revenue - Home page(888) 815-0802
How to build a scorecard to improve win rates

How to Build a Sales Call Scorecard: Criteria and Calibration

Revenue Blog  > How to Build a Sales Call Scorecard: Criteria and Calibration
5 min readFebruary 20, 2026

To build a sales call scorecard, choose one call type, define observable behaviors, write an evidence rule for each criterion, and test whether reviewers apply those rules consistently. Use the result to agree on a specific coaching action, then review a later comparable conversation.

This guide covers rubric design and validation. For a ready-to-adapt discovery example, use our discovery-call scorecard template. For vendor selection, use the sales call scorecard software comparison. A rubric is a coaching aid, not a validated predictor of win rate.

1. Choose a call type and a coaching objective

Start with one role, call purpose, and behavior you want to examine. A prospecting call, discovery conversation, product demonstration, and late-stage evaluation need different expectations. Do not penalize a rep for failing to conduct full discovery during a short call intended to establish relevance.

Write the purpose before the questions. For example: “Help discovery reps explore the business impact of the buyer’s stated problem.” This is more actionable than “improve call quality” and easier to discuss than a total score with no explanation.

If your team follows MEDDPICC, SPIN Selling, or another methodology, identify the parts appropriate to this call. A discovery call cannot establish every qualification fact. Separate information the buyer supplied from a rep’s interpretation.

2. Review a representative set of conversations

Include useful, weak, and ambiguous examples from the same call type. Starting with a few strong calls can help generate ideas, but copying only winning deals risks confusing a rep behavior with favorable account conditions or buyer readiness.

  • Inspect what the rep asked and what the buyer actually answered.
  • Note where timing, buyer access, or the purpose of the call made a criterion inappropriate.
  • Include conversations with incomplete or unclear evidence so reviewers learn how to handle uncertainty.
  • Use only material your organization permits for this review and limit access to the appropriate reviewers.

If you ask AI to draft criteria, treat its output as a proposal. Check each question against the source conversation and your process. Do not submit customer material to an unapproved service or assume an AI-generated rubric has been validated.

3. Define one behavior and an evidence rule per question

“Did the rep conduct thorough discovery?” combines several judgments. “Did the rep ask about the effect of the stated problem on the buyer’s team?” describes a narrower action. Write down what counts, what does not count, and when the question is inapplicable.

Criterion Evidence rule Common mistake
Explored business impact The rep asks about an operational or business consequence of the problem. Counting a generic pain statement as evidence of an impact question.
Clarified the current approach The conversation identifies how the buyer handles the relevant work today. Assuming the current approach from the buyer’s industry.
Confirmed the next action The participants agree on an action and who will take it. Treating a seller’s proposed follow-up as a buyer commitment.

These are illustrative design examples, not a preconfigured Revenue.io rubric. Adapt the wording and response types to the scoring system you use. If a question is binary, define the evidence needed for Yes and No. Keep “not applicable” and “insufficient evidence” separate in your review process rather than forcing every missing answer into a rep failure.

For technical selling, a separate criterion may need to examine quoted specifications or application requirements. Use an appropriate subject-matter reviewer; conversation fluency does not establish technical accuracy.

4. Calibrate reviewers before trusting a total score

Have at least two reviewers independently assess the same sample, then compare the criterion-level decisions. Treat this as a starting exercise, not a statistically validated sample size. Discuss each disagreement before changing the rubric.

  1. Record the decision: Save the criterion, reviewer assessment, and supporting conversation passage or reference.
  2. Identify the disagreement: Was the wording unclear, evidence missing, call type wrong, or context interpreted differently?
  3. Refine the rule: Add a boundary example or split a compound question.
  4. Retest: Use different conversations to see whether the revision resolves the issue.
  5. Review AI separately: If automated scoring is used, compare its assessment with the reviewed evidence and record errors or unresolved cases.

Manager disagreement does not prove the technology is correct or incorrect. It can reflect unclear criteria, missing context, reviewer inconsistency, or a model error. Inspect the cause. Likewise, agreement on a score does not establish that the behavior causes revenue growth.

5. Turn the finding into one coaching action

Illustrative example: the buyer describes delayed onboarding, but its impact on the team remains unclear. A manager first checks whether impact was discussed elsewhere. If it was not, the rep practices a relevant follow-up such as “What does that delay prevent your team from doing?”

Agree when that question would help and when it would interrupt the conversation. Review the next comparable call for the intended behavior and the usefulness of the buyer’s response. A higher score without a better discussion is a reason to inspect the rubric, not declare success.

6. Confirm AI scorecard requirements

Revenue.io’s AI Scorecards and Coaching Feedback supports post-call evaluation with configurable criteria. Feature access depends on licensing, enabled settings, permissions, and qualifying conversations. Review the current product requirements before planning coverage.

Post-call evaluation is separate from Moments live guidance. A scorecard does not establish live prompting, complete capture of every call, or bias-free grading. Confirm the supported channels, processing behavior, and exception-review workflow during a pilot.

7. Maintain the rubric and measure results carefully

Name an owner and review criteria when your messaging, sales process, or buyer requirements change. Keep a version history so a rubric change is not mistaken for an improvement in rep performance. A criterion that almost always receives the same result may be a basic standard, an unsuitable requirement, or an uninformative question; inspect examples before removing it.

Track reviewer disagreements, missing assessments, useful coaching actions, and behavior in later comparable calls. Evaluate commercial outcomes separately, accounting for factors such as territory, deal mix, rep experience, and other programs. Do not promise a fixed win-rate gain, ramp-time reduction, or manager-capacity increase from the rubric alone.

For rollout planning, read how generative scorecards fit the coaching workflow. To evaluate Revenue.io, review packages and bring a sample criterion to a scorecard walkthrough. Ask the team to show an eligible assessment, a coverage exception, and how a manager reviews context.