Therapist Insights / How Therapy Works
When therapy doesn't work, name the failure first.
Two courses of therapy that went nowhere is not evidence that you are untreatable. It is evidence that something specific went wrong, and the something is usually identifiable: the wrong treatment for the presentation, a clinician who was a poor fit, a rupture nobody repaired, or a condition sitting underneath that was never assessed.
THE QUICK TAKEAWAY
Therapy fails often enough that the outcome literature has language for it. Across 228 randomized trials of psychotherapy for depression, about 41% of people responded at roughly two months and only about a third remitted, which means most people in those trials did not meet response criteria on that timetable. Non-response is ordinary. What it is not is a finding about the person, and four situations that get filed under the same sentence have four different fixes: the treatment genuinely did not work, the treatment was wrong for the presentation, the clinician was wrong for the person, or something treatable underneath was never assessed.
§01 / 09 / Definition
When therapy doesn't work, name the failure.
Therapy that is not working is a clinical problem with a differential, not a verdict on the person in the chair. High achievers who have tried therapy before usually need the failure named before anything changes: the wrong treatment, the wrong clinician, a missed condition underneath, or a course that never ran long enough to count as a course.
Most people who say therapy did not work for them are describing something they never got a straight answer about. They went, they talked, they left, and nothing measurable changed. Nobody sat down at week eight and said out loud that the plan was not producing what it should, so the course drifted until it quietly stopped. That drift is the actual failure, and it is a failure of method rather than a failure of the person. The literature is unambiguous that a meaningful proportion of people do not respond to a given course of treatment, and that a smaller proportion get worse during it. Both of those are known, expected, measurable outcomes with established responses. Neither is a reason to conclude that treatment is impossible for you, and neither is a reason to spend another year finding out slowly.
Five things "therapy is not working" can actually mean
The treatment ran and did not work
An adequate course of an appropriate treatment was delivered and the symptoms did not move. That happens, it is measurable, and the response is to change the treatment rather than to repeat it for another year.
The treatment was wrong for the presentation
Supportive weekly talking is not the same intervention as exposure and response prevention, and neither is trauma-focused processing. A person with untreated obsessive-compulsive symptoms can spend two years in general therapy and never receive the protocol with the strongest evidence for what they actually have.
The clinician was wrong for you
Fit is not a soft variable. The strength of the working alliance is one of the most reliable predictors of outcome in the entire psychotherapy literature, and it is measured from the client's side. A clinician who is excellent for someone else can be the wrong clinician for you.
Something underneath was never assessed
Undiagnosed ADHD, an autistic profile recognized late, a sleep disorder, thyroid or other medical contributors, alcohol use that never came up, or a trauma history nobody asked about will all defeat treatment aimed at the wrong target. Assessment is the step that gets skipped most often.
The course never actually ran
Six appointments spread over nine months is not a course of treatment. Where the calendar keeps winning, what failed was the schedule rather than the therapy, and the fix is a format that can realistically be completed. Where the exhaustion itself is what keeps eating the appointments, how burnout gets addressed when stepping back from the role is not an option is the more useful question than whether therapy works.
▶ Research
In a 2021 meta-analysis of 228 randomized trials of psychotherapy for depression, published in Acta Psychiatrica Scandinavica, the overall response rate at about two months was 41%, against 17% for care as usual and 16% for waitlist. Roughly one third of patients remitted, compared with 7% to 13% in control conditions, and deterioration occurred in 5% of psychotherapy patients against 12% to 13% of controls. The authors state plainly that more than half of patients receiving therapy do not respond and only one third remit, and conclude that better treatments and better treatment strategies, including sequencing and combining, are needed.1
What the outcome literature actually says
Non-response is common
A 41% response rate at two months means the majority of people in those depression trials had not responded yet on that timetable. Anyone who tells you therapy works for everyone is not describing the evidence.
Deterioration is real and it is smaller than the alternative
About 5% of psychotherapy patients in that same meta-analysis deteriorated. That is a real number worth knowing, and it was lower than the 12% to 13% who deteriorated in the control conditions.
