QA managers and sales enablement leaders designing call scorecards face the same problem: most scorecards measure activity rather than behavior. A scorecard that checks whether the rep introduced themselves and attempted a close tells you those things happened, not whether they were executed well enough to advance the conversation. Effective scorecard design measures specific behavioral execution, not checkbox completion.

This guide covers how to design scorecards that produce coaching-ready data rather than compliance reports.

Why most scorecards produce low coaching value

The three most common scorecard failures:

  1. Criteria too broad to be actionable: "Communication skills" as a scored criterion tells an agent nothing they can improve. "Asked at least one open-ended clarifying question before proposing a solution" is specific enough to change behavior.

  2. Equal weighting where unequal weighting is warranted: A compliance disclosure failure is categorically different from a suboptimal closing question. Scorecards that weight both equally produce scores that obscure what actually matters.

  3. No connection to training: If QA scores are reported to managers without triggering coaching assignments, the scorecard measures quality without improving it. Scorecard design is incomplete without a defined workflow from score to coaching action.

Step 1: Define the behavioral outcomes your training is intended to produce

Before designing scorecard criteria, identify what your training program is teaching. Scorecard criteria should directly evaluate whether training-targeted behaviors appear in live calls. If your training program teaches objection handling using a three-step acknowledgment-reframe-redirect sequence, your scorecard should evaluate whether reps are executing that sequence, not just whether they "handled the objection."

For sales calls, training-linked scorecard criteria typically include:

  • Discovery question quality (open-ended, probing, customer-revealing)
  • Objection acknowledgment before response (did the rep reflect the concern before redirecting?)
  • Value statement relevance (was the value proposition matched to the customer's stated need?)
  • Closing question directness (did the rep explicitly ask for next steps?)

For support calls, criteria typically include:

  • Empathy acknowledgment timing (within first 30 seconds of problem statement)
  • Resolution completeness (was the stated problem fully resolved?)
  • Proactive escalation criteria (did the rep surface related issues or risk before the customer had to?)

Step 2: Write behavioral definitions for each criterion

Every criterion needs a behavioral definition: what "good" looks like and what "poor" looks like. Without these definitions, two reviewers scoring the same call will reach different conclusions on ambiguous criteria.

A well-written behavioral definition for "Discovery question quality":

Good: Rep asks at least two open-ended questions before proposing a solution. At least one question directly addresses the customer's primary motivation for the inquiry.

Poor: Rep moves to solution proposal within the first 2 minutes without asking clarifying questions, or asks only closed questions (yes/no) that do not surface customer context.

This level of definition produces consistent scoring whether the reviewer is a human QA analyst or an AI scoring engine. Insight7 uses this format, main criterion, sub-criteria, and a context column defining good and poor performance, to produce scoring that aligns with human QA judgment after a calibration period of 4 to 6 weeks.

Step 3: Assign weights that reflect actual business impact

Score weighting should reflect the relative importance of each criterion to your specific business outcomes. A compliance requirement for a regulated financial services contact center might appropriately receive 25 to 30% of the total score. In a sales environment, closing behavior might warrant 20 to 25%. The weighting logic should be defensible: if a manager challenged the weights, you should be able to explain why each criterion carries its percentage.

Common weighting structures for sales calls:

  • Compliance and disclosures: 20-30%
  • Discovery quality: 15-20%
  • Objection handling: 20-25%
  • Value communication: 15-20%
  • Closing execution: 15-20%

Total: 100%

Step 4: Configure both script-based and intent-based evaluation

Some scorecard criteria require verbatim compliance, a specific disclosure must be read in specific language. Others require intent-based evaluation, whether the rep achieved a goal matters more than the exact words used.

Effective scorecard design distinguishes between these two evaluation modes:

  • Script-based (compliance): "The rep stated the required disclaimer within the first 90 seconds of the call." AI can evaluate this with high accuracy through phrase detection.

  • Intent-based (behavioral): "The rep demonstrated empathy when the customer expressed frustration." AI evaluates this based on the semantic content and tone of the response, not exact phrase matching.

Insight7 supports per-criterion switching between script-based and intent-based evaluation, compliance items use exact-match checking while conversational quality items use intent-based evaluation.

