Therapeutic Alliance in Mental Health: Evidence, Measurement & Outcomes

Therapeutic alliance is one of the most replicated predictors of psychotherapy outcome in the research literature. Across treatment modalities, client populations, and clinical settings, the quality of the relationship between client and therapist consistently accounts for a meaningful portion of variance in outcomes — more, in many studies, than the specific techniques or protocols being used.

That isn't an argument against technique. It's an argument for taking the relationship seriously as a clinical variable, one that can be monitored and actively managed rather than left to intuition.

What is the therapeutic alliance?

The dominant conceptual framework comes from Edward Bordin's 1979 tripartite model. Bordin proposed that the working alliance — the collaborative dimension of the therapeutic relationship — consists of three interrelated components:

• Bond: The quality of the personal connection between client and therapist. This includes mutual trust, acceptance, and the sense of being genuinely understood.
• Goals: Shared agreement on the outcomes the client and therapist are working toward. Alliance is weakened when client and therapist hold different ideas of what treatment is for.
• Tasks: Agreement on the therapeutic activities and methods being used. Clients need to experience the work as relevant and credible — not just endure it.

Bordin argued that the alliance was not specific to any one therapeutic model. It was a pan-theoretical construct, present in every effective therapy regardless of orientation. That framing held up well under subsequent empirical scrutiny.

Why it matters: the evidence base

Bruce Wampold's 2001 contextual model of psychotherapy — articulated in The Great Psychotherapy Debate — drew on meta-analytic data to argue that so-called "common factors" (including the alliance, empathy, and client expectations) account for substantially more outcome variance than specific treatment techniques. In his analysis, specific ingredients accounted for roughly 1% of variance; common factors, including alliance, accounted for considerably more.

The most comprehensive recent synthesis is Wampold and Flückiger's 2023 World Psychiatry meta-analysis (doi: 10.1002/wps.21035), which reviewed the alliance-outcome literature and confirmed the relationship as one of the most consistently supported in clinical psychology. Across 17 meta-analyses, the alliance-outcome correlation was consistently significant, with effect sizes in the moderate range. Critically, the authors noted that the alliance is both a predictor and a product of good treatment — early alliance formation matters, but so does the ability to repair it when ruptures occur.

Flückiger and colleagues' 2018 meta-analysis in Psychotherapy (covering 295 studies, n = 30,000+) found an average weighted correlation of r = .278 between alliance and outcome. This makes the alliance one of the most replicated findings in psychotherapy research, cutting across cognitive-behavioural, psychodynamic, humanistic, and integrative approaches.

How to measure therapeutic alliance

There are two main instruments in routine clinical use, suited to different purposes.

Working Alliance Inventory (WAI)

The Working Alliance Inventory (WAI) was developed by Adam Horvath and Leslie Greenberg (1989) directly from Bordin's tripartite model. It operationalises the Bond, Goal, and Task components in three parallel subscales.

Three versions exist:

The WAI-SR (Munder et al., 2010) is currently the most widely used version in clinical research settings. It maintains strong psychometric properties and is well-validated in both outpatient and online intervention contexts.

Typical administration timing: after sessions 3 and 8, or at fixed intervals across a treatment episode. The WAI is not designed for every-session use — it is better suited to periodic structured assessment of alliance quality.

Session Rating Scale (SRS)

The Session Rating Scale (SRS) takes a different approach. Developed by Barry Duncan and Scott Miller (2003), it is an ultra-brief 4-item measure designed to be administered at the end of every session, in under one minute.

Each item is a 10-centimetre visual analog scale covering:

1. Relationship (feeling heard, understood, and respected)
2. Goals and Topics (addressing what matters to the client)
3. Approach or Method (fit of the therapist's approach)
4. Overall (general sense of the session)

Total scores range from 0–40. A score at or below 36 is the clinical cutoff for a meaningful alliance concern — a signal to open a conversation about the client's experience before the session closes.

The SRS is part of the Partners for Change Outcome Management System (PCOMS), a feedback-informed treatment framework. Research by Duncan (2012) shows that practitioners who routinely collect and respond to SRS feedback demonstrate significantly better client outcomes than those who rely on clinical intuition alone.

Real-time alliance monitoring in practice

The clinical case for session-by-session alliance monitoring rests on a straightforward problem: practitioners are poor at detecting when clients are dissatisfied. A 2005 study by Hannan et al. found that therapists predicted which clients would deteriorate at only slightly above chance. The clients' own ratings, collected with brief tools like the SRS, were better predictors.

Feedback-informed treatment (FIT) operationalises this insight. The core practice is simple:

1. Administer the SRS at the end of each session.
2. Score it before the client leaves.
3. Discuss any items that scored noticeably low — not defensively, but with genuine curiosity.
4. Adjust approach based on what you learn.

This loop — measure, review, discuss, adjust — converts the alliance from background noise into a clinical variable. Low SRS scores that are acknowledged and addressed are associated with better outcomes than low scores that go unnoticed or unaddressed (Duncan, 2012).

For practitioners in training or supervision, combining the SRS (session-by-session) with the WAI (periodic structured assessment) provides both real-time feedback and a more granular picture of the Bond/Goal/Task components across the arc of treatment.