An average is not a prognosis
Aggregate response rates describe trial populations, not the person reading this. They are useful for one thing only: establishing that treatment non-response is a known event with known responses, rather than a private verdict about you.
Three fits, and any one of them can be wrong
Whether therapy works is usually treated as one question. It is closer to three, and they fail independently, which is why a single verdict on the whole enterprise is almost always the wrong conclusion to draw.
The treatment
Whether the approach in the room is the one indicated for the problem in front of it. That is a question with an answer, and the answer should be stated in the first few sessions rather than inferred a year later.
The clinician
Whether this specific person and this specific client can build and hold a working alliance. Being unable to is not a character judgment on either side, and it is the most common recoverable reason a course of therapy fails.
The target
Whether what is being treated is the thing driving the impairment. When the target is wrong, everything downstream is competent work aimed at the wrong problem, and it will look exactly like therapy that does not work.
§02 / 09 / Telehealth
"Therapy didn't work for me" is four sentences.
Therapy didn't work for me compresses four separate findings into one sentence, and each has a different response. CEREVITY treats a poor clinician match as a matching problem rather than a motivation problem, which is the reason a nationwide network matters here: the set of clinicians a person can be matched to is not limited to the ones within driving distance of their office.
The pool is national, not local
Matching improves when the choice is not limited to clinicians within driving distance. CEREVITY works as a nationwide network of independent licensed clinicians delivering secure telehealth across all 50 states, so a specific presentation can be matched to a clinician who treats it routinely.
A re-match is not starting from zero
The history of two failed courses is clinical information, not baggage. What was tried, for how long, at what frequency, and where it stalled shortens the next assessment considerably rather than lengthening it.
The specialist presentation gets a specialist
Trauma processing, obsessive-compulsive symptoms, performance and high-stakes anxiety, and late-identified neurodevelopmental profiles each have their own literature and their own clinicians. Someone whose worst hours are the ones before a board meeting needs therapy for the hours before something that matters, not a general course of supportive weekly sessions. Generalist care is not a lesser thing, but it is the wrong thing when a protocol exists.
§03 / 09 / Mechanism
Therapy for people who hate therapy.
Therapy for people who hate therapy is usually alliance work before it is anything else. The alliance is the agreement on goals, the agreement on tasks, and the bond that holds while both are negotiated, and across 295 independent studies covering more than 30,000 patients it correlates with outcome at r = .278. Hating therapy is frequently a rupture nobody named.
The alliance is the most investigated variable in psychotherapy research, and it is not the same thing as liking your therapist. A 2018 meta-analytic synthesis in Psychotherapy examined 295 independent studies covering more than 30,000 patients in face-to-face and internet-delivered treatment, and found an overall alliance-outcome association of r = .278, with an almost identical figure for internet-based work. The relation held across assessor perspectives, alliance measures, outcome measures, treatment approaches, patient characteristics and countries. Two percent of the effect sizes were negative, which is the honest part: this is a robust average, not a law.
A rupture, in the technical sense, is a deterioration in that alliance. The research literature defines it as a disagreement between patient and therapist on the goals of treatment, a lack of collaboration on the tasks, or a strain in the emotional bond between them. Ruptures show up in two recognizable shapes. Withdrawal looks like agreeing with everything, going vague, shortening answers, being pleasant and absent. Confrontation looks like irritation, challenge, complaints about the method or the fee. Both are ordinary events in real treatment. What matters is whether the clinician notices and puts it on the table.
Repair is measurable. A 2018 meta-analysis of 11 studies covering 1,314 patients found a moderate relation between rupture resolution and good treatment outcome, r = .29. The same paper examined whether training clinicians in rupture resolution improved their patients' outcomes across six studies of 276 trainees and found no significant effect, which is worth knowing before anyone sells rupture training as a fix. The practical implication for a person who has fired two therapists is narrow and useful: at least one of those endings may have been an unrepaired rupture, and naming it out loud in the third room is the single highest-value thing you can bring.
