The evidence behind finding the right person

The short version

Three findings shape how we work. If you read nothing else on this page, read these.

The relationship predicts how well things go. It predicts it more reliably than the type of therapy does. This holds across decades of research and thousands of studies.

Pairing people on purpose works. A randomised trial tested it directly. Those paired with a professional strong in their difficulty did better than those assigned as usual.

Getting what you asked for keeps you there. Across 53 studies, people who got the support they wanted were less likely to drop out. They also did better overall.

The rest of this page is the evidence for those three, including where it runs out.

Sources: Fluckiger et al. 2018, Wampold & Fluckiger 2023, Constantino et al. 2021, Swift et al. 2018

Why the relationship matters more than the method

Most people assume the key to good therapy, coaching or counselling is finding someone who uses the right technique. Decades of research tell a different story.

The most consistently replicated finding concerns the relationship between you and the professional you work with.

Researchers call it the "therapeutic alliance". It covers trust, mutual respect and agreement on what you are working towards. It also covers the sense that the person opposite you genuinely understands you.

Of everything that can be influenced before you begin, it has the strongest evidence behind it.

What the numbers actually show

You may have seen a pie chart splitting therapy outcomes into neat percentages. So much for the relationship, so much for the technique. It comes from a 1992 model by Michael Lambert and it is widely reproduced.

We are not going to show it to you.

In 2019 Cuijpers, Reijnders and Huibers reviewed the field. They found the chart's percentages "are in fact no more than rough estimates". No empirical evidence supports them.

A trustworthy version cannot be built from any meta-analysis we have.

We mention this because the chart is persuasive and it would help us if it were true. It is not, so it is not on this page.

What can be measured is how strongly each factor tracks with how well people do. Those figures come from large meta-analyses pooling many studies. The section on different settings below gives them for therapy and for coaching.

One of those meta-analyses holds the most useful comparison on this page. It measured how well a professional understands a client's background, rated two ways, against the same outcomes.

When a client rates how well a professional understands their background, that rating tracks closely with how the work goes. When the professional rates their own understanding, it tracks with almost nothing.

That is the most useful finding on this page, and it shapes how we work. A professional describing themselves is weak evidence. What the people they have actually worked with say is strong evidence.

Sources: Lambert 1992, Cuijpers et al. 2019, Soto et al. 2018

Does pairing people deliberately actually help?

Showing that the relationship predicts outcomes is one thing. Showing that pairing people on purpose changes anything is another. That question has been tested directly.

In 2021, Constantino and colleagues ran a randomised trial across six clinics. It covered 48 professionals and 218 people seeking help. Some were paired with a professional whose track record showed strength in the difficulty they arrived with.

The rest were assigned the usual way. Neither side knew which group they were in.

The deliberately paired group did better. The professionals did not change how they worked. The only thing that changed was who sat opposite whom.

On the trial's main measure, symptoms and day-to-day functioning, the effect size was 0.75.

That is the strongest single piece of evidence for what we do. It is also one trial, in one country. It paired on track record, not on the relational signals we listen for.

It points firmly our way without finishing the argument.

Source: Constantino et al. 2021

Why we ask what you want

This is the reason we ask about the things other services turn into checkboxes.

In 2018, Swift, Callahan, Cooper and Parkin pooled 53 studies covering more than 16,000 people. When someone got the kind of support they had asked for, they were less likely to drop out. They also did better overall.

That matters more than it first appears. The hardest part of getting help is often not the work, it is starting and then staying.

Someone who feels understood in the first conversation is more likely to book and more likely to arrive. They are more likely to still be there in six weeks. Someone who does not is often lost before anything can help them.

So we act on what you tell us you want.

That might be a woman. It might be someone who shares your faith, or someone who will not need neurodivergence explained.

Not because a shared background is a guarantee. Because being able to begin at all is where every good outcome starts.

A conversation you actually turn up to beats a better pairing you do not begin.

