Most coaches set goals with their clients. Far fewer have a way to score whether those goals are actually moving. The gap matters: a coach who can only describe goals can't tell you, at session 12, whether the work has produced more progress than would have happened anyway.
Goal Attainment Scaling — GAS — is the small bit of structure that closes the gap. It's been validated in clinical research since the 1960s and has been quietly underused in coaching ever since. This is what it is and how to use it without becoming a researcher.
What GAS is, in one paragraph
For each goal, you write five short descriptions of what observable behavior would look like at five levels: far behind expectations, a little behind, expected, ahead, and far ahead. You and the client agree on those descriptions before any work starts. Then, at every session (or whatever cadence makes sense), you score where the client actually is, on a scale from −2 to +2. Over time, the score line tells you whether the goal is moving.
That's it. The whole technique is "describe the levels, then score against them."
Why GAS over a SMART goal
SMART goals (specific, measurable, achievable, relevant, time-bound) describe the target. They don't describe the trajectory. A SMART goal tells you whether the client hit the bullseye; GAS tells you whether they're getting closer.
A SMART goal might read:
"Schedule three informational interviews with director-level peers by July 1."
That's a clean, measurable target. But three months in, if the client has done two interviews, what does that mean? Are they on track? Behind? Ahead, given how the role market is right now? Is the meaningful progress in the interviews themselves, or in the way the client now talks about their career? SMART can't say.
GAS would describe the same goal across five levels. Something like:
- −2 (far behind): Avoids the decision; gathers no evidence.
- −1 (a little behind): Lists options but doesn't test assumptions.
- 0 (expected): Chooses a path with coach-reviewed criteria.
- +1 (ahead): Chooses a path and secures two sponsor conversations.
- +2 (far ahead): Chooses a path, secures sponsors, and starts a transition plan.
Now at session 6, when the client says "I had a useful chat with my manager and I think I'm leaning toward staying," you have a clear way to score: that's a +1, because they've made the decision and started a sponsor conversation. Last session was 0; the trajectory is up.
The score is comparable across goals (every goal uses the same five-level scale) and across clients. You can finally answer "are my clients getting more out of this work than they were last quarter?" with data, not vibes.
Why GAS has stronger evidence than alternatives
Three properties make GAS unusually robust:
1. Anchored, not abstract. Each level is described in observable behavior, not in feelings or self-rated confidence. Two different coaches scoring the same client tend to land within one level of each other — far higher inter-rater reliability than open-ended progress descriptions.
2. Personalized but comparable. Each client's −2 to +2 is specific to their goal. But the scale itself is fixed. So you can aggregate scores across clients without losing the personalization.
3. Sensitive to small change. A client who moved from −2 to −1 made meaningful progress. A binary "did they hit the goal" framing would call this a failure. GAS catches the early movement, which is most of what coaching produces.
Meta-analyses across rehabilitation, mental health, and educational coaching put GAS effect sizes at d = 0.6-1.0 — large by most behavioral-research standards. The catch isn't the technique; it's writing levels well.
How to write the five levels
This is the part that takes practice. The levels work when they describe specific, observable behavior; they fail when they describe feelings, intentions, or vague achievements.
A few principles, with examples.
Level 0 is "what would I expect this client to achieve in this engagement?"
Not the moonshot. Not the floor. The honest expectation if the work goes well but normally. For a six-month engagement with a senior IC who wants to move to manager, level 0 is probably "received feedback from manager and identified two leadership behaviors to practice." Not "promoted." Not "considered the idea."
Level −2 should describe genuine stuckness, not "did nothing"
A common mistake is writing −2 as "no progress." That's not useful — clients rarely make literally zero progress. A better −2 describes the pattern you'd see if the work isn't taking. For a goal about delegation: "Continues to personally resolve team escalations." That tells you something. "Did not delegate" tells you nothing.
Level +2 should be possible but unlikely
If +2 is impossible, you'll never see it and it's noise. If +2 is the obvious next step, you'll see it on every plateau and lose the ability to recognize a real breakthrough. A useful +2 happens for maybe 10-15% of clients. For a delegation goal: "Stakeholders report faster decisions and stronger team ownership."
