Solution-focused brief therapy: how it actually works in practice

A client turns up for her third session and says: «this week went better, but I don't know why». In most models that remark is a minor detail and the session carries on where it left off. In solution-focused brief therapy it is the material you work with for the next forty minutes: what she did differently, who noticed, how she managed it, and what it would take for it to happen again. That is not optimism, it is a methodological choice, and behind it sit forty years of clinical practice and a research base of uneven quality that is worth knowing about before anyone claims to «work from solutions».

This guide is written for practitioners: where the model comes from, which questions carry the work, how a first session is structured, what the evidence actually says without overselling it, when it is not the right tool, and how you document a future-oriented process without leaving the clinical record half done.

What solution-focused brief therapy is

Solution-focused brief therapy (SFBT) is a psychotherapy model that builds change out of what the client already does well, rather than analysing where the problem came from and what keeps it going. The APA Dictionary definition describes it as a future-oriented, goal-directed therapy that spends little time exploring the complaint and a lot of time identifying and expanding the person's own resources.

Put another way: solution-focused therapy does not assume you have to understand a problem in order to solve it. That is the claim that draws most resistance, and also the one that defines the model. A client with insomnia does not necessarily have to reconstruct when it started, what maintains it and what function it serves; she can start from the three nights last month when she slept well and work out what was different about them.

From Milwaukee to your practice: where solution-focused brief therapy came from

The model was developed by Steve de Shazer and Insoo Kim Berg with their team at the Brief Family Therapy Center in Milwaukee in the early 1980s. They came out of the Palo Alto strategic brief therapy tradition and Milton Erickson's influence, and their working method was unusual: they recorded hundreds of sessions and kept the questions that, watching the tapes, were followed by change. What is taught today as technique came out of that empirical sifting, not from a prior theory of mind.

That explains two things about the model. First, how concrete it is: these are questions, not constructs. Second, that it has an atheoretical feel which makes practitioners from more explanatory traditions uncomfortable. SFBT does not deny the unconscious, schemas or conditioning; it simply does not need them in order to operate.

The assumptions solution-focused therapy rests on

Almost the whole model fits into four ideas, and the first three are repeated in any training like a refrain:

• If it isn't broken, don't fix it. Whatever works in the client's life is left alone, even if you can see room for improvement.
• If it works, do more of it. A good part of the work is spotting behaviour that already brings relief and increasing how often it happens.
• If it isn't working, do something different. Your own interventions included: if you have spent five sessions pressing the same point with no result, the pressing may be the problem.
• The client is the expert on their own life, and you are the expert at steering the conversation. That is not a friendly slogan: it decides who proposes the goals and who limits themselves to asking about them precisely.

There is a fifth assumption that gets mentioned less and holds up everything else: change is already happening. Days pass between the phone call asking for an appointment and the first session, and in those days something different has almost always occurred. Asking about it opens the conversation somewhere very different from «tell me what brings you here».

What solution-focused therapy is not

Three common confusions, worth clearing up because all three do damage.

It is not positive thinking. The model does not ask the client to see the glass half full, nor does it argue with their account. It takes the complaint as it comes — suffering is always validated — and chooses to spend the time on a different part of the conversation. A solution-focused therapist who interrupts an account of pain to point out the good is not applying the model, they are misusing it.

It is not a set of techniques to drop into another therapy. The miracle question on its own, in the middle of a cognitive-behavioural session, is a pleasant question that leads nowhere. What produces the effect is the sustained sequence: goal, exceptions, scale, amplification, task, review. You can integrate the model with others — plenty of people do — but integrating is not sprinkling.

It is not fast therapy because the client is in a hurry. «Brief» describes the focus, not a session cap imposed from outside. An SFBT process that runs longer because it needs to is perfectly coherent; a six-session therapy with no focus is not brief, it is short.

The questions that carry the work

In this model the technique is the question. There are no homework records and no exposure hierarchies: there is a carefully steered conversation in which the therapist listens, picks up the client's exact words and hands them back turned into the next question.

The miracle question

It is the best known and the worst used. The classic wording is long on purpose, because it needs the client to step into the scene:

«Suppose that tonight, while you are asleep, a miracle happens and the problem that brings you here is solved. Because you were asleep, you don't know it has happened. What would be the first thing you noticed tomorrow morning that would make you think something had changed?»

What matters comes afterwards: the first answer is almost never usable. «I'd feel fine» or «I wouldn't be anxious» are absences, and you cannot work with absences. The craft is in turning them into observable presences with «and what would you be doing instead?», «who would notice first?», «what would that person see?». A good miracle question takes ten or fifteen minutes and ends with a concrete scene, with a time, a place and a behaviour.

The question has many variants and it pays to have them ready, because the literal version jars with adolescents, with people in recent bereavement, and with anyone in an acute crisis. «If this were much better six months from now, what would tell you?» does the same job without the word miracle.