When the alliance ruptures

Even in effective treatments, the alliance does not remain stable. Ruptures — moments of tension, misalignment, or breakdown in the therapeutic relationship — are common. Safran and Muran's rupture-repair model, developed through the late 1990s and formalised in their 2000 book Negotiating the Therapeutic Alliance, describes two types:

• Withdrawal ruptures: The client disengages — becomes compliant but distant, avoids discussing something important, or reduces emotional engagement.
• Confrontation ruptures: The client expresses dissatisfaction, frustration, or criticism of the therapist or treatment.

Confrontation ruptures are often easier to detect. Withdrawal ruptures can persist for sessions before becoming visible, which is part of why routine measurement matters — a sustained drop in SRS scores without obvious explanation is often a withdrawal rupture in progress.

The repair process involves three steps: noticing the rupture, marking it explicitly (rather than working around it), and exploring the client's experience with genuine openness. Safran and Muran's Alliance-Focused Therapy and subsequent research have demonstrated that successful repairs are associated with better outcomes, and in some cases with stronger alliances post-repair than pre-rupture.

Practitioners who avoid ruptures by adjusting their approach before the client disengages show different patterns from those who allow ruptures to develop unaddressed. Real-time monitoring makes early course correction possible.

Integrating alliance measurement into clinical workflows

Measurement-based care frameworks that include alliance monitoring have a practical advantage over those that track symptoms alone: they capture a dimension of treatment quality that symptom scales cannot. A client's PHQ-9 scores may be stable while their alliance is deteriorating — an early indicator that treatment is drifting before the symptom picture moves.

Practical integration points:

• Session-end: Administer the SRS every session. Score it before the client leaves.
• Mid-treatment review: Administer the WAI-SR at sessions 3–4 and again at session 8–10 to get subscale-level data on where Bond, Goal, or Task alignment is weakest.
• Case formulation: Use WAI subscale profiles to guide supervision — a low Task score suggests the client finds the approach irrelevant; a low Goal score suggests the treatment target needs renegotiation.
• Rupture detection: A sustained SRS score at or below 36, or a pattern of declining scores, warrants direct metacommunication before the alliance deteriorates further.

HiBoop includes both the SRS and WAI in its session-by-session measurement stack, so practitioners can track alliance alongside symptom outcomes within a single workflow.

The bottom line

Therapeutic alliance is not soft-skills territory. It is a measurable clinical variable with a well-documented evidence base, standardised instruments, and a clear relationship to treatment outcomes. Practitioners who monitor it — using tools like the SRS at every session and the WAI at regular intervals — can detect early signals of misalignment, repair ruptures before they compound, and make more informed decisions about when and how to adjust their approach.

The Wampold and Flückiger 2023 meta-analysis is unambiguous: the alliance is among the most reliable predictors of psychotherapy outcome, and it responds to deliberate clinical attention. Measurement is how that attention becomes systematic.

References

• Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of the working alliance. Psychotherapy: Theory, Research & Practice, 16(3), 252–260. https://doi.org/10.1037/h0085885
• Duncan, B. L. (2012). The partners for change outcome management system (PCOMS): The heart and soul of change project. Canadian Psychology, 53(2), 93–104. https://doi.org/10.1037/a0027762
• Duncan, B. L., Miller, S. D., Sparks, J. A., Claud, D. A., Reynolds, L. R., Brown, J., & Johnson, L. D. (2003). The session rating scale: Preliminary psychometric properties of a "working" alliance measure. Journal of Brief Therapy, 2(1), 3–12.
• Flückiger, C., Del Re, A. C., Wampold, B. E., & Horvath, A. O. (2018). The alliance in adult psychotherapy: A meta-analytic synthesis. Psychotherapy, 55(4), 316–340. https://pubmed.ncbi.nlm.nih.gov/29792475/
• Hannan, C., Lambert, M. J., Harmon, C., Nielsen, S. L., Smart, D. W., Shimokawa, K., & Sutton, S. W. (2005). A lab test and algorithms for identifying clients at risk for treatment failure. Journal of Clinical Psychology, 61(2), 155–163. https://doi.org/10.1002/jclp.20108
• Horvath, A. O., & Greenberg, L. S. (1989). Development and validation of the Working Alliance Inventory. Journal of Counselling Psychology, 36(2), 223–233. https://doi.org/10.1037/0022-0167.36.2.223
• Munder, T., Wilmers, F., Leonhart, R., Linster, H. W., & Barth, J. (2010). Working Alliance Inventory-Short Revised (WAI-SR): Psychometric properties in outpatients and inpatients. Clinical Psychology & Psychotherapy, 17(3), 231–239. https://doi.org/10.1002/cpp.658
• Safran, J. D., & Muran, J. C. (2000). Negotiating the Therapeutic Alliance: A Relational Treatment Guide. Guilford Press.
• Wampold, B. E. (2001). The Great Psychotherapy Debate: Models, Methods, and Findings. Lawrence Erlbaum.
• Wampold, B. E., & Flückiger, C. (2023). The alliance in mental health care: conceptualization, evidence and clinical applications. World Psychiatry, 22(1), 25–41. https://doi.org/10.1002/wps.21035