► Standard advice vs. CEREVITY's approach
Standard therapy
"Assume the discomfort means the therapy must be working"
CEREVITY
"Ask whether the discomfort is a rupture, and say so in the session"
Standard therapy
"Go quiet, agree with everything, then cancel the next three"
CEREVITY
"Name the disagreement out loud and watch what the clinician does with it"
Standard therapy
"Start over from scratch and mention nothing about the last two"
CEREVITY
"Bring both previous courses, what was tried, and where each one stalled"
| Standard insurance-based therapy | CEREVITY's specialized approach |
|---|---|
| "Assume the discomfort means the therapy must be working" | "Ask whether the discomfort is a rupture, and say so in the session" |
| "Go quiet, agree with everything, then cancel the next three" | "Name the disagreement out loud and watch what the clinician does with it" |
| "Start over from scratch and mention nothing about the last two" | "Bring both previous courses, what was tried, and where each one stalled" |
A break from the page
A third clinician is not a third failure.
A first conversation is confidential and commits you to nothing. CEREVITY is a nationwide network of independent licensed clinicians working private-pay, with no insurance claim submitted and no diagnosis sent to a payer. If you want the structure before the sales pitch, read how the clinical model is structured and bring the history of what already failed.
§04 / 09 / Cases
Common challenges we address.
The executive who has already fired two therapists
The patternTwo courses, each ended somewhere between session four and session eight. The stated reason is usually that it was not going anywhere, or that the sessions felt like paid conversation with no plan behind them. The private conclusion is that therapy is for other people, and that conclusion is now three years old.
What we addressThe first job is a structured history of both courses: what was tried, at what frequency, for how long, what the stated target was, and what happened in the last two sessions before each ended. That usually separates a wrong-method failure from a rupture that was never repaired, and it makes the third course a different course rather than a third attempt at the same one. Where the depletion is coming from the volume of judgement calls rather than from the role itself, clinical work for people who decide for a living is a different target with a different plan.
The person told they were treatment-resistant
The patternYears of treatment for anxiety or depression with partial results, a growing file of approaches tried, and a working assumption that this is simply how their brain is. Nobody has ever run a formal assessment. Executive function, sleep, alcohol, and a lifetime of compensating at high cost have never been examined as contributors rather than as symptoms.
What we addressAssessment changes the target. An ADHD profile recognized in adulthood, an autistic profile identified late, an untreated sleep disorder, or a trauma history nobody asked about will each defeat treatment aimed elsewhere, and each has its own indicated approach. If the discouragement has moved into hopelessness or thoughts of suicide, that is an immediate matter rather than a scheduling one: call or text 988 now, and treat the therapy question as the second conversation rather than the first.
§05 / 09 / Methods
Evidence-based treatment approaches.
Good clinical practice when treatment is not working is specific and unglamorous: measure, review on a schedule agreed in advance, change the plan when the measurement says to, and refer out when the presentation needs something the clinician does not do. CEREVITY clinicians set review points at the start rather than discovering after a year that nothing moved.
Measurement-based care and progress feedback
Brief validated symptom measures completed at intervals, with the scores looked at together rather than filed. A 2021 multilevel meta-analysis of 58 studies and 21,699 patients found a small but real effect of progress feedback on symptom reduction, d = 0.15, and a favorable effect on dropout. Small effects are worth having when the alternative is drifting unmeasured.
Formal psychological assessment
Structured assessment when the picture does not fit the treatment being given. Attention, executive function, learning history, autistic traits, sleep, and substance use are asked about directly rather than assumed, because a wrong target is the most expensive error available in this work.
Cognitive Behavioral Therapy (CBT)
Targeted, structured, and easy to evaluate on a schedule, which is exactly why it is a reasonable place to restart when a previous course was open-ended and unmeasured. Explicit targets make non-response visible early instead of late.