Scaling questions

These are the most portable tool in the model and the one most often borrowed by other approaches: «on a scale of 0 to 10, where 10 is the day after the miracle and 0 is the worst it has been, where are you today?». They do three different jobs and it is worth not mixing them up: measuring the current state, defining the next step («what would be different at a 5?») and gauging confidence and motivation («how confident are you, 0 to 10, that you can get to that 5?»).

A technical detail that gets missed: a client who says 3 and says 3 again the following week is not necessarily stuck. The scale is subjective and its value lies in the conversation it opens, not in the number. «You told me 3 last week and it's 3 today, and you still came — how did you manage that?» is a perfect coping question.

Exceptions, coping and compliments

Exception questions look for the moments when the problem does not show up, or shows up less: «tell me about a day last week when the argument didn't end in shouting». If the client cannot find any — which happens often with severe depression — the model switches track and asks about coping: «you've been like this for three months and you still get up every morning to take the children to school; where does that come from?». The practical difference is large: the exception question looks for change, the coping question looks for resources where there is no change yet, and it is the one that rescues sessions that were sinking.

Compliments are the third pillar and the most easily misread, because the word sounds like flattery. It is not: it is specific feedback about something the client has done, with the evidence in front of you. «You booked the appointment yourself, you arrived on time and you brought notes on what you wanted to tell me» is a useful compliment. «You're doing so well» is noise.

How a first session runs

The classic BFTC structure still works, although hardly anyone takes the break so formally today. A first SFBT session looks fairly close to this:

1. Frame and legal ground (10 min). Same as any model: informed consent, data protection, fees and cancellation terms. There are no shortcuts here; if you want to review the script, it is in our first session protocol.
2. Problem-free talk (5 min). What she does with her life, who she lives with, what she enjoys. This is not filler: the resources you will use later come from here.
3. What brings her and what she expects (10 min). You listen to the complaint without squeezing it, and ask early about the goal: «what would have to happen for you to say that coming here was worth it?».
4. Miracle question and scene (15 min).
5. Exceptions and scale (10 min).
6. Break and feedback (5 min). The therapist summarises, offers two or three concrete compliments and proposes a task.

The first-session task is nearly always observational: «between now and next week, notice what happens that you would like to keep happening, and bring it written down». It looks like very little. In practice it redirects the client's attention for seven days, which is far longer than the time she spends with you.

Visitor, complainant, customer

De Shazer described three ways the client-therapist relationship can be positioned, and they are among the most useful things the model offers for everyday work, including outside it. The key point — constantly misread — is that they describe the relationship right now, not the person's character:

• Visitor: does not acknowledge a problem of their own, or has been sent by someone else (the court, a partner, the school). You do not ask them to change; you look for what they would want so that people stop pressing them.
• Complainant: acknowledges the problem but places it outside themselves and expects the other person to change. You work with observation and with whatever is within their reach.
• Customer: acknowledges the problem and sees themselves as part of the solution. This is where action tasks make sense.

Handing an action task to someone in the visitor position is the quickest way to make sure they do not come back. Much of what we call «resistance» is a mismatch between what we ask for and where that person actually stands, and it shows up in dropout rates; we write about that separately in therapy adherence and dropout.

Well-formed goals

The model is demanding about goals, and that is why it works with short processes. A goal is usable when it meets most of this: it is small, described in observable behaviour, framed as the presence of something rather than the absence of something («phone my sister once a week», not «stop being angry»), set in a concrete context, realistic within the life that person actually has, and described as something that will take effort.

That last point gets forgotten and it matters: if the goal looks easy, the client does not credit the achievement to her own work. Solution-focused therapy is quite deliberate about handing the credit back to the person who earned it, because that is where the sense of agency comes from — the one that holds the change up once you are no longer there.

What the research says, without overselling it

This calls for precision, because the model drags a promotional literature behind it. The honest summary is: there is a fair amount of research, the effects it finds tend to be small to moderate and comparable to other established treatments for several problems, and methodological quality is uneven. Recurring meta-analyses point to small samples, highly heterogeneous outcome measures, and allegiance effects on the part of those doing the research.

A recent, accessible example: a systematic review with meta-analysis published in Frontiers in Psychology on SFBT with cancer patients finds improvements in anxiety, depression and fatigue while at the same time warning about heterogeneity between studies. That double reading — favourable results plus methodological caution — is the one to hold when someone asks whether «this works».

What is reasonably well established is how well it fits settings with limited time: public services with waiting lists, school programmes, family support, social services teams. When the number of sessions is fixed in advance, a model that works with a focus from the first session has a practical advantage over one that spends three on assessment. The Solution Focused Brief Therapy Association keeps training standards and practice materials if you want to go further.

