Pulse Reviews and Analysis
How do you coach reps to act on AI call-coaching feedback?
Curated by Kory White · Fractional CRO, CRO SyndicateDirect Answer
To coach reps to act on AI call-coaching feedback, separate the machine's job from yours: let tools like Gong AI, Chorus, and AI scorecards flag *what* happened on the call, then sit beside the rep and coach *why it matters and what to do differently*. In 2027, AI is the tireless observer, not the coach — it surfaces patterns (low talk-ratio, no next-step booked, monologuing through pricing) at scale, but reps only change when a human manager turns one flagged pattern into one concrete rep, runs a role-play, and checks the next three calls.
Pick one behavior per rep per cycle, co-watch the actual clip, agree on a single change, and verify it on the next live call. Reps ignore AI feedback when it arrives as a wall of red metrics with no human owning the change; they act on it when their manager makes it personal, small, and tracked.
Why This Happens — Diagnose Before You Coach
Reps don't act on AI feedback for four very different reasons, and the fix for each is different. Coaching the wrong cause is why "we bought Gong and nothing changed" is so common.
- Knowledge gap — the rep doesn't understand *why* the flagged behavior hurts (e.g. A 70% talk-ratio in discovery). They need teaching, not nagging.
- Skill gap — they understand it but can't yet execute the better behavior live. They need role-play and reps, not another dashboard.
- Will / belief gap — they think the AI is wrong, the score is unfair, or "that's not how I sell." They need a credibility conversation and proof from their own deals.
- System / overload gap — the AI surfaces 15 things at once, the rep is buried, and nothing gets prioritized. They need *you* to pick one.
A common 2027 failure mode is over-reliance: managers forward the AI scorecard and assume the software coached the rep. It didn't. AI flags the behavior; the human coach drives the change. The diagnosis below routes you from the symptom ("rep isn't improving on the AI metric") to the real cause.
The Coaching Conversation
Use the GROW model — Goal, Reality, Options, Will — and anchor every step to a real clip from the rep's own Gong or Chorus library. Reps dismiss abstract metrics; they cannot dismiss the sound of their own voice. Open the 1:1 with the rep watching, not you lecturing.
Goal — set the focus before opening the dashboard:
"Before we look at the AI feedback, what's the one part of your calls you most want to get better at this month? I want us to pick a single thing and actually move it, not boil the ocean."
Reality — co-watch one clip and let the rep self-assess first:
"Gong flagged that on the Acme discovery call your talk-ratio was 68% and there was no next step booked. Let's watch the two minutes around pricing together. ... Okay — what do you notice? What would you do differently if you ran that moment again?"
Letting the rep diagnose first is the highest-leverage move in the room. If the rep names the gap, it's coaching; if you name it, it's criticism. When they spot it themselves, you skip the defensiveness entirely.
Options — generate the better behavior together, don't hand it over:
"What's one question you could have asked there that would've flipped the ratio and surfaced their timeline? ... I like that. Here's one more I use: 'What would have to be true for this to be a priority this quarter?' Which feels more like you?"
Will — lock a single, testable commitment:
"So the change is: on your next three discovery calls, you book the next step out loud before you hang up, and you keep your talk-ratio under 55% in discovery. I'll review those exact three calls in Gong on Friday and we'll watch the best one together. Deal?"
Notice what this script does NOT do: it doesn't dump every AI flag, it doesn't blame the tool, and it doesn't let the manager off the hook for follow-through. The AI gave the data point; the human turned it into one rep-owned commitment with a verification date.
The Coaching Plan / Cadence
Acting on AI feedback is a loop, not an event. Run it weekly per rep, and structure the first 90 days of any rollout so the team learns to trust the tool instead of resenting it.
30 days — build trust in the signal. Co-watch one AI-flagged clip per rep per week. No leaderboards yet. The goal is for every rep to believe the feedback is fair before you ever attach it to accountability.
60 days — one behavior per rep. Each rep owns a single flagged behavior (talk-ratio, next-step rate, discovery question count, monologue length). You verify it on real calls every Friday and log whether it moved.
90 days — rep-led review. Reps bring their *own* best and worst AI-scored call to the 1:1 and self-coach; you add the nuance the model misses (deal context, buyer mood, strategy). This is where the team stops needing you to police the dashboard.
The loop only works because a human owns steps B through E. Hand those to the software and you get a noisy dashboard nobody reads.
Drills & Role-Play
Behavior change happens in reps before the live call, never during it. Run these short, specific drills:
- The clip-pair drill. Pull one of the rep's calls where the flagged behavior was bad and one where it was good (or borrow a top rep's clip from Chorus). Watch both at 1.25x. Ask: "What's different in the first 90 seconds?" Reps learn faster from contrast than from a score.
- The 55% talk-ratio role-play. You play the prospect. The rep runs discovery for five minutes with a hard rule: ask three questions before making any statement. You time it. Repeat until it's natural.