Acceptance and Commitment Therapy (ACT)
Useful where symptom reduction has been the only goal and has stalled, since the working target becomes what the person is able to do rather than how much the symptom has moved. For high achievers whose functioning is intact and whose life has narrowed anyway, that is often the more honest measure.
Trauma-focused protocols
Prolonged Exposure, Cognitive Processing Therapy and EMDR are specific procedures with specific evidence, not a general orientation. Where a trauma history was never asked about, this is frequently the treatment that was missing rather than the treatment that failed.
§06 / 09 / Investment
Understanding the investment in private-pay care.
Private-pay, nationwide, and built around discretion
At CEREVITY, our online individual therapy sessions are structured as a direct investment in your mental agility and overall well-being. The investment includes:
- Licensed mental health professional specializing in assessment-guided therapy for high achievers
- Evidence-based, one-on-one approaches proven effective for depression, anxiety, and burnout
- Flexible online scheduling including evenings and weekends
- Complete privacy with no insurance involvement or red tape
- High achievers who have tried therapy before expertise and understanding
- Outcome tracking and progress measurement
The cost of therapy that is not working going unaddressed
Consider what is at stake when therapy that is not working goes unaddressed:
What private-pay changes after two failed courses
Working outside of insurance means the length and shape of the plan are clinical decisions rather than authorization decisions, and it means no claim is submitted and no diagnosis reaches a payer. It also means the honest answer to a stalled course can be a referral out rather than a retained client. Practical questions about how starting works are collected in answers to the questions people ask first. View our current rates here: cerevity.com/our-pricing-for-therapy/.
Where to start when you have started before
Sessions are delivered by secure telehealth nationwide across all 50 states. Where a previous course died in a run of cancellations, the format is worth choosing deliberately this time rather than defaulting to the standard weekly session that already failed twice. The 3-hour intensive format exists for calendars that will not protect a weekly hour, and 90-minute sessions exist where the procedure genuinely needs the length. For people who want a clinician reachable between sessions, who this model tends to suit and who it does not is worth reading before choosing it.
§07 / 09 / Evidence
What the research shows.
The most useful number in this literature is not an effect size, it is a proportion. A 2021 meta-analysis of 228 randomized trials of psychotherapy for depression, restricted to therapies with at least ten trials, reported a 41% response rate at roughly two months against 17% for care as usual, with about one third of patients remitting and 5% deteriorating. The authors' own conclusion is the sentence worth carrying: more than half of patients receiving therapy do not respond, and only one third remit. That is not an argument against treatment, since the control conditions did considerably worse on every one of those measures. It is an argument for treating non-response as a clinical event with a name. Treatment non-response is what routine measurement exists to detect, and detecting it early is the difference between changing the plan at week eight and abandoning the whole idea at month fourteen.
► Three numbers worth knowing before session one
response rate for psychotherapy for depression at about two months, against 17% for care as usual.
Cuijpers et al., Acta Psychiatrica Scandinavica, 2021
alliance-outcome association across 295 studies and more than 30,000 patients.
Fluckiger, Del Re, Wampold and Horvath, Psychotherapy, 2018
effect of routine progress feedback on symptom reduction across 58 studies and 21,699 patients.
de Jong et al., Clinical Psychology Review, 2021
Two honest limits belong on this page. The first is that progress feedback is not a large effect. The 2021 multilevel meta-analysis that produced d = 0.15 also found no significant effect of feedback on the rate of deteriorated cases, and its authors are careful about when and why feedback helps. The case for measuring is that it makes non-response visible at week eight rather than month fourteen, not that measurement is itself a treatment. The second limit is that a missed condition is a real and common story rather than a rhetorical device. The National Institute of Mental Health puts current ADHD prevalence among United States adults aged 18 to 44 at 4.4%, with an estimated lifetime prevalence of 8.1% in that age band, and those estimates come from structured diagnostic interviews rather than from clinic records. A presentation that has never been assessed is not the same thing as a presentation that has been ruled out.
§§ / 09 / Recap
Key takeaways.