When solution-focused therapy is not the tool

No model does everything, and the marketing around this one sometimes forgets it. There are situations that call for something else, or at least something else first:

• Active risk: suicidal ideation with a plan, ongoing violence, a child at risk. Assessment and safety first; the rest comes later.
• Presentations that call for a specific protocol: OCD with established rituals responds to exposure with response prevention, and that is not improvised from a solution-focused stance. The same goes for complex trauma, where EMDR and trauma-focused therapies have their own track record.
• Acute psychotic decompensation or severe active substance use, where the priority is stabilising and coordinating with medical care.
• Clients who explicitly ask to understand. If someone has spent years with the question «why is this happening to me» and that is what they came for, refusing it in the name of the model is poor relational practice. You can do SFBT afterwards, or integrate it.

There is also an honest limit on the therapist's side: the model looks easy and it is not. The questions are written down, but choosing which one fits, in what tone and at what moment takes practice and supervision, like anything else.

How many sessions «brief» actually means

In the classic BFTC literature the median was around three or four sessions, and services working this way usually land between three and eight. But the number is not the goal: the criterion is the model's own, «as many as needed and not one more», with the closing question on the table from the start. Every session reviews the scale and asks directly whether another appointment is needed, and when.

One practical consequence follows: sessions spread out as the process advances. Weekly at first, every two or three weeks later, then a follow-up at one or three months. That is a different pattern from the classic «every Thursday at six», and your scheduling needs to allow it without a fight. The ending, when it comes, deserves a session of its own; we go into it in therapy discharge and closing the process.

Documenting solution-focused therapy without betraying the model

Here a real tension shows up. The model looks forward and works in the client's language; the clinical record, by contrast, is a legal and ethical obligation that does not depend on the approach you use. In Spain, Law 41/2002 on patient autonomy sets a minimum retention period of five years from the discharge of each episode of care, and several regions extend it; in other jurisdictions the periods differ, so check your own. Your professional code — the Spanish General Council of Psychology publishes its version — adds its own requirements on confidentiality and custody.

The way to resolve it is simple: record what any clinical record needs — presenting problem, assessment, risk explored, plan, interventions, progress — and add what this model produces and others do not. In practice, four things per session are enough:

• The goal as the client put it, in her words, in quotation marks.
• Today's scale and the previous session's, with one line on what would be different at the next point up.
• Any new exceptions that turned up.
• The task agreed, and what happened with the previous one.

Written like that, the process reads in one pass before the next session and progress is visible without any chart: the sequence of numbers tells the story by itself. And it has one unromantic but very practical advantage — if a report is ever needed, or the discharge has to be justified, the material is already in order.

How it differs from other brief models

SFBT is often confused with three neighbouring models, and the differences are more about focus than technique:

• Narrative therapy: shares the postmodern root and also works with exceptions (there, «unique outcomes»), but its unit of work is the identity story and its main tool is externalising the problem. We cover it in our guide to narrative therapy.
• Motivational interviewing: not a complete therapy but a conversational style for working with ambivalence, and it fits before change rather than during it. If what you have in front of you is a «yes, but», that is motivational interviewing.
• CBT: shares the interest in what is observable and the brevity, but it does run a functional analysis and works with cognitive content and exposure. It is the model with the most accumulated evidence across most presentations, and many practitioners use both depending on the case: see CBT techniques and, for the acceptance variant, ACT.

Training, supervision and practice

Reading about the model takes an afternoon; doing it well takes considerably longer. What tends to make the difference is listening to yourself: recording sessions with consent, reviewing which questions you asked and what happened immediately afterwards, and taking that to supervision. It is exactly the method SFBT was born from, and it is still the one that works best for learning it. If you are setting up your own supervision arrangement, we write about how to organise it in clinical supervision.

Two practical warnings. One: weekend courses hand you the questions, not the judgement; judgement comes from cases and from someone pointing out when you have started persuading the client instead of asking them. Two: if you come from a heavily explanatory model, the hard part will not be learning the miracle question, it will be sitting with the silence without interpreting it.

How My Psico Agenda supports a brief process

My Psico Agenda does not do therapy and does not tell you what to ask; it handles the administrative side, which in a model with spaced sessions and short processes weighs more than it seems. What it does solve:

• The clinical record is free text, so the four-line scheme above — goal, scale, exceptions, task — is written exactly as it is, without forcing it into fields designed for a different model. Each session also keeps its own note.
• Scheduling does not impose a weekly pattern: you can space appointments as the process requires, and block the hours you do not work.
• WhatsApp reminders cut no-shows, which hurt far more in a four-session process than in a forty-session one.
• Per-client statistics show the days since the last session, the average spacing and attendance, which is exactly what you need when you space appointments on purpose and want to know whether a case has been left hanging.
• Session packs fit short processes agreed in advance, with the remaining sessions worked out for you instead of counted by hand.

Individual plans start at €4.99/month + VAT — the Seed plan, with 25 patients, 20 automatic reminders and 20 VeriFactu invoices a month — and continue at €19.99 (Junior) and €29.99 (Senior), with unlimited patients. If you work in a team, practice plans start at €124.99/month + VAT for up to five therapists, each with their own account and their own patients. The first month is free, with every feature and without asking for a card, and there is no lock-in. If you work on your own, the page for self-employed psychologists sums up what is usually needed.