Source: Swift et al. 2018

The research across different settings

Therapy and counselling

The most comprehensive evidence comes from psychotherapy research. A 2018 meta-analysis by Fluckiger, Del Re, Wampold and Horvath pooled 295 studies covering more than 30,000 patients.

They found a steady, moderate link between the strength of the relationship and how treatment went: a correlation of .278. It held regardless of the type of therapy, the difficulty someone arrived with or their background.

Wampold and Fluckiger returned to the question in 2023. They concluded the alliance is among the most robust findings in psychotherapy research. It holds across treatment types, settings and measures.

They are careful about what that does and does not prove, and so are we. See the limitations below.

In 2018 the APA Division 29 Task Force named the relationship an evidence-based element of effective practice. Norcross and Lambert led it. Its recommendation was that professionals actively monitor and adjust the relationship throughout treatment.

Coaching

The alliance effect is not limited to clinical settings. A 2020 meta-analysis by Grassmann, Scholmerich and Schermuly synthesised 27 samples covering 3,563 coaching processes. The link between the working relationship and how well coaching went was stronger still: a correlation of .41.

The strongest effects appeared in how people thought and felt, rather than in what they did. So the relationship matters most when coaching aims to shift your view of a challenge.

Hypnotherapy

Here the evidence base is thinner. There is no large-scale meta-analysis of the working alliance in hypnotherapy. So we cannot put a number on this page next to the others.

What we can say is that the alliance findings hold everywhere researchers have looked carefully. That covers different treatments, different problems and different settings. It would be surprising if hypnotherapy were the exception.

That is a reasonable expectation, not a measured result. We would rather tell you which of the two you are reading.

Sources: Fluckiger et al. 2018, Wampold & Fluckiger 2023, Norcross & Lambert 2018, Grassmann et al. 2020

How we put it together

When you talk to us, we listen for what research says matters.

How you communicate. How you process things. What would make you feel heard, and what you have told us you want.

We use that to rank professionals on relational fit, alongside practical requirements like language, location and availability.

Some things are not negotiable. If you need trauma-focused work, we only put forward people trained for it. Within that qualified group, we rank on fit rather than credentials alone.

Where we can, we lean on what previous clients report rather than on how a professional describes themselves. The professional's own rating in the chart above is why.

What the research does not say

Honesty about limitations is important.

None of this means technique is irrelevant. For some conditions NICE names particular approaches rather than general supportive therapy. For PTSD it recommends trauma-focused CBT and EMDR, and for OCD it recommends CBT that includes ERP.

That is why the type of professional and their speciality are requirements for us. They are settled before anything else is weighed.

Shared background is worth being precise about. In 2011 Cabral and Smith pooled 52 studies of what people want. They found a clear, moderate preference for a professional who shares their racial or ethnic background, with an effect size of 0.63.

Then they looked at 53 studies of how treatment went, and found almost no difference. The effect size was 0.09. So the research does not show that a shared background alone produces a better result.

We treat your preference as what it is. A strong signal about what will let you begin and keep going.

We do not dress it up as a clinical guarantee. What does track with outcomes is whether a professional understands your context, judged by the people they have worked with. That is a different thing from sharing a label with you.

The biggest single influence on how therapy goes is you. What you bring, your readiness, the people around you, whatever else is happening. Nobody's algorithm changes that.

What a good pairing can do is remove the obstacles between you and the work.

Finally, correlation is not causation. A strong relationship might partly reflect people who are already improving feeling warmer towards the professional.

Researchers have taken this seriously. They use designs that track the relationship and symptoms over time, and the relationship still predicts what happens next. But this is an active question, not a closed one.

The research gives us a clear direction, not a guarantee. We use it to make the introduction as informed as we can.

Sources: NICE NG116, NICE CG31, Cabral & Smith 2011, Soto et al. 2018, Wampold & Fluckiger 2023

How we handle this evidence

We cite the papers behind every claim above. Where a finding is thinner than we would like, we say so next to it, not in a separate note.

We are not a research institution. We connect adults with a professional who fits. We are careful about the difference between a measured result and a reasonable expectation.