Somewhere around the third month, a lot of people sit down in the same chair, in the same room, and quietly wonder whether they're wasting their money. The first few weeks felt like opening windows in a stuffy house. Now the air has stopped moving. Nothing is wrong, exactly. Nothing is happening, either. They start doing arithmetic they would never say out loud: how many of these do I need? How long is this supposed to take?

There is a real research answer to that question. It doesn't say "twelve sessions" or "a year." It says something stranger and more useful: the shape of improvement is a curve, and almost no one is told what the curve looks like.

The question researchers actually asked

In 1986, Kenneth Howard and colleagues published an influential analysis of what they called the dose-effect relationship in psychotherapy. They pooled data from many earlier studies and asked a simple question borrowed from medicine: if therapy is a dose, how does the response change as the dose increases?

The pattern they reported has been described ever since as negatively accelerating. Improvement rose steeply at the beginning, then kept rising, but more and more slowly. By roughly the eighth session, about half of patients had measurably improved. It took many more sessions, on the order of half a year of weekly therapy, for that figure to reach around three quarters. After that, the curve flattened further.

So the first stretch of therapy buys a lot, and each later stretch buys less. This is the basic shape of the finding, and it has held up well enough to keep being cited. It is also easy to misread, so it's worth being careful about what it does and doesn't claim.

What the curve does not say

It does not say therapy stops working after twenty-six sessions. A flattening curve means the rate of measurable change slows, not that change ends. It also describes groups, not you. Averages hide enormous spread: some people improve in a handful of sessions, and some need years.

Later research complicates the neat numbers too. Studies of routine clinical care, such as work by Michael Lambert and colleagues using symptom questionnaires session by session, found that the number of sessions needed for reliable improvement varies a lot by problem, by person, and by how "improvement" is defined. A person with a recent, specific stressor and a person untangling decades of attachment trauma are not on the same timetable, and no honest reading of the evidence pretends they are.

The literature also distinguishes symptom relief from deeper change. Symptoms like acute anxiety or low mood often ease earlier. Changes in how you relate to people, how you treat yourself, or what you expect from the world tend to move slower and are harder to measure with a checklist.

Why the curve feels like failure from the inside

Here is the cruel part. The early steep section of the curve is partly made of relief: finally saying things aloud, finally being listened to, finally having a name for something. That is real, and it is also the cheapest improvement therapy ever delivers. It arrives when you have the most room to improve and the least resistance.

Then the work changes. You're past the easy gains and into the material that was hard enough to avoid for years. The sessions get less dramatic and sometimes more uncomfortable. If you expected the early pace to continue, the slowdown reads as a verdict: it stopped working.

It's the same trap as learning an instrument. Your first month feels like leaping forward because you start from zero. Month eight feels like standing still because the improvements are finer. Nobody concludes the cello has stopped working. But in therapy, there is no audible difference between a note you can finally hit and a flat note, so we draw the wrong conclusion more often.

The shifting baseline makes it worse

As you improve, your sense of normal moves with you. The hard things from month one no longer register as hard, so you forget they were ever problems. Your brain keeps grading the present against a present that has already changed. If you've ever wondered how to tell if therapy is working when your memory of the starting point has gone soft, this is why: the evidence exists, you just aren't holding it.

Put the two effects together and you get a predictable mid-therapy crisis. The curve is flattening, which is statistically normal. And your internal reference point has drifted, which makes the flattening feel like zero. Both are ordinary. Neither tells you whether to stop.

So when is it reasonable to stop, change, or keep going?

The dose-response research suggests a better question than "how many sessions?" It suggests asking what is the curve doing for me, specifically? A few distinctions help.

A slow curve is not a flat line. Slow movement toward something you can name is different from months of circling with no change in what you do or feel outside the room.

Deterioration is a different signal. Lambert's work on feedback-informed treatment found that clients who were getting worse were easy to miss without tracking, and that sharing simple progress data with therapists helped. If you're steadily feeling worse over many weeks, not just uncomfortable after hard sessions, that deserves a direct conversation, and possibly a different approach or provider.

"Done" is a choice you define. Some people aim for symptom relief and are finished in a couple of months. Others use therapy as an ongoing place to think. Neither is the correct dose. What matters is that you decided, rather than drifting out on a bad week or staying out of habit.

Your next moves

  • Write down your starting line today. In five sentences, describe what a normal week looked like when you began: how you slept, what you avoided, what you said yes to when you meant no. Date it. It's the only defense against a shifting baseline.
  • Rate three specific things from 0 to 10 once a month. Pick concrete ones, such as how often you spiral at night, how hard it is to ask for something, how long a bad mood lasts. A number you wrote in March can't be rewritten by your October mood.
  • Ask your therapist where you are on the curve. Try: "If you had to describe what's changed since we started and what's still stuck, what would you say?" It's a fair question, and a good therapist will answer it.
  • Name what "enough" would look like. Write one sentence: "I'll know I'm done, or ready to space sessions out, when ___." Bring it to a session and see whether your therapist agrees.
  • Separate a hard week from a flat quarter. Before deciding the therapy has stalled, check whether you're judging the last seven days or the last ninety.

Where a record helps

Everything above depends on one thing that human memory is bad at: keeping an honest record of where you started and what shifted along the way. It is the thread that lets you see a curve instead of a bad Tuesday. That's the reason we built Sesh, a private place to capture what came up in a session while it's fresh, so months later you can look back and see the slope for yourself. If that sounds useful, you can look at it at sesh.lumenlabs.works. And if it doesn't, the first action item costs you nothing but a pen.