You may have a dose problem. The same applies to what happens between sessions: if therapy is one hour out of 168 and nothing carries over into the other 167, the hour is doing all the work alone.

Version four: it’s working and it feels terrible

The hardest one to read from the inside, because the evidence of progress and the evidence of harm look the same for a while.

The best data here come from trauma-focused treatment, where the fear of making people worse has historically been strongest. Larsen and colleagues tracked symptom exacerbations across cognitive processing therapy and prolonged exposure. Between 14.7% and 28.6% of participants had a temporary symptom increase depending on the treatment. Those spikes were not associated with dropping out, and were unrelated to how people ended up: those who exacerbated still showed clinically significant improvement by the end (Larsen et al., 2016).

Feeling worse partway through is not, by itself, evidence that something has gone wrong. What separates this version from real harm is direction and shape. Productive discomfort is topic-specific, it settles between sessions, and you can point at what you’re working on. Harm is diffuse, it doesn’t settle, and you feel smaller as a person rather than raw about a subject.

The conversation that covers all four

You don’t have to diagnose which version you have on your own. That’s a legitimate thing to hand to your therapist, and a good one will take it seriously rather than treat it as a complaint.

Something like: “I want to check in on how this is going. I don’t feel like much has shifted since we started, and I’d rather say that now than quietly drift. Can we look at what we’re working on and whether the approach still makes sense?”

That sentence does several things at once. It surfaces a fit problem if there is one, it opens the approach question, it invites a dose conversation, and if the answer is that you’re mid-way through something hard, it gives your therapist the chance to say so.

When switching is the right move

Sometimes it is. If you’ve raised it, given it a few more sessions, and nothing has changed, switching is a reasonable and common decision rather than a dramatic one.

Practically, the friction is usually logistical rather than emotional: waitlists, insurance, starting your history again. Platform-based services tend to be the fastest route to a different provider without repeating the whole search, which is one of the things the BetterHelp and Talkspace comparison covers. If you’d rather go back to a local search with better filters this time, our complete guide to finding the right therapist is built for exactly that second attempt, and you’ll be much faster at it than you were the first time.

And sometimes stopping is right

This deserves saying plainly, because most articles on this topic won’t.

Not every ending is a failure to fix. Around one in five people leave therapy before completing it (Swift & Greenberg, 2012), and that figure includes people who got what they came for, people whose circumstances changed, and people for whom the timing was wrong. Therapy asks for money, hours and emotional bandwidth, and there are periods in a life when one of those genuinely isn’t available.

Stopping for now is a different decision from concluding that therapy doesn’t work for you. The first is a scheduling call. The second is a much bigger claim, and the research on what therapy does and doesn’t achieve is a better basis for it than one course with one person.

The takeaway

Eleven sessions and no change is real information, and it’s worth acting on. It just isn’t yet a conclusion about therapy, or about you.

Work through it in order: get a measurement, pick a review point, then ask which of the four versions you’re in. Most people who feel stuck are in the fit or dose version, and both respond to a conversation rather than an exit. If you’ve had that conversation and nothing moved, you now have something you didn’t have before, which is a clear reason to change one variable rather than abandon the whole project.

And if the answer is that now isn’t the time, that’s allowed too. The door stays open.

Frequently asked questions

Sources

1. Lambert MJ, Whipple JL, Kleinstäuber M. (2018). Collecting and delivering progress feedback: A meta-analysis of routine outcome monitoring. Psychotherapy, 55(4), 520-537. (Two-thirds of studies favored monitored therapy over the same practitioners’ usual care; feedback reduced deterioration and nearly doubled clinically significant improvement among clients predicted to have a poor outcome.)
2. Beard JI, Delgadillo J. (2019). Early response to psychological therapy as a predictor of depression and anxiety treatment outcomes: A systematic review and meta-analysis. Depression and Anxiety, 36(9), 866-878. (Symptom change in the early sessions predicts overall treatment outcome.)
3. Delgadillo J, Ali S, Fleck K, et al. (2022). Stratified care vs stepped care for depression: A cluster randomized clinical trial. JAMA Psychiatry, 79(2), 101-108. (Thirty clinicians across four NHS services; matching people to treatment intensity at baseline outperformed escalating intensity over time, at modest additional cost.)
4. Larsen SE, Wiltsey Stirman S, Smith BN, Resick PA. (2016). Symptom exacerbations in trauma-focused treatments: Associations with treatment outcome and non-completion. Behaviour Research and Therapy, 77, 68-77. (Temporary symptom increases occurred in 14.7% to 28.6% of participants depending on treatment, were unrelated to non-completion, and did not predict worse outcomes.)
5. Hansen NB, Lambert MJ, Forman EM. (2002). The psychotherapy dose-response effect and its implications for treatment delivery services. Clinical Psychology: Science and Practice, 9, 329-343. (Trials: 57.6% to 67.2% improved within an average of 12.7 sessions. Routine care: fewer than five sessions attended on average across 6,000+ patients, improvement rate about 20%. Consensus estimate 13 to 18 sessions for half of clients to improve.)
6. Swift JK, Greenberg RP. (2012). Premature discontinuation in adult psychotherapy: A meta-analysis. Journal of Consulting and Clinical Psychology, 80(4), 547-559. (669 studies, 83,834 clients: weighted dropout rate 19.7%.)