Sales Call Scorecard: A Template for Each Call Type, and How to Score Calls the Same Way
A sales call scorecard is a short list of things a rep should do on a call, each written so that two managers listening to the same recording give the same score. Use five to seven items, score each one 0, 1, or 2 against a written description, and keep a separate scorecard for cold calls, discovery calls, and demos. This guide gives a template for each, the weights, a way to calibrate your scorers, and a test for whether AI scoring can take over.
Discovery call · 31 minutes
7 of 10 · coach one thing
The call went well until the last minute: no date for the next meeting.
What Is a Sales Call Scorecard?
A sales call scorecard is a list of five to seven things a rep should do on one type of call, each with a written description of what a 0, 1, and 2 sound like, and a weight. A manager, a peer, or an AI listens to the recording and scores each item. The result shows which skill to coach, not just how good the call was.
It is not the same as a sales scorecard, which tracks a rep's activity and results over a month: calls made, meetings booked, pipeline, revenue. That tells you who is behind. A call scorecard tells you why, one recording at a time.
We read the pages that rank for this term in November 2026. Most are contact-center QA forms (greeting, compliance, soft skills, often scored 1 to 5 with no description of what a 3 means) or one template for every kind of call. They mention calibration in a line at most. None explains how to run it, how many calls a score needs before it means anything, or how to test an AI scorer before you trust it. Those are the parts this guide covers.
One disclosure: SnapSDR is an AI demo agent that runs on Tough Tongue AI's agent infrastructure. Tough Tongue AI comes up once, as one of several practice tools, and SnapSDR once, in the section on scoring demos an agent gives.
What Should a Sales Call Scorecard Include?
Five rules decide whether a scorecard gets used or ignored.
Score what you can hear. Every item should be something a listener can point to on the recording, with a timestamp. “Good discovery” fails that test. “Asked what the problem costs and got a number” passes it.
Five to seven items. Long QA forms with ten or more items turn listening into form-filling, and the scorer stops hearing the call. If an item never changes the coaching, cut it.
Three levels with written anchors. 0 means it did not happen, 1 means it happened in part, 2 means it happened fully, and each level is described for each item. On a 1 to 5 scale, two managers can agree that discovery was decent and still give a 3 and a 4. Three described levels leave much less room for that.
One item for the outcome. Every call type has a job: a meeting, a known problem, an agreed next step. Score whether it got done, and weight it heavily.
Weight what moves the deal. Not every item matters equally. Put the weight on the items that decide whether the deal reaches the next stage, and revisit the weights once you can compare scores with outcomes.
Rewrite Vague Items Before You Score Anything
Most scorecards fail on wording. Here are common items and the version two managers can score the same way.
| Vague item | What to score instead |
|---|---|
| Good discovery | Found what the problem costs, and got a number or a concrete example from the buyer |
| Active listening | Summarized the problem in the buyer’s words, and the buyer agreed |
| Handled objections well | Asked a question about the objection before answering it |
| Strong close | Ended with a meeting that has a date and named attendees |
| Product knowledge | Answered each question, or logged it with a promised follow-up date |
| Confident tone | Leave it out. Nobody can hear it the same way, so scorers disagree |
Sales Call Scorecard Templates by Call Type
A cold call, a discovery call, and a demo have different jobs, so one scorecard cannot judge all three. Scoring a cold call on how deep its discovery went punishes the rep for keeping it short, which is the point of a cold call. Pick a call type below, read the anchors, and copy the one you need into a spreadsheet.
Three scorecards, one per call type. Each item is scored 0, 1, or 2 against the descriptions shown.
Goal: know the problem, its cost, and how the buyer decides.
1. Agreed an agenda and the time available
10%0 No agenda
1 The rep stated an agenda
2 The buyer confirmed or added to it
2. Learned how they do it today, and who does it
15%0 Not asked
1 A one-line answer
2 The steps, the tools, and the people involved
3. Found what the problem costs, in the buyer’s numbers
25%0 Not asked
1 The buyer agreed it is a problem
2 A number or a concrete example: hours, money, a missed target
4. Learned who else decides, and how they will choose
20%0 Not asked
1 Names only
2 Names plus the steps or the criteria
5. Summarized the problem and got agreement
10%0 No summary
1 A summary in product terms
2 A summary in the buyer’s words, and the buyer agreed
6. Booked the next meeting with a date and the right people
20%0 No next step
1 Agreed to meet, no date
2 Date set and attendees named
Weights add to 100%. Score = sum of (score ÷ 2 × weight).
The weights are a starting point. The next step carries 20% to 30% on every card, because a call that ends without one tends to stall, whatever else went well. Discovery puts its largest weight on cost, because a problem nobody has put a number on is hard to get a budget for.
