A sales call scorecard is a fixed list of selling skills, each with a defined scale, that a manager or a scoring engine applies to one recorded conversation. It answers one question per skill: at what level did the rep perform this on this call? It is a different object from the seller scorecard that tracks meetings booked and quota attainment, and from the contact-center checklist that confirms script adherence. This page gives the template we use, the rules for scoring it, and the part most templates leave out: how to write a level descriptor that two people will score the same way.
Which scorecard this is
Three different things are called a sales scorecard.
| Type | Unit | Question it answers | Typical owner |
|---|---|---|---|
| Seller performance scorecard | One rep over a period | Is this rep on pace against targets? | Sales leadership, RevOps |
| QA or compliance checklist | One interaction | Did the agent follow the required steps? | Contact-center QA |
| Capability scorecard (this page) | One call, one skill at a time | At what level did the rep perform each skill? | Front-line manager, enablement, the coach |
The first tells you a rep is behind. The third tells you why, on a specific call, in terms the rep can change before the next one.
The template
The template below has twelve skills across the arc of a call, each scored on five levels. The scorecard we used in our own scoring research carries 20 skills; twelve is enough for a team starting out and the structure is the same. Level 1 is the lowest and level 5 is expert. Descriptors here are abbreviated to fit the table; the section after shows how to write the full version.
| Phase | Skill | Level 1 | Level 3 | Level 5 |
|---|---|---|---|---|
| Opening | Agenda and purpose | No agenda; launches into questions or pitch | States an agenda without confirming it with the buyer | Proposes an agenda, asks what the buyer wants from the call and adjusts to the answer |
| Opening | Framing | The buyer does not know why the rep is there | The rep explains why the call matters from the rep's side | The rep frames the call around the buyer's situation in the buyer's words |
| Discovery | Problem discovery | Takes the first stated problem at face value | Asks follow-ups and stops at the second layer | Reaches the underlying cause and who it affects |
| Discovery | Impact quantification | No attempt to size the problem | Asks for impact and accepts a vague answer | The buyer states a number, a date or a consequence in their own words |
| Discovery | Stakeholder mapping | Treats the contact as the decision | Asks who else is involved | Establishes who decides, who influences and how the decision gets made |
| Discovery | Listening and acknowledgement | Interrupts or moves on without acknowledging | Paraphrases some answers | Paraphrases accurately and builds the next question on the answer |
| Value | Linking capability to the stated problem | Generic feature description | Mentions the buyer's problem while describing the product | Every capability named is tied to a specific thing the buyer said |
| Value | Emotional control | Defensive or flustered under push-back | Stays calm but loses the thread briefly | Stays composed and uses the push-back to learn more |
| Objections | Handling objections | Argues or concedes immediately | Acknowledges and answers the stated objection | Finds what sits behind the objection and answers that |
| Objections | Negotiation | Discounts at the first mention of price | Holds price but cannot explain value | Trades on terms and anchors value to the buyer's own impact statement |
| Close | Gaining commitment | Ends with "I'll follow up" | Proposes a next step the buyer does not confirm | The buyer agrees a specific action, owner and date |
| Close | Summarizing and confirming | No summary | Summarizes the rep's own view | Summarizes in the buyer's terms and asks the buyer to correct it |
Score each skill in whole levels. Where the moment for a skill never arrives (no objection raised, no price discussed), mark it not applicable and leave it out of any average.
How to write a level descriptor
A level descriptor is a sentence that tells the scorer what that level looks like in a transcript. The usual way to write one is to describe good behavior in the abstract or to quote a textbook.
- 1Level 1Ends with "I'll follow up"
- 2Level 2Takes a "yes" as commitment, nothing scheduled
- 3Level 3Proposes a step the buyer does not confirm
- 4Level 4Next step, date and attendee agreed
- 5Level 5The buyer agrees an action, owner and date
In our call-scoring research, textbook examples added nothing to how closely scores matched the trained assessors. What moved the fit were marker reasons: the assessors' own notes, in their words, on why a particular call sat at a particular level. Adding marker reasons to the criteria lifted average fit with human scores from 0.62 to 0.72. The skills that improved most were framing, gaining commitment, emotional control and negotiation, the four where "good" is hardest to describe in the abstract.
