What scoring a sales call involves
Scoring a sales call means rating how well the rep performed each skill the call needed, in whole levels, with a quote from the call behind every level, and then choosing the one skill to work on before the next conversation on that deal. A front-line manager usually does it. Allow the length of the call plus about ten minutes once you know the scorecard.
- 01Name the job of the call
- 02Pull the stage scorecardFour to eight skills
- 03Listen once without scoringMark the moments that matter
- 04Climb the ladder, quote each levelStop at the first no; whole levels
- 05Score whether it did its jobSeparate from the skill scores
- 06Choose one skill for next timeTied to a named meeting
The method is the same at every stage of the deal. What changes is which skills the call should show: a cold call and a demo need different criteria. This page is one part of the AI call scoring guide; for first meetings in depth, see How to score a discovery call.
The method draws on our research program: August to October 2026, more than 5,000 tests and more than 250,000 scoring decisions, with blind marking by PeerLab's trained assessors against a 20-skill capability scorecard (field guide).
Skills to score at each stage
Build one skill model for the team, then score each call on the four to eight skills its stage calls for. Think of it as a shared core plus a module per call type.
| Stage of the deal | Job of the call | Skills to score |
|---|---|---|
| First discovery | Find a problem worth solving and who owns it | Problem discovery, impact, stakeholder mapping, agenda, gaining commitment |
| Second meeting or demo | Show the buyer their problem solved | Linking to the buyer's stated problem, handling questions, confirming fit, gaining commitment |
| Champion follow-up | Equip the champion to sell internally | Decision process, stakeholder mapping, risk discovery, gaining commitment |
| Pricing or proposal | Tie the price to the cost of the problem | Value link, objection handling, emotional control, negotiation |
| Close | Agree the path to signature | Paper process, decision criteria, negotiation, gaining commitment |
Gaining commitment appears at almost every stage because every call should end with a dated next step the buyer agreed to.
A scorecard template to copy
Every skill on the scorecard needs each level written out, so two managers reading the same call land on the same number. Here is one skill done in full, as a pattern for the rest:
SKILL: Problem discovery (first discovery call)
1 The rep described the product; the buyer never stated a problem.
2 The buyer named a problem in general terms ("reporting is slow").
3 The buyer described the problem in their own words, with an example
from their work.
4 Level 3, plus the buyer said what the problem costs or what happens
if it stays.
5 Level 4, plus the buyer said who else is affected and why it matters
now.
EVIDENCE: the line from the call that earned the level, with timestamp.
NEXT: the question that would have reached the next level.
Each level adds one thing the buyer did, which is what makes the ladder in step 4 work: the scorer asks whether the call reached each level in turn. Write levels about what the buyer said or agreed to wherever you can. "The rep asked good questions" is a judgment; "the buyer named a cost" is something a second manager can find in the transcript.
The eight steps
1. Name the job of the call
Before you listen, write one line: what should this call have moved on the deal? "Get the CFO's view of the business case" is a job. "Catch up" is not. Good looks like: a sentence that the call either achieved or did not.
2. Pull the scorecard for that stage
Use the stage table. Do not score skills the call had no reason to show. A first discovery call that never touched pricing is not weak on negotiation. Good looks like: four to eight skills, each with every level described.
3. Listen or read once without scoring
Mark the moments that matter: where the buyer named a problem, raised a concern, mentioned another person, or agreed to something. Scoring while listening for the first time pulls you toward the last thing you heard. Good looks like: six to twelve timestamps or quotes.
4. Climb the level ladder for each skill
For each skill ask in turn: is this at least level 2? At least level 3? Stop at the first no. Our AI scoring method asks the same questions, 80 per call (field guide). Score whole levels only. Good looks like: one whole number per skill, reached by a yes at each level below it.
5. Write the quote behind each level
Copy the exact line from the transcript that earned the level, or the moment the next level was missed. A score without a quote turns into an argument about memory. Good looks like: every score has a line a rep can find in the recording.
6. Score whether the call did its job
Go back to step 1. Did the buyer commit to the thing the call was for? Keep this separate from the skill scores, because a good call can fail its job (the CFO canceled) and a weak call can succeed. Good looks like: yes, partly or no, with the agreed next step and its date.
7. Choose one skill for the next conversation
Pick the lowest-scoring skill that matters for the next meeting on this deal. If the next call is with procurement, work on negotiation before rapport. Good looks like: one skill, one question or move the rep will use, tied to a named meeting.
8. Check yourself against another manager
Once a month, score five calls with a peer, blind, then compare the order you put them in. Trained experts in our research disagreed on 18% of individual scores, so expect some. The order matters more than any single number: in our tests, a scorer could be close on average and still put calls in the wrong order. Good looks like: you and your peer put the calls in close to the same order, and you write down the reasons behind the scores where you differed.
Worked example: a pricing call
The deal, names and scores are illustrative. An AE is on the pricing call for a warehouse software deal. The job of the call (step 1): get the operations director to agree that the price is justified by the cost of late shipments the director described in discovery.
The manager pulls the pricing scorecard (step 2) and marks four moments on a first listen (step 3): the director says "that's more than we planned," the rep offers a discount, the director asks about implementation time, and the call ends with "send me the proposal."