- The next-step close. Drill only the last 60 seconds of a call: the rep must book a specific calendar next step out loud, ten times in a row, with you throwing a different objection each time.
- The AI-disagreement drill. When a rep insists the AI scored them unfairly, have them pull the exact moment and argue it. Half the time they're right (context the model missed) and half the time they hear themselves and concede. Both outcomes build trust in using the tool.
- Scorecard calibration. Once a month, the whole team scores the same call using your AI scorecards rubric, then compares to the AI and to each other. This is how reps internalize the standard instead of gaming the metric.
What to Measure
Track leading indicators of behavior change, not just quota, or you won't know the coaching worked until it's too late.
- The targeted behavior itself — e.g. Talk-ratio trending from 68% toward 55% over the rep's next ten discovery calls. This is the cleanest proof the AI feedback was acted on.
- Next-step booked rate — percentage of calls ending with a confirmed calendar next step, straight from Gong.
- Feedback-to-change lag — how many days between a flag and a measurable change on live calls. Shrinking lag means your coaching loop is tightening.
- Self-coaching rate — how often reps surface their own AI-flagged moments before you do. The strongest signal that the team has stopped resenting the tool.
- Conversion movement — discovery-to-demo or demo-to-proposal rates for the coached behavior, watched over a quarter, not a week.
Hold lagging quota in your peripheral vision but coach to the leading behavior. A rep can't control whether a deal closes this month; they can control their talk-ratio on the next call.
Common Mistakes Managers Make
- Forwarding the scorecard and calling it coaching. The email is not the intervention. Over-reliance on the tool is the single biggest failure of AI coaching rollouts — the software flags, the human coaches.
- Dumping every flag at once. Fifteen red metrics paralyze a rep. Pick one. Your job is prioritization the model can't do.
- Coaching to the deal, not the skill. "What should we do on Acme?" rescues the rep on one deal. "What pattern does the AI keep flagging in your discovery?" builds a skill that wins ten deals.
- No verification. Agreeing on a change without booking a date to re-watch the next three calls means it never happened. Always close the loop.
- Coaching everyone identically. The AI gives every rep the same metrics; you must not give every rep the same conversation. A knowledge gap and a belief gap need opposite moves.
- Treating a performance problem as a coaching problem. If a rep won't engage with clear, fair feedback over a full cycle, that's a documentation-and-deadline conversation, not more role-play.
FAQ
How do I get a rep to stop ignoring AI feedback they think is wrong? Make them argue it. Have the rep pull the exact flagged moment and explain why the AI is mistaken. Often they're partly right — the model missed deal context — and acknowledging that buys credibility.
Just as often they hear themselves and concede. Either way they engage with the tool instead of dismissing it, which is the real goal.
Should reps watch their own calls or should I do it for them? Reps watch first and self-diagnose; you add what they miss. When the rep names the gap, it's ownership; when you name it, it's a verdict they'll resist. Co-watching the clip together in the 1:1 is the format that consistently changes behavior.
How many behaviors should a rep work on at once? One. AI tools surface dozens of signals, and that abundance is exactly why reps freeze. Pick a single behavior per rep per cycle, verify it on live calls, then move to the next. Sequential beats simultaneous every time.
Will AI eventually replace the sales manager's coaching? No. AI is the best call observer ever built, but it can't read a wavering rep's confidence, judge deal strategy, or earn the trust that makes someone change. In 2027 the model is the data layer; behavior change still runs through a human relationship.
How do I roll this out without the team feeling surveilled? Spend the first 30 days co-watching clips with zero accountability attached — pure development, no leaderboards. Let reps prove the feedback is fair to themselves before you ever tie it to numbers. Trust in the signal has to precede pressure on the signal.
What if a rep acts on the feedback but their numbers still don't move? Check whether you coached the right behavior. If talk-ratio improved but win-rate didn't, the real gap may be qualification or next-step discipline, not discovery. Re-run the diagnosis — and if the behavior is genuinely fixed and results still lag, look at fit, territory, or pricing before you coach harder.
Bottom Line
AI call-coaching only changes behavior when a human owns the change. Let Gong, Chorus, and your AI scorecards flag the pattern at scale, then do the irreplaceable part yourself: pick one behavior, co-watch the rep's own clip, role-play the fix, and verify it on the next three live calls. The tool is the observer; you are the coach.
Sources
- Gong Labs — What separates top reps (call analytics research)
- Harvard Business Review — The Best Sales Managers Don't Chase Revenue
- RAIN Group — Sales Coaching Research and Best Practices
- Sales Hacker — How to Build a Sales Coaching Program
- The GROW Model — Performance Consultants (Sir John Whitmore)
- Chorus by ZoomInfo — Conversation Intelligence for Coaching
- Salesforce Blog — Sales Coaching Tips for Managers
*Sales coaching for AI call-coaching feedback — how to coach reps to act on Gong and Chorus AI feedback, sales manager coaching guide, rep coaching framework, and an AI call-coaching playbook for 2027.*