CBT has more supporting evidence than any other psychological therapy, and it does not work for everyone. Both statements are true, and people who are only told the first frequently conclude that failure to benefit is their own fault.
Where the evidence is strongest
Panic disorder, specific phobias, social anxiety and OCD, where specific protocols substantially outperform general support.
PTSD, where trauma-focused CBT is a first-line recommendation.
Insomnia, where CBT outperforms medication over the long term.
Depression and generalised anxiety, where it performs comparably to medication in many trials and maintains gains better afterwards.
Where it is an addition rather than a treatment
In psychosis and bipolar disorder, CBT is used alongside medication rather than instead of it, and can reduce distress and relapse.
In eating disorders, an adapted version has good evidence for bulimia and binge eating disorder, alongside nutritional and medical care.
What the numbers look like
Typical findings are that around half of people receiving CBT for depression or anxiety improve substantially, with a further group improving somewhat and a minority not benefiting.
These are good results for a psychological treatment and they are not universal, which is worth knowing before starting.
Who tends to do well
People with a clearly defined difficulty rather than diffuse distress.
People able to identify thoughts and feelings, and to complete work between sessions.
People whose circumstances are reasonably stable, since therapy cannot resolve an ongoing crisis.
Who tends to do less well
People with long-standing difficulties rooted in early relationships, where a short structured course may not reach the material.
People with several conditions occurring together, where a single protocol addresses one part of a complex picture.
People in ongoing adverse circumstances: poverty, unsafe housing, an abusive relationship or insecure immigration status.
People who find the structure constraining, or who want to explore rather than to work to an agenda.
Where adaptation is needed
Standard CBT assumes certain capacities that not everyone has.
Difficulty identifying emotions, common in autistic people and in alexithymia, makes emotion-focused work harder. Adapted versions exist and work considerably better.
Executive function difficulties, as in ADHD, make between-session tasks harder to complete. This is not a lack of motivation and needs accommodating rather than interpreting.
Where these apply, saying so at the outset changes what is offered.
Why people do not benefit
The wrong protocol for the condition, which is common. General CBT for OCD without exposure and response prevention is substantially less effective.
Too few sessions, particularly where a low-intensity intervention was offered for a difficulty needing more.
A poor therapeutic relationship, which reduces effectiveness across all approaches.
Something else driving the difficulty: undiagnosed ADHD, autism, sleep apnoea, thyroid disease or trauma.
Circumstances that no therapy can address.
What to do if it has not helped
Establish which version you had, how many sessions, and whether it matched your condition.
Ask about stepping up, about a different approach, or about assessment for something not yet considered.
Alternatives with evidence include interpersonal therapy, psychodynamic therapy, EMDR for trauma, and third wave approaches including acceptance and commitment therapy and dialectical behaviour therapy.
Give it a fair trial
People frequently judge CBT after three or four sessions, which is generally before any active intervention has begun.
The early sessions are assessment and formulation, and it is normal for nothing to have changed at that point. Most protocols expect improvement to become apparent somewhere around the sixth to eighth session.
Where nothing at all has shifted by then, raising it is appropriate. Concluding before then is judging the preparation rather than the treatment.
Combining it with medication
These are frequently presented as alternatives, and for moderate to severe depression guidance generally supports offering both.
Medication can reduce symptoms enough to make therapy possible, particularly where concentration and energy are too impaired to engage with the work.
Choosing between them is unnecessary in most cases, and asking about both is reasonable.
It is not a personal failing
People frequently describe feeling they did CBT wrongly, which reflects how it is presented rather than how it works.
A treatment that helps roughly half of people substantially is a good treatment, and being in the other group is a finding about fit rather than about effort.
This is general information rather than medical advice.