Step 5: Connect scorecard results to training assignments

A scorecard that produces a score without triggering a coaching action is a reporting tool, not a training improvement tool. The scorecard-to-training connection requires:

  • A defined threshold below which a coaching assignment is triggered (for example: any dimension scoring below 60% on two consecutive calls)
  • A mapping of each scorecard criterion to a specific training module or practice scenario
  • A follow-up measurement: did QA scores on the coached criterion improve after training?

Insight7's coaching module automates this connection. When a rep scores consistently below threshold on a scorecard criterion, the platform generates a targeted practice scenario for that specific skill and queues it for supervisor approval. QA score trends on the coached criterion are tracked over subsequent calls to measure training application.

Step 6: Calibrate scorecard criteria with your QA team before full deployment

Before deploying a new scorecard at scale, run a calibration session:

  1. Select 10 to 20 calls representing the range of quality your team encounters
  2. Have two to three QA reviewers score each call independently using the new criteria
  3. Compare scores and identify where reviewers disagreed
  4. Clarify criterion definitions wherever inter-rater agreement is below 80%

Calibration catches definition ambiguity before it produces inconsistent scoring data. For AI scoring engines, the same calibration logic applies: Insight7's criteria tuning process compares AI scores against human QA judgment and adjusts criteria definitions until alignment is reached.

What is the right number of criteria for a sales or support call scorecard?

Between 5 and 8 core criteria is the practical range. Below 5, the scorecard is too coarse to identify specific development priorities. Above 8, scoring time increases and inter-rater reliability declines. If compliance requirements produce a longer list, consider grouping related items into parent criteria with sub-criteria rather than expanding the main scorecard. ICMI research on contact center QA shows that scorecards with 6 to 8 well-defined criteria produce more actionable coaching data than longer, broader scorecards.

How do you prevent scorecard criteria from becoming outdated?

Review criteria quarterly against your training content and business outcomes. If your product or sales methodology changes and scorecard criteria do not update to match, reps are scored on behaviors that training no longer teaches. The practical test: for every scorecard criterion, identify the training module that teaches it. If no module maps to a criterion, either update training or retire the criterion. Insight7's configurable criteria let QA teams update definitions and weights without platform rebuilds. According to Training Industry research, scorecards tied to current training content improve training application rates by 30% compared to generic QA frameworks.

How Insight7 applies scorecard design to 100% of calls

Insight7 applies configurable scoring rubrics to every recorded call. Weighted criteria with behavioral definitions, script-based and intent-based evaluation modes, and evidence-backed scoring that links every criterion rating to the specific transcript quote that supported it. QA managers click through to verify any score without listening to full recordings.

When scorecard data identifies training gaps, Insight7 connects those gaps to targeted coaching assignments. The scorecard does not just measure training application, it drives the next training cycle. See how scorecard-to-coaching works in Insight7.

FAQ

How many criteria should a sales or support call scorecard have?

Between 5 and 8 criteria is the practical range for scorecards that produce actionable coaching data without creating reviewer fatigue. Below 5 criteria, the scorecard is too coarse to identify specific development priorities. Above 8 to 10, scoring time per call increases significantly and inter-rater reliability tends to decline. If you have more than 8 essential criteria, consider whether some can be grouped into parent criteria with sub-criteria rather than expanding the main list.

Should different call types have different scorecards?

Yes, in most contact center environments. Inbound service calls have different quality criteria than outbound sales calls. Compliance conversations have different requirements than escalation handling. Running all call types through the same scorecard produces scores that are hard to interpret because the criteria are not relevant to all call types equally. Insight7's dynamic evaluation auto-detects call type and routes the correct scoring rubric, eliminating the need to manually sort calls before scoring.

How do you know if your scorecard is measuring what training intended to teach?

Compare scorecard criteria to your training content. Every criterion on the scorecard should correspond to a behavior your training program specifically teaches. If a criterion appears on the scorecard but has no corresponding training module, agents have no development pathway when they score poorly. If your training teaches behaviors that do not appear on the scorecard, those behaviors are not being measured in live calls, creating a gap between what training promises and what QA verifies.

Designing scorecards that connect training to live call performance? See how Insight7 applies weighted behavioral criteria to 100% of calls and automates the connection from scorecard gaps to coaching assignments.