What turning a score into coaching means
Turning call scores into coaching means choosing one skill from a rep's scores, tying it to a live deal, and running a short conversation that ends with the rep's next move on that deal. A front-line manager can do it in about 20 minutes per rep once the scores exist.
Most teams now have the scores, from a conversation intelligence tool or a scoring add-on reading recorded calls. The step that decides whether any of it changes the next call is the conversation, and most of what gets written about it stops at "coach the low score". This piece is the procedure for that conversation. It sits within our AI call scoring guide, which covers how the scores themselves are produced and checked.
How to turn call scores into a coaching conversation in seven steps
1. Read the pattern across calls
One call is one data point. A rep who scores 2 on impact quantification once may have met a buyer who refused to give a number. Three calls at 2 is a pattern. Look at the last three to five scored calls of the same type before choosing anything. Good looks like: a short list of skills that are low on the rep's recent calls.
2. Pick the skill the deal needs next
From that list, choose the one skill that matters most to a deal the rep is working right now. A low score on differentiation matters less this week than a low score on stakeholder mapping in a deal where the buyer just mentioned a new approver. Good looks like: one skill, one named deal, one reason connecting them.
3. Check the evidence before using it
Open the call and read the quotes behind the score. In our scoring research, with cheap automated extraction a fifth of quotes were not in the call; every newer model generation marked reps harsher than the human assessors; and 37% of one production engine's scores were fractional on a whole-level scale (field guide). Coaching on a score that is wrong costs the rep's trust in every score after it. If your scorer has not been measured against your managers, calibrate it first. Good looks like: you have heard the moment yourself and agree with the level.
4. Open with the deal
Start with the deal and the rep's read of it: "Where's that deal at? What do you still need from them?" The number can wait. A rep asked about a score starts defending it. Good looks like: the rep names the gap in the deal before the skill comes up.
5. Explain the level with the reason
Play or read the moment, then explain the level the way an expert marker would: what the call showed and what the next level up sounds like. This is the marker reason. In our research, giving the scorer trained assessors' marker reasons lifted average fit with human scores from 0.62 to 0.72, more than any other change, because a reason says why a call sits where it does. The same reason tells a rep what to do differently. Good looks like: "That's a 2 on stakeholder mapping because the buyer mentioned the board and we moved on. A 3 asks who else is involved. A 4 learns how they decide."
6. Rehearse the next move on the deal
Turn the gap into the specific thing the rep will say or send next on this deal. Have them say it out loud, then once more in their own words. Saying the line is the rehearsal. Good looks like: a line in the rep's words, tied to a date on the deal.
7. Score the next call on that skill
After the next conversation on the deal, check that one skill again. Movement of one level is a result; no movement is a reason to try a different approach to the same skill. Good looks like: a short running note per rep: skill, deal, score before, score after.
Worked example
This example is illustrative; the people, deal and scores are invented.
Lena manages six AEs. Her scorer flags that Dev's last four discovery calls scored 2, 2, 3 and 2 on gaining commitment, and his active deal with a hospital group has gone quiet since a good first call.
Steps 1 to 3. The pattern is real across four calls. The hospital deal needs a next step more than anything else. Lena opens the last of the four calls and finds the moment: Dev closed with "I'll put something together and send it over." The quote is in the transcript and the level is right.
Step 4. Lena asks: "Where's the hospital deal at?" Dev: "They liked it. I sent the deck, haven't heard back." Lena: "What did they agree to do next?" Dev pauses. "Nothing, really."
Step 5. Lena plays the last minute of the call. "That's a 2 on commitment. You proposed a step and they didn't confirm anything. A 3 gets a date. A 4 gets a date, who's coming, and what they need to see."
Step 6. They plan an email that asks for a 30-minute session with the buyer and their IT lead in the next two weeks, and the line for the next live conversation: "Before we hang up, can we put the follow-up on the calendar now, and who else should be there?" Dev says it twice.
Step 7. On the following call, gaining commitment scores 4: date, attendees and an agenda item from the buyer.
Closing the feedback loop between sales calls
The feedback loop between sales calls is closed when the skill coached after one call is checked on the next call with the same buyer, and the result decides what gets coached after that. Most teams break the loop in one of two places. The coaching happens, and nobody looks at whether the next call changed. Or the next call is scored on everything, and the one skill that was coached disappears into the total.
