Reflection

Does Online Therapy Work? What the Evidence Says, and What It Cannot Measure

The question arrives in almost every enquiry about remote work, usually apologetically, as though it were an admission: would it be better if I came in person?

It deserves a straight answer, and the straight answer has two halves that do not sit comfortably together. The research is more reassuring than most people expect. And it does not measure the thing that clinicians who work at depth are actually worried about.

The objection everybody raises, and what the data says about it

Nobody’s first concern about online therapy is whether the technique survives the wire. The concern is the relationship. Therapy works through a relationship, so if the relationship is thinner over video, everything downstream of it should be thinner too.

This has been measured directly. A 2024 systematic review and meta-analysis pooled eighteen studies comparing therapeutic alliance in videoconferencing psychotherapy against in-person psychotherapy. Alliance rated by patients showed a difference of Hedges’ g = −0.09, with a confidence interval running from −0.26 to 0.07 and a p-value of .28. Alliance rated by therapists came out at g = 0.04, marginally in the other direction.

Both of those numbers are, in practical terms, zero. Patients did not report a weaker bond with a therapist on a screen. Neither did the therapists.

That finding is worth sitting with, because it contradicts something most people feel certain about before they have tried it — and it contradicts what a good many experienced clinicians, including me, would have predicted.

What the outcome research does and does not establish

On symptom outcomes, the evidence is broadly favourable and more limited than the enthusiasm around it suggests.

A 2025 systematic review comparing digital and face-to-face delivery of systemic psychotherapy screened 3,633 references to find four eligible trials covering 754 participants. Digital delivery came out superior on 18% of outcomes and face-to-face on 5%. The remaining 75% were indeterminate: neither superiority nor demonstrated equivalence.

That last number is the honest one. Most head-to-head comparisons are not large enough to prove that two formats are the same; they are only large enough to fail to find a difference. Those are not identical statements, and the gap between them is where a lot of confident marketing lives. The broader picture from the large psychotherapy meta-analyses, including Cuijpers and colleagues’ 2024 review across eight disorders in World Psychiatry, is consistent with remote delivery performing in the same range as in-person work. It is not proof that the formats are interchangeable for every person and every problem.

For eating disorders, where I would have expected format to matter most, the delivery-format evidence is more reassuring than I anticipated. A 2025 network meta-analysis of 36 trials and 3,136 participants found large effects for cognitive behavioural therapy across individual, group and guided self-help formats, with no significant difference between them — individual g = 1.06, group g = 1.08, guided self-help g = 0.94. Telehealth and face-to-face delivery were comparable for bulimia nervosa and binge eating disorder across six studies and 698 participants.

That concerns the psychological treatment. It does not touch the question of medical monitoring, which is a separate matter and does not travel down a video link.

The finding that does not flatter online work

There is one place where the evidence is genuinely awkward, and it is worth naming rather than skipping.

Earlier research found that videoconferencing therapy had higher attrition than in-person therapy — people left it sooner. A later meta-analysis of twenty randomised trials comparing teletherapy with in-person therapy found no significant difference in attrition at all, with a risk ratio of 1.006. Face-to-face psychotherapy has its own dropout problem regardless, with meta-analytic estimates ranging from 12% to 27%.

So the literature contradicts itself, which usually means the format is not the dominant variable. What I think is being measured here, badly, is not the screen at all. It is whether the work has a container around it — and that is a question with a name.

The same person is not the same person

Here is where I have to depart from the data, because the data does not measure what I want to talk about.

Some of my work is hybrid: one session a week in the consulting room, one session a week online, with the same person, about the same material. That arrangement turns out to be a natural experiment, and what it shows is not what I expected.

We tend to speak about a person as though there were one of them. There is not. I am not the same at work as I am at home, or with my children, and neither are you. Analytical psychology has language for this — the persona, the complexes, the parts of a personality that hold authority in one setting and go quiet in another — but you do not need the vocabulary to recognise it. You already behave as though it were true every time you brace slightly before a particular conversation.