Each level should be one sentence, observable, and self-checkable
If the client can't read a level and tell you whether they're at it, the level is too abstract. Compare:
Bad: "Has greater confidence in their leadership." Better: "Took the lead on the strategy review without asking the manager to review beforehand."
The second one passes the "would two coaches agree?" test. The first one doesn't.
Avoid measurement hell at the extremes
Some coaches try to put metrics into every level: "10 informational interviews", "5 stakeholder conversations". Resist this for the extremes (−2 and +2). Specific numbers at the floor and ceiling create perverse incentives — clients chase the count instead of the underlying behavior. Keep numbers if they genuinely track the thing you care about, drop them if they're proxy metrics.
How to use GAS in a session
Once the levels are written, the in-session protocol is:
- Open with the goal. "Let's revisit the director path goal. Last session you were at 0 — chose a path with criteria. Where are you now?"
- Use the levels as a structured probe. "What evidence are you noticing? Have you tested any assumptions yet? What's stopping the sponsor conversation from happening?"
- Score collaboratively. The coach scores. You can ask the client what they think first, but the score isn't a vote. The coach is the one with the rubric and the evidence base.
- Note what shifted the score. A score going from 0 → +1 isn't useful unless you and the client know what action moved it. That action is what to repeat. A score going 0 → −1 isn't useful unless you know what got in the way.
The protocol takes 3-5 minutes. Done at every session, it's the difference between a coach saying "you're doing great" because the client looks animated, and saying "you moved from 0 to +1, here's exactly what changed" with three months of evidence.
How many goals per engagement
The honest answer: 2-4. Three is the sweet spot for most adult learners.
One goal is too narrow — coaching almost never produces change in a single dimension. Five+ goals overwhelms the cadence; you can't score five goals every session in any depth, and you'll find yourself scoring some of them by inertia.
Some practices use a primary goal + secondary goals model, where the primary gets scored every session and secondary ones every 2-3 sessions. That works, but it adds bookkeeping. Three equally-weighted goals, scored every session, is the simpler default.
What to do with the scores once you have them
This is where most coaches stop, and where the technique pays back the most. The scores are signals about your practice, not just signals about the client.
A few patterns worth watching:
A goal that stays at −1 for four consecutive sessions is rarely the client. It's almost always either a goal that wasn't well-anchored to start with (rewrite it with the client) or a goal that's blocked by something outside the engagement (acknowledge it explicitly).
A goal that jumps from 0 to +2 in a single session is suspicious. Either the level descriptions weren't well-calibrated (revise them) or something happened outside the work that the score is over-attributing to your coaching.
A goal that stays at 0 for the entire engagement is a "met expectations and didn't move" — useful to acknowledge as a stable baseline, but a signal the work didn't produce above-baseline change. That's not a failure, just honest data.
A pattern across clients where most goals plateau in months 2-3 points at something in your coaching style, not at the clients. Coaches who notice this and adjust their cadence — more concrete experiments, more accountability between sessions, more specific feedback — typically see the plateau lift by month 4-5.
A worked example
Here's how a goal might look in practice, anchored properly.
Client: Senior individual contributor, 6-month engagement, wants to be promoted to manager.
Goal: Build a leadership signal at work that managers and peers can see.
- −2: No leadership feedback collected.
- −1: Feedback requested from one stakeholder.
- 0 (expected): Feedback collected from manager and two peers.
- +1: Feedback converted into three leadership behaviors that are being practiced.
- +2: New behaviors observed by manager and peer group.
Session-by-session, the score might progress: −2 → −1 → −1 → 0 → 0 → +1 → +1 → +1 → +2. The plateau at −1 is real (the client got stuck on whom to ask). The plateau at +1 is the productive grind of practicing the behaviors before observers notice. Both are useful information for the coach.
How Arcline handles this
In Arcline, every goal is GAS-anchored by default. When a coach adds a goal, the form requires the five level descriptions before saving. At every session, the coach scores against those descriptions on a −2 to +2 scale. The arc plots over time, and the roster screen surfaces any goal that's been static for too long.
You don't have to know the term "Goal Attainment Scaling" to use Arcline. You just write five honest descriptions and score progress every session. The technique is in the engine; the surface stays plain English.
If you want to try the discipline without committing to anything, a 14-day trial lets you set up a real client with anchored goals and run a coaching cycle.