Demo calls deserve the most attention, because they are where most of the time goes. Our guide to demo-to-close rate shows how to tell whether your demos turn into revenue, and our product demo best practices cover how to build a demo around the buyer's problem, which is what the second demo item scores.
How Do You Get Managers to Score Calls the Same Way?
A scorecard is only fair if the score depends on the call, not on who listened. Before you use scores to compare reps, test that. Try it yourself first: score the excerpt below on three items from the discovery scorecard.
Score this excerpt from a discovery call on three items, then compare your scores with the anchors.
Rep: What happens today when a demo request comes in after hours?
Buyer: It sits until morning. Honestly, by then about half of them have booked with someone else.
Rep: Half? How many requests is that in a month?
Buyer: We get around 200. So, a lot.
Rep: Got it. Who else would look at a tool like this?
Buyer: My VP, Dana. She signs anything over ten thousand.
Rep: Great. I’ll send over some info and we can find a time next week.
Buyer: Sure, send it over.
Found what the problem costs, in the buyer’s numbers
Learned who else decides, and how they will choose
Booked the next meeting with a date and the right people
Illustration. The call excerpt is invented to show how written anchors settle a score.
Scorers usually disagree on the wording of the anchor rather than on what they heard. Fix that with a calibration session, which takes about an hour.
Pick three recent calls
One strong, one weak, one in between, all the same call type. Do not say which is which.
Everyone scores alone
Each manager scores all three on the same scorecard, without talking, and writes one line on why for every item.
Compare item by item
Put the scores side by side. Count the items where everyone gave the same score. That share is your agreement rate.
Settle each disagreement on the anchor
Read the anchor aloud and agree the right score. If two people read the anchor differently, rewrite the anchor, not just the score.
Repeat monthly with one call
New managers drift, and so do old ones. One shared call a month keeps everyone scoring the same way.
Our rule of thumb: use scores to compare reps only once managers give the same score on at least eight items in ten. Below that, use the scorecard for coaching conversations, where a one-point difference does no harm, and keep rewriting anchors.
How Many Sales Calls Should You Score?
Fewer than you fear for coaching, more than you think for ranking. Gong's help center suggests that frontline managers start with one call per direct report each week. That is enough to pick one skill and check next week whether it changed.
It is not enough to rank reps. One call is one buyer, one day, and one deal stage. A rep who scores 60% this week and 80% next week may not have changed at all. Compare reps on a month or more of scored calls, and on the same call type.
Choose calls by rule, not by request. If reps pick, they pick their best call, and you coach a version of them that rarely shows up. A fixed rule works better: the most recent discovery call each week, plus one won and one lost deal a month so you can see whether scores line up with outcomes.
The time adds up quickly. A 30-minute call at 1.5x speed takes 20 minutes to hear, before notes. One call a week for eight reps is close to three hours of listening. That cost is why teams look at AI scoring, and why the next section starts with how to test it.
Manual vs AI Call Scoring: Which Should You Use?
Four ways to get calls scored, and what each is good for. Most teams that score well use two: AI or a manager for coverage, and a manager for the coaching conversation.
| Method | Calls covered | Cost | Good at | Weak at |
|---|---|---|---|---|
| Manager scores | One or two calls per rep a week | About 20 minutes per 30-minute call at 1.5x speed, plus notes | Judgment, context about the deal, and a coaching conversation | Small samples, and managers drift apart without calibration |
| Peer review | Whatever the team agrees to swap | Rep time instead of manager time | Reps learn by hearing how others handle the same moments | Scores run kind; best used for discussion, not for ranking |
| Self-review | Any call the rep chooses | Almost none | Reps hear their own habits, which is often enough to fix them | Reps pick their best calls and score generously |
| AI scoring | Every recorded call that fits the filters | Software cost; no listening time | Coverage, and the same reading of an item every time | Items that need context, and errors nobody checks unless you test it |
Test an AI Scorer Before You Trust It
Vendors describe how their AI scores calls. Few tell you how often it agrees with your best manager on your calls. You can find out in a week:
- Have a calibrated manager score 20 recent calls of one type by hand.
- Run the same 20 calls through the AI with the same scorecard and the same anchors.
- For each item, count the calls where the two agree.
- Let the AI score the items where it agrees on 16 or more of the 20. Keep a person on the rest, and try rewording those questions.
Expect the split to follow the wording. “Was a next meeting booked with a date?” has an answer in the transcript. “Was the demo tied to the buyer's problems?” needs the discovery notes too, and a model that sees only this call can miss it.