Marker reasons for gaining commitment (illustrative, written for this page):
- Level 2: "The rep asked 'does that sound good?' and took the 'yes' as commitment. Nothing was scheduled and no one on the buyer side owns anything."
- Level 3: "The rep proposed a follow-up with a date. The buyer said 'sounds reasonable' and the rep moved on without confirming who would attend."
- Level 4: "Next step, date and attendee agreed. The rep did not ask what the buyer needed to see at that meeting for it to be worth their time."
- Level 5: "The buyer named the action, the date and the person they would bring. The rep asked what would make the meeting a success for them and wrote it back."
The procedure for producing a set like this:
- Take a batch of calls that your two most trusted managers have scored independently.
- Where they agree on a level, ask each to write two sentences on why. Keep the sentences verbatim.
- Where they disagree, get them on a call to talk it through and record the reason they settle on. That reason becomes the descriptor for the boundary between the two levels.
- Attach the collected reasons to the skill, grouped by level. The descriptor is now a set of real cases written by the people whose judgment you want to reproduce.
- Re-score a fresh set of calls and compare how often the two managers agree now with the first batch. In our call-scoring research, trained assessors with a shared scorecard still disagreed on about 18% of scores, so 18% is the number to beat, and the descriptors are the lever.
Scoring rules that keep the numbers honest
Whole levels only. In our call-scoring research, 37% of one production engine's scores were fractional, on a scale defined in whole levels. Rounding alone added 16 points of exact agreement with the human assessors. A 3.4 is a scorer hedging.
Not applicable is a result. When the trigger for a skill never arrives, the skill goes unscored. A skill that comes back N/A on most calls is either rare or defined too narrowly, and both are worth knowing.
One call is one data point. A rep who scores 2 on impact quantification on one call may have met a buyer who refused to give a number. Three calls at 2 is a pattern, and the pattern lives with the coach.
Score the whole call. In our call-scoring research, scoring single turns out of context ranked calls at between 0.57 and 0.63 against the assessors, compared with 0.89 for our final setup, which reads the whole call. Emotional control shows up in how the rep handles the second push-back, which only makes sense alongside the first.
Check every quote. If the scorer cites evidence, human or machine, the quote has to exist in the transcript. In the same research, with cheap evidence extraction, a fifth of extracted quotes were absent from the call.
Versions by call type
The twelve skills cover a full sales conversation, and not every call has every moment. One scorecard, with the skills that do not come up marked not applicable, covers most teams. Where a team runs very different kinds of call, three versions help:
| Call type | Skills that matter most | Skills often N/A |
|---|---|---|
| Discovery | Problem discovery, impact quantification, stakeholder mapping, listening, gaining commitment | Negotiation, often handling objections |
| Demo or evaluation | Framing, linking capability to the stated problem, handling objections, gaining commitment | Negotiation |
| Proposal or close | Summarizing and confirming, handling objections, negotiation, emotional control, gaining commitment | Problem discovery, if it was done earlier |
Keep the descriptors the same across versions. If gaining commitment means one thing on a discovery call and another on a close call, scores stop being comparable from one call to the next.
Should you weight the skills?
Weight the report, and leave the scoring alone. Score every skill on the same five levels, then decide which skills count most when you read the results.
The reason is practical. If a weighting is built into the scores themselves, a change in priorities next quarter changes every past score, and nobody can compare a rep with their own last month. Keeping the levels raw and applying weights in the summary lets you change the weights without rescoring.
Weighting by call type is the simplest scheme. On a discovery call, problem discovery, impact quantification and gaining commitment usually carry the most weight, because a call that misses those leaves the deal where it started. On a close call, negotiation and summarizing carry more. A simple rule: the skills that decide whether the deal moves forward should outweigh the skills that make the call pleasant.
Check your weights once against real outcomes. Take ten calls from deals that progressed and ten from deals that stalled, and see whether the weighted summary separates them. If it does not, the weights are on the wrong skills.