Five things to remember
- Non-response is a known event In 228 randomized depression trials, 41% responded at about two months and roughly a third remitted. Most people in those trials had not responded on that timetable, which makes non-response ordinary rather than diagnostic of the person.
- Four failures, four fixes A treatment that did not work, the wrong treatment, the wrong clinician, and a missed condition underneath all read identically from the inside and need completely different responses.
- The alliance is measurable and ruptures are repairable Alliance correlates with outcome at r = .278 across 295 studies, and resolved ruptures correlate with good outcome at r = .29. Hating therapy is often an unrepaired rupture rather than a verdict on treatment.
- Set review points before you need them Agree at the start when the plan gets reviewed and what would count as it not working. A clinician who will refer you out when the answer is no is the one worth staying with while the answer is still yes.
- CEREVITY provides this through online individual therapy nationwide, with full privacy through its private-pay concierge network and no insurance involvement.
§08 / 09 / FAQ
Frequently asked questions.
Why isn't therapy working for me?
Therapy that is not working usually has one of four causes, and they are worth separating because they have different fixes. The treatment may have been delivered adequately and simply not worked, which happens and is measurable. The treatment may have been the wrong one for the presentation, which is common where a specific protocol exists and was never used. The clinician may have been a poor fit, which is the most recoverable of the four. Or something treatable underneath, such as an undiagnosed attention or autistic profile, a sleep disorder, alcohol use or a trauma history nobody asked about, may never have been assessed. A structured review of what was actually tried, for how long, and against what target will normally tell you which one you are dealing with.
How long should therapy take before I see progress?
Progress in therapy should be a question with a date attached rather than an open-ended hope. In the depression trials that produced the widely cited response figures, outcomes were assessed at roughly two months, so a plan with no review point inside that window is a plan nobody is checking. A reasonable arrangement is to agree in the first two sessions what the target is, what measure will be used, and when the two of you will look at it together. High achievers who have tried therapy before frequently discover that the previous course had none of those three, which is why nothing was ever formally declared to have failed.
Is it normal to feel worse after starting therapy?
Distress that rises early in treatment is common and is not the same thing as deteriorating. Trauma-focused and exposure-based therapy in particular can raise distress before it lowers it, and that is an expected feature of those procedures rather than a sign of harm. Sustained deterioration is a different matter and it is real: in a meta-analysis of 228 randomized depression trials, 5% of psychotherapy patients deteriorated, against 12% to 13% in the control conditions. The distinction is made by measurement and by saying it out loud, not by waiting. If distress has moved into hopelessness or thoughts of suicide, contact the 988 Suicide and Crisis Lifeline immediately by call or text.
Should I fire my therapist or give it more time?
A decision to leave a therapist is sometimes right and often premature, and a single direct conversation in the room usually settles which. Alliance ruptures, meaning a disagreement about the goals of therapy, a breakdown in collaboration on tasks, or a strain in the bond, are ordinary events in real treatment, and a 2018 meta-analysis of 11 studies covering 1,314 patients found that resolved ruptures correlate with good outcome at r = .29. The test is what happens when you name the problem directly. A clinician who takes it seriously, adjusts the plan, and revisits it is worth staying with. A clinician who becomes defensive, or who has no answer to what is being measured and when it gets reviewed, has told you what you needed to know.
Is there a test that tells me whether my therapy is working?
No webpage can assess whether a course of therapy is working, and CEREVITY does not publish a self-scoring quiz that claims to. A score generated by a website has no clinician looking at it, no history behind it, no measurement of the thing you were actually being treated for, and no ability to distinguish a temporary rise in distress during exposure work from genuine deterioration. What does exist is measurement-based care: brief validated instruments administered and interpreted by the clinician treating you, reviewed with you on a schedule set in advance. A 2021 meta-analysis of 58 studies and 21,699 patients found progress feedback produced a small improvement in symptom reduction and a favorable effect on dropout. Ask your clinician what is being measured and when it is reviewed.