The ladder (step 4) and quotes (step 5):
| Skill | Level | Evidence |
|---|---|---|
| Value link | 2 | Rep restated the product's features after "more than we planned"; never returned to the cost of late shipments |
| Objection handling | 2 | Offered 10% off within a minute of the objection without asking what "planned" meant |
| Emotional control | 3 | Steady when the director asked about implementation time: a plain answer with a realistic go-live week, no hedging |
| Negotiation | 1 | Gave a concession and got nothing back |
Did the call do its job (step 6)? No. "Send me the proposal" has no date and no agreement on value.
The one skill (step 7) is value link, because the proposal will be judged against budget unless the cost of the problem is on the table. Before the follow-up, the rep gets on a call with a coach, talks the deal through and leaves with one move: open the next call by restating the director's own figure for late shipments and asking whether it still holds.
How reps can score their own calls
Reps can run the same ladder on their own calls, and it is one of the cheapest ways to build the habit. Ask the rep to score one call a week before the 1:1, using the same scorecard and the same rule: a quote behind every level. Then the manager scores the same call blind.
The interesting part is the gap between the two. When a rep scores a call 4 on problem discovery and the manager hears a 2, the rep usually remembers the buyer saying more than they did. Reading the quotes together settles it in a minute. Over a few weeks the two scores tend to move closer, and the rep starts hearing the missing level while still on the call.
Keep self-scores out of performance reviews. Reps who know the number will be judged mark themselves up, and the exercise stops teaching anything.
How many calls to score each week
One scored call per rep per week is enough for coaching, if it is the right call. Pick the call that matters most for the next conversation on a live deal, usually the last call before a decision meeting. For a manager with eight reps, that is eight calls a week, each taking the length of the call plus about ten minutes.
Scoring more calls by hand adds little. The pattern in a rep's skills shows within a few weeks of one call each, and the time is better spent on the coaching that follows step 7. Teams that want every call scored, for example to spot which deals need attention, use a calibrated AI scorer for coverage and keep the manager's hand-scored call for coaching.
Is call scoring micromanagement?
Call scoring is not micromanagement when it leads to one coached skill and a better next call. It turns into micromanagement when the scores are used to rank reps, posted on a board, or collected with nothing coming back to the rep.
Two habits keep it on the right side. Share the quotes along with the score, so the rep can see exactly which moment earned it. And tie every scored call to a conversation before the next meeting on that deal, as step 7 says. Experienced reps benefit as much as new ones, especially when the method or the product changes, though they need to see that the scorecard describes the selling the team agreed is good.
Where it goes wrong
Scoring the outcome. A call that ended well pulls every skill score up with it, and a deal that later closed makes its early calls look better than they were. Step 6 exists to keep outcome and skill apart.
Half scores. A 3.5 means the scorer did not decide. In our research, 37% of one production engine's scores were fractional on a whole-level scale, and rounding alone added 16 points of exact agreement with experts.
One total. Adding skills into one number hides which skill to coach, and a rep who is strong on discovery and weak on commitment looks average on both.
Vague levels. A level described as "good" invites every manager to read it differently. Giving our AI scorer the reasons trained assessors wrote for each level lifted its average fit with human scores from 0.62 to 0.72 in our tests, the biggest single gain we measured.
A score that goes nowhere. A scorecard in a folder is a measurement. If step 7 does not lead to a conversation before the next meeting, the score changed nothing.
Scoring every call for every rep
The method above takes a manager's time, so it works for a handful of calls a week. To score every call, a calibrated AI scorer can run the same ladder. Our final method changed 1.7% of scores on a re-run, and averaged with a second model's blind whole-call score it ranked calls in the same order as trained experts at 0.89. In the study, that final method cost about $0.007 per call to run for scores and feedback, with every quote checked against the transcript.
With our AI sales experts, the score feeds the coaching call: the rep talks the live deal through with the expert and prepares the next conversation against the standard your managers set. Test any scorer first; Can AI score a sales call? shows how. To have one of your own first meetings scored against your standard, talk to us.
FAQ
What criteria should I use to score a sales call?
Use the four to eight skills that call's stage should show, drawn from one skill model for the team. Discovery calls focus on problem discovery, impact and stakeholders; pricing calls on value link, objection handling and negotiation. Add gaining commitment to almost every call, since each should end with a dated next step.
How do you grade a sales call fairly?
Score each skill in whole levels by asking whether the call reached each level in turn, write the quote behind every score, and keep the call's outcome separate from the skill scores. Then check your order of calls against another manager's once a month, since even trained experts disagree on some scores.
Should I use a 1 to 5 scale or pass/fail?
Use a described 1 to 5 scale for skills with real levels of quality, such as discovery depth, and pass/fail only for steps that either happened or did not, such as a required disclosure. Every level needs a written description, or managers will read the middle of the scale differently.
How long does it take to score a sales call?
Plan on the length of the call plus about ten minutes once a manager knows the scorecard. Most of the time goes into the first listen and writing the quotes. Calibrated AI scoring cuts that to a review of the AI's evidence, which lets the manager spend the saved time on coaching.
Can I use the same method for every type of sales call?
You can use the same method for every type of sales call. The eight steps stay the same at every stage; only the skills change. A first discovery call is scored on problem discovery and stakeholders, a pricing call on value and negotiation. That way a rep's level on a shared skill, such as gaining commitment, means the same thing across the deal.