Versions by role
The same scorecard can serve SDRs, account executives and account managers, with different skills in front.
| Role | Calls scored | Skills to lead with |
|---|---|---|
| SDR or BDR | Cold calls and first meetings booked | Framing, problem discovery (first layer only), gaining commitment to a meeting |
| Account executive | Discovery, demo and close calls | The full card, weighted by call type |
| Account manager | Renewal and expansion calls | Problem discovery on new needs, stakeholder mapping, summarizing and confirming |
Keep the level descriptors identical across roles. An SDR at level 3 on gaining commitment and an account executive at level 3 should mean the same thing, so a rep's scores still make sense after a promotion.
Coaching scores and evaluation scores
Decide before the first call is scored whether scores are for coaching or for evaluation, and tell the team. Reps who think a call score feeds their performance review start choosing safe calls to share and stop bringing the ones that went wrong, which are the ones worth coaching.
We suggest keeping call scores in the one-on-one and out of reviews and rankings. Share patterns with the team (most of us stop at level 3 on impact this month) and keep individual scores between the rep and the manager.
Using the scorecard for practice
The same scorecard works on roleplay practice. When a rep lands at level 2 on gaining commitment, the coach can build a short roleplay from that deal so the rep practices the close before the next conversation, then debrief it on the same descriptors. Using one scorecard for practice and live calls means the rep is working toward the standard their manager uses.
Rolling it out in four weeks
Week 1: calibrate. Two trusted managers score the same five calls independently and note every disagreement of more than one level.
Week 2: write marker reasons for the week 1 disagreements, using the procedure above. A few skills usually cause most of them, so start there.
Week 3: score live calls. Each manager scores and debriefs one call per rep, and agrees one change for the next conversation with that buyer.
Week 4: check agreement on a fresh set of five calls. A skill where the gap has not closed needs more marker reasons.
After the first month, re-check agreement once a quarter and whenever a new manager starts scoring.
The 15-minute call review
A manager can review one call in about 15 minutes:
- Read or skim the transcript once without scoring, to get the shape of the call (3 minutes).
- Score each skill that applies, in whole levels, with the quote that decided the level (7 minutes).
- Pick the one skill where a change would matter most in the next conversation with this buyer (2 minutes).
- Write the question or line the rep could use on that call (3 minutes).
Steps 3 and 4 are the part the rep uses. A sheet of twelve scores with no next move gets read once and filed.
A worked example
An illustrative discovery call, 31 minutes, mid-market software. The buyer, a head of operations, opened by saying the current tool "is fine, mostly, but reporting is a pain".
Problem discovery, level 3. The rep asked what "a pain" meant and learned that monthly reports take two days to assemble by hand. The rep did not ask who assembles them or what happens when they are late. Marker reason: "Follow-up reached the second layer (the time cost) and stopped before the cause and the consequence."
Impact quantification, level 4. The buyer gave "two days a month" unprompted and the rep asked whose two days. Answer: a senior analyst. The rep did not ask what that analyst would otherwise do with the time. Marker reason: "A number and an owner from the buyer. The consequence was left on the table."
Gaining commitment, level 2. The rep closed with "I'll send over some info and we can go from there." The buyer said "great." Marker reason: "Vague next step, no date, no owner. The buyer's 'great' is politeness."
The debrief: the coach plays back the closing 90 seconds and names the level, and the rep hears it in the buyer's tone. The next conversation with this buyer is in six days. The coach and the rep agree the opening line for it, a question about who reads the monthly report and what they do with it, and a close that asks the buyer to name the person who would join a working session.
Setting it up in a spreadsheet
A shared sheet is enough to start, with one row per scored call:
- Call details: date, rep, buyer company, call type (discovery, demo, close).
- One column per skill: the level, 1 to 5, or N/A.
- One column per skill for evidence: the quote that decided the level.
- The one change: the skill picked for the debrief and the line the rep will use next time.
- Debriefed: the date the manager and rep talked it through.
Filter by rep and skill to see patterns, and count rows with no debrief date. Those are calls that were scored and never used.
A self-review for reps
Reps can score their own calls with the same card before the manager does. Ask them to pick the two skills they think went worst, give each a level with the quote, and write what they would say differently. When the rep's levels and the manager's differ by more than one, that gap is the first thing to talk about in the debrief.
Common mistakes
Adding a skill for every new initiative. Each added skill makes scoring slower and the debrief thinner. Retire a skill before adding one.