Is there therapy for people who hate therapy?
Therapy for people who hate therapy is a real request and usually a specific one. Most people who describe hating it are describing something identifiable: sessions with no agenda, a method that never matched the problem, a clinician who could not be disagreed with, or an ending that was never discussed. Structured, time-limited, target-driven work tends to suit people who found open-ended sessions intolerable, because it has explicit goals, homework between sessions, and a stated point at which it is reviewed. Being told in advance what the plan is and when it gets evaluated is not a small preference for high achievers who have tried therapy before, it is often the difference between a third course and a third failure.
What should a therapist do when treatment is not working?
Competent practice when therapy is not working has four parts and none of them is waiting. The clinician should be measuring outcomes on validated instruments rather than relying on impression, should review the plan against those measures on a date agreed in advance, should change the treatment or the target when the measures say to, and should refer out when the presentation needs a protocol or an assessment they do not provide. Referring a client onward is a normal clinical act, not an admission of defeat, and clinicians reading this rather than clients can see what referring someone to CEREVITY involves. If a clinician cannot tell you what is being measured, when it is being reviewed, and what would change their mind, that is the answer to your question.
How does your private-pay pricing structure work?
As a private-pay concierge network, we offer structured investments in your mental health without the restrictions or privacy risks of insurance. You can review our full fee schedule and specific session lengths directly on our website. While this costs more than insurance copays, it provides the flexibility, total privacy, and highly specialized care that standard options cannot offer. View our current rates here.
How do you protect my privacy?
Privacy is foundational to our network. As a private-pay network, your sessions never appear on insurance records or EOBs that could be seen by employers, boards, or family members. We use HIPAA-compliant nationwide telehealth platforms, and you can attend sessions from anywhere with a private internet connection.
§09 / 09 / Begin
Start from what already failed.
Two courses that went nowhere are the most useful clinical history you can bring to a third. CEREVITY is a nationwide network of independent licensed clinicians providing confidential private-pay care in 50-minute, 90-minute and 3-hour formats, matched to the presentation rather than to the postcode. Call (562) 295-6650 or send a private inquiry.
Seven days a week, early morning to late evening · Current session and support hours are on the contact page, shown in your time zone§§ / Author
About Christa Smith, PhD.
Christa Smith, PhD
Dr. Smith is a Licensed Clinical Psychologist who specializes in psychological and neuropsychological assessment and evidence-based therapy for adults. Her clinical work integrates cognitive behavioral therapy, acceptance and commitment therapy, and trauma-informed approaches with formal assessment when clarity on diagnosis or cognition is needed. She sees clients through CEREVITY's nationwide private-pay telehealth network. View full bio →
§§ / Sources
References.
- Acta Psychiatrica Scandinavica. The effects of psychotherapies for depression on response, remission, reliable change, and deterioration: A meta-analysis. 2021. research.vu.nl
- Psychotherapy. The alliance in adult psychotherapy: A meta-analytic synthesis. 2018. doi.org
- Psychotherapy. Alliance rupture repair: A meta-analysis. 2018. doi.org
- Clinical Psychology Review. Using progress feedback to improve outcomes and reduce drop-out, treatment duration, and deterioration: A multilevel meta-analysis. 2021. pure.uva.nl
- National Institute of Mental Health. Attention-Deficit/Hyperactivity Disorder (ADHD) statistics. 2026. nimh.nih.gov
- CEREVITY. High-stakes anxiety therapy. cerevity.com/high-stakes-anxiety-therapy
- CEREVITY. Decision fatigue therapy. cerevity.com/decision-fatigue-therapy
- CEREVITY. Executive burnout therapy. cerevity.com/executive-burnout-therapy
⚠ Crisis resources
If you are experiencing a mental health crisis or having thoughts of suicide, please reach out immediately. 988 Suicide & Crisis Lifeline · Call or text 988 Crisis Text Line · Text HOME to 741741 National Alliance on Mental Illness · 1-800-950-NAMI (6264)