Tools That Score Calls Against Your Scorecard
Three examples of the options, read from each vendor's own pages in November 2026. Prices change, so check before you buy.
| Tool | How it scores | Published price |
|---|---|---|
| Gong | AI Call Reviewer answers scorecard questions automatically on calls that match your filters. It scores the call host by default, skips any question it cannot find an answer to, and can back-score matching calls from the past 90 days. | Not listed on the pages we read |
| Avoma | Auto Call Scoring with AI and Custom AI Scorecards come in the Conversation Intelligence add-on, alongside AI coaching recommendations. | Add-on $29 per seat a month billed annually ($35 monthly), on top of a base plan from $19 per recorder seat a month billed annually |
| Outdoo | Scores discovery, demo, and closing calls against custom scorecards (MEDDIC, BANT, SPIN, or your own) and sends low-scoring behaviors to AI roleplay practice. | Not listed on the page we read |
Already on Gong: start with AI Call Reviewer on one call type, and use the test above on its first 20 scored calls. Note that it skips questions it cannot answer, so check how often each question comes back empty.
A smaller team watching cost: Avoma publishes its prices, and the scoring features sit in one add-on, so the total per seat is easy to work out.
You want scores to lead straight to practice: Outdoo links low-scoring behaviors to roleplay drills. Live prompts during the call are a different category, covered in our guide to real-time sales coaching.
How Do You Turn Call Scores Into Coaching?
A score is not feedback. “You got 64%” tells a rep nothing they can do on the next call. Turn it into one instruction.
Pick one item. Take the item with the biggest weighted gap across the rep's last few calls, not the lowest score on a single call. In the hero example above, that is the dated next step: worth 20% and scored 0.
Play the moment, then practice it. Play the 30 seconds where it went wrong, ask the rep what they would say instead, and have them say it out loud five times against a buyer who pushes back. A colleague can play the buyer. So can an AI roleplay tool such as Hyperbound, Second Nature, or Yoodli, or Tough Tongue AI, whose agents can also play the buyer in practice calls. Our sales role play scenarios include buyer briefs for most call types, and our roundup of Hyperbound alternatives compares the practice tools.
Score the same item next week. Only that one. If it moved, pick the next item. If it did not, practice again before you add anything new.
Score agents with the same card. If an AI demo agent such as SnapSDR gives your first demos, score its recordings on the demo scorecard you use for reps. It is the fairest way to compare the two, and it shows you which items to teach the agent next.
Frequently Asked Questions
What is a sales call scorecard?
A sales call scorecard is a short list of things a rep should do on a specific type of call, each written so that two people listening to the same recording give the same score. Managers use it to compare calls fairly, pick one skill to coach, and check whether coaching changed anything. It scores behavior on one call, not a rep's monthly results.
What should be on a sales call scorecard?
Five to seven items that a listener can hear on the recording, each tied to the goal of that call type. A discovery scorecard should cover the agenda, how the buyer works today, what the problem costs in the buyer's own numbers, who else decides and how, a summary the buyer agrees with, and a dated next step. Leave out items nobody can hear, such as confidence or energy, because scorers disagree on them.
How do you score a sales call?
Score each item 0, 1, or 2 against a written description: 0 means it did not happen, 1 means it happened in part, and 2 means it happened fully. Multiply each score by the item's weight, add them up, and divide by the maximum to get a percentage. Then coach the one item with the biggest gap rather than the total, because the total does not tell the rep what to change.
How many sales calls should a manager score each week?
Gong's help center suggests starting with one call per direct report each week. That is enough to coach one skill at a time, but not enough to rank reps: scores from one or two calls move a lot by chance. Compare reps only on a month or more of scored calls, and pick calls by a fixed rule, such as the most recent discovery call, rather than letting reps choose their best one.
Can AI score sales calls accurately?
It depends on the item. AI scoring does well on items with a clear answer in the transcript, such as whether a next meeting was booked with a date. It is less reliable on items that need judgment about context. Before trusting it, have a calibrated manager score 20 calls by hand and compare item by item. Keep AI scoring for the items where the two agree, and keep a person on the rest.
What is the difference between a sales call scorecard and a sales scorecard?
A sales scorecard tracks a rep's activity and results over time: calls made, meetings booked, pipeline, and revenue. A sales call scorecard grades what the rep did on one recorded call. The first tells you who is behind; the second tells you why, and what to coach.
Score What You Can Hear, Coach One Thing
A sales call scorecard works when anyone who listens to the call gives the same score, and when the score ends in one thing for the rep to change. Most scorecards fail the first test because their items describe feelings rather than things a listener can hear.
Our recommendation: start with the discovery scorecard, since discovery drives everything after it. Run one calibration session with your managers and reach eight in ten agreement before comparing reps. Score one call per rep a week, chosen by rule, and coach one item at a time. When listening time runs out, test an AI scorer on 20 calls against your best manager, and hand it only the items where the two agree.
Score an AI Demo the Way You Score a Rep
Bring your demo scorecard to a live SnapSDR demo. The agent joins Google Meet or Zoom, runs the demo in a live browser, asks discovery questions as it goes, and hands qualified deals to your team with the answers attached.
Related Reading
SnapSDR Team
AI Sales Automation Experts