Sales Coaching Framework: Diagnose, Demonstrate, Drill, Deploy
A four-stage coaching operating model — the methodology underneath ThinkWork's competency rubric work — adapted as a standalone system any manager can install to close one named skill gap at a time.
What's inside
- Diagnose stage: manager actions and exit criteria
- Demonstrate stage: manager actions and exit criteria
- Drill stage: manager actions and exit criteria
- Deploy stage: manager actions and exit criteria
- Cadence guidance for how long each stage should realistically take
- Worked example carried through all four stages
- Common failure mode for each stage
Most coaching fails not because managers don't care, but because feedback is delivered without ever being rehearsed, and skills are "covered" without ever being drilled to fluency before they're needed live. This framework installs four sequential stages, each with an exit criterion — you don't move to the next stage on a schedule, you move when the evidence says the rep is ready.
Stage 1 — Diagnose
Purpose: Pinpoint the single highest-leverage skill gap, backed by evidence — not by impression.
Manager actions:
- Review 2-3 recent calls using a structured rubric (not a general "how did it go" review)
- Compare the rep's pattern against their own historical baseline, not just against other reps
- Resist the urge to name three gaps — pick the ONE gap that, if closed, would move the most other metrics
Exit criteria: You can name one specific skill gap, with a specific piece of evidence (a quote, a timestamp, a scored rubric dimension) anchoring it — not a general impression like "needs to handle objections better."
Common failure mode: Diagnosing from a single bad call, or from a gut feeling not tied to any specific evidence — this produces coaching that feels arbitrary to the rep and gets contested rather than accepted.
Stage 2 — Demonstrate
Purpose: Show the rep what the skill actually looks like executed well, so they're working toward a concrete standard, not a verbal description of one.
Manager actions:
- Model the specific skill live, in a role-play, or via a recorded example (your own call, a peer's call, or a library clip)
- Narrate the decision points as you demonstrate — what you noticed, why you responded that way, not just what you said
- Ask the rep to articulate, in their own words, what was specifically different from how they currently do it
Exit criteria: The rep can describe the specific difference between the demonstrated version and their current version, unprompted — not just "that was good."
Common failure mode: Skipping straight from Diagnose to Drill without ever showing the rep a concrete standard — reps then practice toward an abstraction and plateau below where a clear model would have taken them.
Stage 3 — Drill
Purpose: Build the skill to fluency in low-stakes reps, before it's live in front of a real buyer.
Manager actions:
- Run the specific skill through 3+ role-play repetitions using scenarios matched to the gap
- Give immediate feedback after each rep — don't stack up feedback for the end
- Increase difficulty each repetition (easier buyer objection first, harder pushback last)
Exit criteria: The rep hits a consistent scoring threshold (e.g., 3+/4 on the relevant rubric dimension) across at least 3 consecutive role-play repetitions — not just once.
Common failure mode: Declaring the skill "drilled" after a single good repetition — one clean rep under low pressure doesn't predict performance under real, higher-pressure conditions.
Stage 4 — Deploy
Purpose: Apply the drilled skill live, on a real call, with accountability tied back to the original diagnosis.
Manager actions:
- Set a specific pre-call commitment with the rep ("today, isolate before you respond to any pushback")
- Observe or review the resulting call using the same rubric used in Diagnose
- Close the loop explicitly: did the specific gap identified in Stage 1 show up differently?
Exit criteria: The skill shows up unprompted on a real call, scored against the same rubric used at Diagnose, with a measurable improvement over the original evidence.
Common failure mode: Moving on to a new skill gap immediately after one good live rep, without confirming it holds up across 2-3 calls — a single successful deployment can be luck, not fluency.
Cadence Guidance
| Stage | Typical duration |
|---|---|
| Diagnose | 1 call review session (15-20 min) |
| Demonstrate | 1 session (15-30 min) |
| Drill | 1-2 sessions across 3-5 days |
| Deploy | 1-2 real calls across 1-2 weeks, then loop back to Diagnose |
A full cycle on one skill gap typically runs 2-3 weeks. Running more than one skill gap through the cycle simultaneously is the most common reason this framework fails to produce visible improvement — pick one gap, run it all the way to Deploy, then start the next cycle.
Worked Example — Closing the "Reframe" Gap
- Diagnose: Call review shows the rep acknowledges and isolates objections well (rubric scores of 3), but reframes with generic proof every time (rubric score of 2) — evidence: three separate calls, same pattern, verbatim quotes pulled
- Demonstrate: Manager plays back a peer's call where the same price objection was met with a specific, relevant customer comparison instead of a generic reassurance; rep identifies the difference as "they used a real example instead of a general statement"
- Drill: Rep runs the Price & Budget scenarios from the Role-Play Script Library three times, each time swapping in a specific, relevant proof point instead of a generic one, scoring 3+ on Reframe by the third repetition
- Deploy: Rep commits pre-call to using one specific proof point on the next live price objection; call is reviewed against the rubric; Reframe score moves from 2 to 3 on the real call — cycle closes, next gap (Isolate, still occasionally skipped under pressure) becomes the new Diagnose target
How to use it
Run one rep through all four stages on ONE named skill gap at a time, using the exit criteria to decide when to move to the next stage rather than a fixed calendar, then loop back to Diagnose for the next gap.