Averaging the card into one number. An overall 3.2 hides a rep who is at level 5 on problem discovery and level 1 on gaining commitment. Report skills separately.
Scoring the script. If a level depends on saying a set phrase, reps learn the phrase. Write descriptors around what the buyer said or did in response.
Editing descriptors without re-calibrating. Any change to a descriptor changes what a level means, so re-score a small set of calls before comparing new scores with old ones.
What the scorecard is for
A scorecard is a measurement. Coaching is what happens around it: a brief before the call on what the rep is going to try, observation of the call itself, and a debrief after it. A scorecard dropped in a folder with a 3 next to "handling objections" is a record nobody acted on.
The debrief is a call. The rep brings the deal to the coach and talks it through; the coach brings the scorecard and the marker reasons; together they find the moment on the transcript where the level was decided and agree what changes in the next conversation with that buyer.
On our platform an AI sales expert runs the team's own method on the live deal (Gap Selling, ValueSelling Framework®, MEDDPICC, or the team's own) and remembers every previous conversation about it, so the third time a rep lands at level 2 on gaining commitment the expert says so. A rep who cannot talk at that moment can type, and the expert picks up the same deal either way. How a scorecard is configured for a team is in the admin docs. To see the scorecard run on one of your own calls, talk to us.
FAQ
What is a sales call scorecard?
A sales call scorecard is a fixed list of selling skills, each with a defined level scale, applied to one recorded call. It answers "at what level did the rep perform this skill on this call?" for each skill. It is a different object from a seller performance scorecard, which tracks activity and quota over a period, and from a compliance checklist.
What should a sales call scorecard include?
A sales call scorecard should include skills across the arc of the call (opening, discovery, value, objections, close), each on a five-level scale with a written descriptor per level. Add marker reasons drawn from real calls your most trusted managers have scored, and a not-applicable option for skills whose moment never arrives. The template on this page has twelve skills; our research scorecard carries 20.
How many skills should a call scorecard have?
A call scorecard should have between ten and twenty skills, scored in whole levels. Fewer than ten tends to collapse distinct skills (framing and problem discovery, for instance) into one score. More than twenty makes a single call slow to score by hand and spreads the debrief too thin. Start at twelve and add from your own reviews of won and lost calls.
How do you score a sales call consistently across managers?
You score a sales call consistently across managers by writing level descriptors as marker reasons: your assessors' own notes on why a real call sat at a given level. In our research, descriptors written this way matched trained assessors more closely than textbook examples did (figures in how to write a level descriptor). Score whole levels on whole calls, and re-check agreement on a fresh set of calls each quarter.
Should a scorecard use a 1 to 5 scale or a 1 to 10 scale?
A sales call scorecard should use a 1 to 5 scale, scored in whole levels. In one production engine we tested, 37% of scores were fractional on a whole-level scale, and rounding alone added 16 points of exact agreement (scoring rules). Five levels with a written descriptor each is enough to separate a rep who stopped at the second layer of a problem from one who reached the cause.
What is the difference between a sales call scorecard and a sales scorecard?
A sales scorecard (sometimes called a seller scorecard) tracks a rep's activity and results over a period: meetings booked, pipeline created, quota attainment. A sales call scorecard scores the skills shown on one conversation. The first tells you who is behind. The second tells you what to change on the next call.
How is a call scorecard used in coaching?
A call scorecard is used in coaching as the standard under the debrief. The rep brings the call or the deal to the coach and talks it through; the coach names the level on each skill and shows the moment on the transcript that decided it; the two agree what changes in the next conversation. The scorecard measures; the coaching is what happens around it on that call.
How many calls should a manager score per rep?
One call per rep per week is enough for most teams, as long as each one is debriefed. A pattern needs three or more calls at the same level on the same skill, so a month of weekly scoring is usually enough to see one.
Can AI score sales calls with this scorecard?
AI can score calls against a scorecard like this one, once the setup has been tested against your own assessors. In our call-scoring research, the final setup matched the experts' ranking of calls at 0.89 rank agreement, while scoring single turns out of context dropped that to between 0.57 and 0.63. Test any engine on calls your managers have already scored before the team relies on it.