How to set up a sales call coaching program that lifts win rates

Before a single call enters your coaching library, the consent policy has to be documented, reviewed by counsel, and operationally enforced. In the United States, roughly a dozen states require all-party consent, and most of Europe treats call recording as processing of personal data under GDPR. The baseline is a short verbal notification at the start of every outbound call, a one-click opt-out that stops recording in real time, and a retention window that matches your legal and privacy obligations. Document who can access recordings, how long they live, and the exact script reps read. Train reps on the script until it sounds natural, not legalistic.

• List every state and country your team dials into and tag each as one-party or all-party consent.
• Write the exact disclosure sentence reps use at call open, in plain, friendly language.
• Define the opt-out path: what reps say, how the recording stops, and where that event is logged.
• Set a retention window (90 to 365 days is typical) and document the deletion job that enforces it.

A good rubric is the contract between coach and rep. Keep it short, measurable, and tied to outcomes you can see in the pipeline. Four to six dimensions is the right ceiling: more than that and scoring becomes inconsistent, fewer and you miss the behaviors that actually move deals. Score each dimension on a simple 1 to 5 scale with written anchors so two different managers would land within one point of each other on the same call. Publish the rubric, review it with the team, and freeze it for a full quarter before you tune it.

• Talk ratio: is the rep listening more than they talk during discovery (target 40 to 55 percent rep talk time).
• Next-step clarity: did the call end with a specific meeting on the calendar, named attendees, and an agreed objective.
• Discovery depth: how many layers of impact questions landed on a quantified business consequence.
• Objection handling: did the rep acknowledge, isolate, and resolve objections, or did they default to feature dumps.

The most common failure mode in coaching programs is trying to review too many calls. Managers burn out, feedback gets shallow, and reps stop reading recap notes. Set a hard budget per rep: three to five calls per week, pulled by the rep from their own activity and surfaced to the manager for review. The rep picks because that drives ownership, but the manager retains the right to add one blind pull per week to catch what reps avoid showing. Keep the budget stable even in busy quarters. Coaching that pauses for a month never fully resumes.

• Give reps a Monday deadline to flag three calls from the prior week for review.
• Reserve one blind pull per rep per week where the manager picks a call the rep did not flag.
• Cap total manager review time at 90 minutes per rep per week including the 1:1 conversation.
• Track completion as a manager-level metric so coaching does not get crowded out by forecasting.

Generic feedback like "get better at discovery" does not change behavior. Specific feedback like "at minute 12 you asked three questions in a row without pausing, so the buyer never got to answer the first one" does. Every 1:1 coaching session should open with the rep self-scoring the flagged calls on the rubric, then jump to two or three timestamped moments the manager wants to work on. End every session with one behavior the rep will change on the next five calls and a note in the CRM so you can track whether the behavior actually shifted. Keep each 1:1 to 30 minutes.

• Have the rep self-score each flagged call before the 1:1 and send scores ahead of time.
• Pick two or three specific timestamps per call rather than reviewing full recordings end to end.
• Close with a single behavior commitment, written down, that the rep will apply on their next five calls.
• Log the commitment in the CRM against the rep profile so next week's 1:1 can measure follow-through.

Weekly team call reviews turn individual coaching into team learning, but only if the format protects psychological safety. Pick one call per week, rotate the submitter, strip or anonymize any sensitive buyer detail, and run a 30 minute session with a tight structure. The rep introduces context and what they want feedback on, the team listens to two or three clips, and feedback is framed as "what I would try next time," not "what you did wrong." Managers moderate for tone and keep anyone from monopolizing the air. The goal is pattern-level learning, not individual critique.

• Rotate who presents so no single rep becomes the perpetual example.
• Share the clips in advance so the live session is spent on discussion, not listening.
• Use a "what I would try next time" frame for all feedback to keep tone constructive.
• Capture the two or three team-wide takeaways in a running playbook doc the whole team can reference.

Conversation intelligence tooling can transcribe every call, tag talk ratio, flag monologues, surface filler words, and spot missing next steps automatically. Use it as a force multiplier, not a replacement for human coaching. The AI handles the mechanical scoring at scale so managers can focus their 90 minutes a week on the judgment calls humans do best: tone, trust, strategic framing, and buyer psychology. Describe the capability to reps honestly, confirm it respects the same consent and retention policy as human review, and make clear that AI scores inform the 1:1 conversation but never trigger automated performance actions.

• Pilot automated scoring on the mechanical rubric dimensions first (talk ratio, next-step presence, filler words).
• Spot-check AI scores against manager scores for a month before you trust them in reporting.
• Keep AI scores advisory in rep-facing dashboards and never attach them to compensation or PIP triggers.
• Document how AI-generated transcripts and scores are stored, retained, and deleted alongside the raw recordings.

Rubric scores are leading indicators. Pipeline outcomes are the proof. Pick two or three outcome metrics that should move if coaching is working (stage conversion rate, discovery-to-demo progression, average cycle time, win rate by segment) and track them against individual rubric trends across a quarter. Expect lag: coaching behavior shifts in weeks, pipeline math shifts in months. If a rep's rubric scores improve for a full quarter and their pipeline conversion does not, the rubric is scoring the wrong things or the rep is gaming the dimensions. Either way, that is a signal to iterate on the rubric or the coaching focus.

• Pair each rubric dimension with the pipeline metric it should influence (talk ratio to discovery-to-demo rate, next-step clarity to stage velocity).
• Review the paired trend together in a quarterly skill-lift report, not weekly where noise dominates.
• Segment by tenure: new hires should show the fastest rubric lift, tenured reps the slowest absolute gain.
• Celebrate improvements publicly and privately, not just the top absolute performers.

Freeze the rubric for a full quarter, then open it up for revision. Pull the quarterly skill-lift report, poll reps and managers on which rubric dimensions felt most useful, and swap in one or two new dimensions that reflect what the pipeline numbers tell you is now the binding constraint. Do the same with cadence: if 1:1s are landing but team reviews are stale, change the team review format. If reps are gaming the flag-your-own-calls rule, raise the blind-pull quota for a quarter. Treat the coaching program itself as a product you are iterating on with real users.

• Pull the quarterly skill-lift report and share it with the whole team before the retro.
• Run a 60 minute retro with reps and managers to vote on rubric changes for the next quarter.
• Freeze the new rubric for a full quarter before touching it again to keep scoring stable.
• Archive the old rubric and scores so year-over-year trends remain comparable.