Consider a state I have met more than once. In the consulting room: courteous, contained, carefully spoken. Softly spoken, in fact, week after week, and easy to be with. Then the same hour the following week, on video, from a room in their own house — and the voice rises. Anger arrives that had never once come into the room. Not anger at me, particularly, though I am the one receiving it. Anger that had apparently been there the whole time and had never been able to get into a consulting room in Harley Street.

Nothing had changed in the material. The setting had changed, and a different part of the person had the floor.

Another kind of thing happens too, quieter than that one. Someone whose manner is flat and rather blunt — not unfriendly, but plain, and plain in appearance: clothes chosen to be unremarkable, nothing about the way they dress inviting a second look. The work moves online, and behind them is a room of real beauty. Colour chosen with obvious care. Objects placed deliberately. An eye at work, and evident pleasure taken in the work of it.

None of that care had gone into what they wore. All of it had gone into the room.

I would not have known. Years in a consulting room and I would not have known, because it would never have been mentioned — not out of concealment, but because it would not have occurred to them that it counted as anything. And it changes what you understand: the capacity for beauty and attention was not missing from this person at all. It had simply been placed somewhere other than themselves. That is a question worth asking about, and I would not have had it to ask.

That is not a distortion to be corrected for. It is information I would not otherwise have had. My consulting room is my territory, and the person who arrives in it is shaped by that, partly by the courtesy the room asks for. On video they are on their own ground, and their own ground gives a different part of them permission.

The house tells you other things too. How quickly attention is pulled away by something happening elsewhere in it. Whether the hour has been protected or fitted in around domestic life. Whether someone arrives with a coffee in their hand, which tells you something about how the hour is being held, and which I will come back to.

So the honest correction to online gives you less is not that it gives you the same. It gives you something different, and some of what it gives you is unavailable in the room.

What is genuinely reduced is the peripheral, bodily register: the shift in breathing, the way a body organises itself just before it says something difficult, what happens inside a silence when you are both in it rather than each holding your own end of a connection. In depth work that material is not decoration, and a face on a screen does not supply it.

Both things are therefore true at the same time. The relationship survives the format, on the evidence and in my experience of it. Some information is lost. And other information arrives that the consulting room was quietly holding down — sometimes because the setting gives a suppressed part of someone permission to speak, sometimes because you are simply shown a life that would never have been described to you. Anyone deciding what to do deserves to be told all three.

Which direction the work goes

There is an assumption buried in the original question: that the consulting room is the real thing and the screen is the substitute, so the aim must be to get people into the room eventually.

That is not the aim, and it is not what happens.

Some people I will always see online, because they live too far away for anything else, and that is simply the arrangement rather than a compromise anyone is tolerating.

Sometimes the direction runs the other way entirely. Someone is frightened of coming — intimidated by the idea of it, or carrying a phobia that makes the journey itself the obstacle rather than the appointment. They start on video because that is what is possible. Moving from the screen into the room then becomes the therapeutic work, not the precondition for it. Online is where it starts. The room is what it achieves, if it is achieved, and that is a very different proposition from a downgrade.

And a good deal of the time the room is not worth what it costs. Sitting in the same space has a payoff, and I would not pretend otherwise. It also costs a journey, an hour or two of the day, and for a lot of people a quantity of stress that has to be set against the benefit. That trade comes out differently for different people, and differently for the same person in different weeks.

I work with a number of mothers who cannot simply leave — a young child, no one to hold the afternoon, a school run that makes a consulting-room hour structurally impossible. Some of them take the session in the car. Parked, mid-afternoon, in the twenty minutes before the pickup.

That is not a failure of the setting. It is frequently the only fifty minutes in the week when nobody needs anything from them, and the car is the one room in their life with a door that closes.

The temenos is not the room

Analytical psychology has a word for the consulting room, and it is not a neutral one. The room is a temenos — the Greek term for the precinct marked off around a temple. A piece of ground cut out from ordinary ground, inside which different rules apply. Jung used it for what analysis requires: a protected space, a vessel that holds what is being worked on for long enough that something can happen to it, rather than letting it leak back out into the rest of the week.

The word carries the thing people find hardest to say about therapy, which is that the space is not merely private. It is sacred, in the ordinary and unembarrassing sense: set apart, and treated differently from the rest of life.

When people ask whether online therapy works, this is the real question underneath it. Not whether the technique survives the wire. Whether a temenos can exist without a room.