A simple running note keeps the loop closed. One line per coaching conversation:
Rep | Deal | Skill | Level before | Line rehearsed | Next call date | Level after | Next step
Dev | Hospital group | Gaining commitment | 2 | "Can we put the follow-up on the calendar now?" | 14 Nov | 4 | Move to stakeholder mapping
Read it at the start of each one-on-one. If the level moved, the rep moves to the next skill the deal needs. If it did not, look at the call before trying again: did the rep use the line, and what did the buyer say? A line that was used and still did not land needs rewriting. A line that was never used needs another rehearsal, with the rep saying it in their own words.
Using scores across the team
The same scores answer a different question when a manager reads them across all reps at once. If one rep is low on a skill, that is a coaching conversation. If most of the team is low on the same skill, that points to training, a playbook gap or a hiring profile, and a one-on-one will not fix it.
Look across the team once a month. Sort skills by how many reps scored low on them over the last few weeks. A skill near the top for most reps is a topic for the team meeting, with a good call from a peer as the example. Skills that are low for one or two reps stay in their one-on-ones. Keep the two lists apart, because a team session on a skill most reps already have wastes the time of everyone who has it.
Where it goes wrong
Leading with the number. "Your discovery score was 2.4" starts a debate about the scorer. The deal starts a conversation about the buyer.
Coaching the average. An overall score hides the one skill a deal is short of. Coach the skill and leave the total alone.
Coaching a harsh scorer's opinion. If your scorer marks harder than your managers, every debrief becomes a telling-off. Newer models in our research did exactly that, and the gap widened with each generation.
Assigning practice and calling it coaching. Some tools close the loop by assigning a practice roleplay built from the low score, and practice has its place. A low score on a live deal is fixed on that deal, in the next conversation with that buyer, which is where the rehearsal in step 6 points.
Coaching too many skills at once. Three skills in one conversation produce agreement and no change. One skill, one deal.
Coaching from scores for every rep
A manager with eight reps can run this conversation for a few of them each week. The rest wait, and their scores sit unread. That gap is manager capacity, and it is the gap our AI sales experts fill.
The rep brings the live deal to Caty on a call. She works in the team's own method, has the scorecard and the last call's evidence, and follows the same steps: the deal first, the skill it needs, the reason behind the level, the next move rehearsed out loud. A rep who cannot talk at that moment can type.
Managers get a read across the team from Bryce, the manager coach, and spend their own time where judgment matters most. The steps of the session itself are in How to review a sales call, the one-page version is the call review checklist, and the coaching practice around it is the sales coaching framework. Talk to us.
FAQ
How do you use call scores for coaching?
Read a rep's scores across several calls to find a pattern, pick the one low skill that a live deal needs next, and check the evidence behind it. Open the conversation with the deal, explain the level using the reason behind it, and rehearse the rep's next move on that deal out loud. Re-score the next call on that skill.
Should I share AI call scores with reps?
Yes, once the scorer has been checked against your managers. Shared scores from an unmeasured scorer erode trust quickly, especially if the model marks harder than people do, which newer models did in our research. Share skill scores with quotes and the reason behind each level, and talk about them in the context of a deal.
How many skills should one coaching conversation cover?
One coaching conversation should cover one skill. A conversation about one skill on one deal ends with a specific next move the rep can try. Covering several skills tends to produce agreement in the room and the same call next time. Note other low scores and come back to them once the first one moves.
What if a rep disagrees with their call score?
Play the moment and read the quote together. If the quote is wrong or the level does not fit the descriptor, fix the score and tell whoever owns the scorer. If the score holds, move to the deal: what would a higher level have got from this buyer? The point is the next call.
How quickly should coaching follow a scored call?
Before the next conversation on the same deal, ideally within a few days. Coaching that arrives after the deal has moved on teaches a skill in the abstract. Scores that land the same day make it possible to prepare the rep for the very next call with that buyer.
Can AI coach reps from call scores?
AI can coach reps from call scores if the scores are calibrated and the coaching works the live deal. In our scoring research, the final method ranked calls like expert assessors at 0.89 and changed 1.7% of scores on a re-run. A coach using those scores can run the deal-first conversation with every rep on a call, which no manager has the hours for.