If you ask therapy researchers what predicts good client outcomes, the answer most of them give is some version of: not the specific technique, not the orientation, not even the experience of the practitioner. It's the working alliance.
The same is true in coaching. And the same is invisible to most coaches because the alliance feels like atmosphere — you know it when it's good, you sense it when it's strained, but you can't name what you're sensing.
The Working Alliance Inventory, Short Revised (WAI-SR) is a small piece of structure that turns the alliance into something you can see. Three questions, 30 seconds, every fourth session.
What the alliance actually is
The working alliance has three parts, in the framework Edward Bordin proposed in 1979 and that's held up in research ever since:
Bond. The personal connection between coach and client — feeling liked, respected, taken seriously, talked to as a whole person rather than a case.
Goals. Mutual agreement on what the work is for. Not just the topline ("I want to be promoted") but the underlying agreement on what success looks like, what the engagement is fundamentally about.
Tasks. Agreement on how the work happens. The coach's approach — the questions, the homework, the structure of sessions, the cadence — feels useful to the client.
A coach who's strong on bond but weak on tasks ends up with clients who like them but feel the work isn't moving. A coach who's strong on tasks but weak on bond gets results in the short run but loses clients to drop-out. A coach who's strong on goals but weak on bond and tasks runs technically correct sessions that feel transactional. The three legs are needed together.
Why the alliance predicts outcomes more than technique
Across thousands of psychotherapy outcome studies, alliance correlates with outcome at r ≈ 0.27. That doesn't sound huge until you compare it to almost everything else: specific techniques (CBT vs. psychodynamic vs. humanistic) correlate with outcomes at r ≈ 0.05-0.10. The alliance is several times more predictive than which method you use.
In coaching, the literature is smaller but pointing the same direction. The McKinsey-style "executive coaching ROI" framing is dominated by alliance-style variables — perceived chemistry, perceived alignment on the work — far more than by what specific assessments or methodologies the coach uses.
The mechanism is reasonable. When the alliance is strong, the client brings their actual messy reality to the work, accepts challenge from the coach, sustains effort between sessions, and stays with the engagement long enough for change to compound. When the alliance is weak, the client filters what they say, deflects challenge, drifts on accountability, and finds reasons to drop off.
What the WAI-SR actually asks
The Working Alliance Inventory was developed by Tracey and Kokotovic in 1989, with the Short Revised version (WAI-SR) refined by Hatcher and Gillaspy in 2006 to 12 items. For day-to-day coaching practice, even 12 items is too long — you'll do it once, never again.
In Arcline, we use a three-item version that captures the three Bordin dimensions cleanly:
- Trust with my coach — bond.
- Shared goals — goals.
- Useful work together — tasks.
Each rated on a 4-20 sub-scale (the WAI-SR uses 5 items per dimension scored 1-5, totaling 5-25; we collapse each dimension to a single 4-20 slider). Sum is 12-60. The cadence is every fourth session by default — long enough that the relationship has texture, short enough to catch a rupture before it becomes a drop-out.
The coaching score thresholds
Working ranges, calibrated against published WAI-SR norms and adjusted for typical coaching populations:
- 52+ (Strong alliance) — bond, goals, tasks all aligned. Continue what you're doing.
- 44-51 (Stable alliance) — solid working relationship. Watch for any one dimension trending down.
- 36-43 (Watch) — one or more dimensions sliding. Surface it explicitly in the next session.
- Below 36 (At risk) — alliance is strained enough that drop-out is plausible. Direct conversation needed.
The dimension-level scores matter as much as the total. A 50/60 with bond=18, goals=18, tasks=14 tells a different story from a 50/60 with bond=14, goals=18, tasks=18. The first is "the relationship is strong but my approach isn't fitting"; the second is "we're aligned on the work but the personal connection is thin."
When to run the check
Two things have to be true for an alliance check to produce useful signal:
1. There has to be enough relationship to assess. A check after the first session is mostly noise — the client is rating a vibe. The first useful check is around session 4 (in a weekly cadence, that's a month into the work).
2. It can't be a default ritual the client tunes out. Running every session burns the signal — clients pattern-match to "another check" and click through. Every fourth session keeps the salience.
So: first check around session 4, then sessions 8, 12, 16. Some coaches add a check after any major engagement event — a difficult conversation, a goal pivot, a missed-session pattern — even if it doesn't fall on the every-fourth cadence.
How to use the score
The score itself is a starting point. The conversation it triggers is the point.
A few patterns worth recognizing:
Bond down, goals stable, tasks stable. Something happened in the personal relationship — a comment that landed wrong, a moment the client felt unseen, a tone shift. Worth surfacing directly: "I noticed our last few sessions have felt different. Anything come up that we haven't named?"
Bond stable, goals down, tasks stable. The client's underlying goal has shifted but they haven't said so. Common after a life event (job change, relationship change, health). Worth re-contracting: "Are we still working on the same thing?"
Bond stable, goals stable, tasks down. Your approach isn't fitting anymore — the questions feel rote, the structure feels constraining, the homework isn't sticking. Worth asking: "What would change if we worked differently for the next month?"
All three down. The alliance has fractured. This is rarely sudden; usually one dimension led and the others followed. A direct conversation about whether to continue the engagement is appropriate — ending well is better than coasting through.
What the check is NOT for
Coaches sometimes treat the WAI score like a performance review of themselves. It isn't. A 48/60 doesn't mean you're a bad coach; it means there's a specific dimension to attend to with a specific client.
A few things the alliance check is not:
Not a customer-satisfaction survey. Satisfaction is downstream of fit. A client who rates alliance high may not be progressing; a client who rates alliance moderate may be doing the deepest work of their life. Don't optimize for the score.
Not a contract renewal trigger. Don't tie next-session scheduling to alliance scores. The check should never feel like a vote of confidence.
Not the same as outcome. The alliance is a leading indicator of outcome, not the outcome itself. Goal scores (see Goal Attainment Scaling) and session feedback (see ORS+SRS) measure different things and shouldn't be combined.
A coach's protocol
A practical four-session protocol for using the alliance check:
- Session 4. First check. Note the score and dimensions. Don't bring it up unless something's notably off.
- Session 5-7. No check; do the work.
- Session 8. Second check. Compare to session 4. If anything's moved by more than 4-6 points on the total, or by more than 2-3 on a single dimension, it's worth a conversation.
- Session 9-11. Continue. If session 8 surfaced something, bring it back: "Last month we talked about X. How is it landing now?"
- Session 12. Third check. By now you have a trajectory. Three data points are enough to see whether the alliance is solid, slowly fracturing, or volatile.
A typical strong engagement looks like 50 → 53 → 54 over the first three checks. A typical struggling one looks like 48 → 44 → 40, and that 40 is the moment to either pivot the work meaningfully or end the engagement well.
How Arcline handles this
The alliance check is built in. Three sliders, 30 seconds, prompted every fourth session automatically:
- The client's My Progress page surfaces the check when it's due.
- The score is visible to the coach with the dimension breakdown.
- The trend is visible across all checks the engagement has logged.
- A "Watch" or "At risk" band on the roster surfaces clients whose last alliance check landed below threshold.
We deliberately hide the formal name "WAI-SR" from the client view — they see "Alliance check," and three short questions in plain language. The instrument is in the engine; the surface is uncluttered.
If you want to try it without committing, a 14-day trial gives you a full coaching cycle with the alliance check running, no credit card required.