A consulting room supplies one almost for nothing. The journey there, which is already a transition. The threshold. The waiting. The door closing. Fifty minutes that belong to nothing else because you are not anywhere else. None of it has to be thought about, because the building is doing it for you.

Online, nothing supplies it. It has to be made on purpose, and I ask people to make it. The same hour every week. A door that shuts, and the certainty that nobody will come through it. No drinking during the session, no smoking, nothing in your hands — which is why I notice the coffee, and why I say something about it. Not out of severity. Because the hour is not an ordinary hour and it needs to be treated as though it is not.

That is more effort than walking into a building, and it is the actual work of making online therapy hold.

Which is why the mother in the car is not the poor relation in this essay. The car is a temenos. Cut off, enclosed, belonging to nobody else, with a door that closes and a period of time with a hard edge at the end of it. It is doing precisely what the building would have done.

And it is why the failure mode online has nothing to do with the screen. It is the session taken at a kitchen table with the door open and somebody else’s afternoon carrying on behind it. Not because the picture is worse, but because there is no precinct — the vessel leaks, and nothing stays in it long enough to be worked on.

Where I say no

Sometimes I do say no, and it is worth being exact about when, because the reasons are not the ones people expect.

When someone needs more containment than a screen can give them. Containment is not a metaphor here. It means that when something overwhelming arrives, there is a person in the room whose presence holds it, so that it does not have to be held alone.

When I need to see how someone is in their body. Not their face — the screen gives me that. Their body: how they are sitting, what is happening in their breathing and their hands, whether the state they are describing to me is the state they are actually in.

Where there is psychosis, or where the grip on reality is unstable.

And where someone is very highly emotional and an abreaction is likely — the sudden discharge of feeling attached to something that has been held down for a long time, arriving faster than the person can manage it. This is the one that decides it for me. If that happens in my consulting room, I am there. We can stay inside it together, it can be worked with, and they leave when they are able to leave rather than when the hour formally ends. If it happens on a video call, the abreaction stops when the connection stops, and the person is alone in their house with everything that has just come up and nobody to bring them back. That is not a small difference. It is the reason I will decline remote work with certain people until something has changed.

There is also the early phase of eating disorder treatment, where physical observation is part of the assessment and medical oversight has to be real rather than reported.

EMDR is delivered online routinely and effectively, with bilateral stimulation adapted to the format, and the questions that decide it are the same ones asked of any EMDR referral: is the person stable enough, is there containment between sessions, is there anybody around them if the processing brings up more than expected. Those questions occasionally produce a no. The answer can change in either direction as a piece of work develops.

Where online is not the compromise

There is a group of people for whom this whole framing is wrong, because for them the consulting room was never the realistic option.

People who live abroad. People who move every two or three years. People whose work puts them on planes for a third of the month. People who have started good therapy somewhere and lost it to a relocation — which tends to happen at exactly the point when the relocation is the thing they most need to think about.

For them, weekly work with the same clinician over video is not a diminished version of proper therapy. It is the only version that survives contact with their life. The alternative is not the consulting room; it is starting again with somebody new every few years, or not going at all. Set against that, the question of whether a screen costs you some peripheral information is a small one.

What to do with all this

If you are deciding, the useful question is not whether online therapy works in general. It does, for a great many presentations, and the relationship holds up better than almost anyone predicts.

The useful question is whether it suits your particular situation — what you are bringing, how unwell you are, what support exists around you, and whether the work you need to do is the kind that can be done at that distance. That is a clinical judgement, and it is a reasonable thing to spend a first conversation on.

But the framing I would let go of is the one where the room is the real therapy and everything else is a reduced version of it. That is not what nearly two decades of working this way has shown me. Online is not less. It is different, and sometimes it is different in ways that give you more than the consulting room would have.

And for some people it is not a lesser way of getting help at all. It is the first way they have ever had.

Dr Philippe Jacquet is a UKCP registered psychotherapist and Jungian analyst with a doctorate from the University of Essex, working from Harley Street and Fitzrovia and internationally by secure video, in English and French. To discuss whether online work suits your situation, arrange a consultation, or read more about online therapy at this practice.