Sales training scorecard: criteria for scoring a practice call

Short answer

A sales training scorecard defines which criteria a practice conversation is scored on, with what weights and against which descriptions. A good scorecard has five to seven criteria, a one- or two-sentence observable description for each, weights that reflect the company's sales priorities, and a separate "good conversation" standard for each customer profile. With too many criteria employees lose track of what to focus on; without descriptions the score turns into opinion.

A practice scorecard differs from a real-call quality form: its purpose is learning, not control. Below are how to choose criteria, write descriptions and set weights, a sample table and typical mistakes.

Practice scorecard versus QA form

A form for quality assurance (QA) of real calls is built around compliance, critical errors and risk: did the operator give the mandatory information, did they use a banned phrase. An example is in the agent evaluation form.

A practice scorecard measures skill and shows development. Its criteria should answer "what did they do well, and what should they do differently?". So it has fewer mandatory items and more behavioural descriptions, and the outcome is not a penalty but the next session's topic.

Choosing criteria

A typical set for a sales conversation looks like this; adapt it to your sales model:

• Needs discovery: before the pitch, were open questions used to learn what the customer wants, why, their budget, timing and who decides?
• Product knowledge: is the information accurate, specific and relevant, with no invented facts?
• Objection handling: was each objection acknowledged, clarified, answered with substance and checked?
• Communication: listening, not interrupting, speaking clearly and politely, adapting to the customer.
• Sales structure: do greeting, discovery, presentation, objections and close follow a logical order?
• Next step: does the conversation end with an agreed, concrete step with a time?

Writing descriptions

The description, not the name, defines the scoring. "Communication" means something different to everyone; "doesn't interrupt the customer, reuses the customer's phrase in the reply, speaks briefly and clearly" can be observed. A good description has three properties:

1. 1ObservableBehaviour that can be seen in the text or transcript of the conversation.
2. 2One criterion, one ideaTwo different skills should not be mixed in one description.
3. 3Fits the contextDifferent descriptions for chat and calls: in chat "no walls of text", on a call "speaks with pauses".

Weights

A weight is a criterion's share of the total score and shows the company's priority. In a team where price objections often lose sales, giving objection handling more weight makes sense. How weights are calculated and the maths of a weighted score are explained in the weighted scorecard.

As an example, Vexvon AI Training's built-in rubric for call practice is: needs discovery 20, product knowledge 15, objection handling 25, communication 15, sales structure 10, next step 15 — 100 in total. It is a ready starting point; the company can set its own criteria and weights.

A standard per profile

Scorecard criteria can be the same for everyone, but a "good conversation" differs by profile. A good conversation with a price objector is one thing; with an investment buyer, another. So each profile gets its own standard, and the criteria are applied in the light of that standard. How a profile is built is in the AI customer profile.

Statuses: a score is not always needed

Not every criterion applies to every conversation. If the customer raised no objection, giving a low score for objection handling is unfair. The scorecard should have a status for each criterion: met, partly met, not met, not applicable. A criterion marked "not applicable" is removed from the total, not counted as zero.

Illustrative example: a results table

The result below is invented and only shows how the table is read. Profile: a customer who says "it's cheaper elsewhere".

• Needs discovery (20): partly — budget asked, decision-maker not. Note: "Moved to the pitch after the third message."
• Product knowledge (15): met — warranty terms stated accurately.
• Objection handling (25): partly — the objection was acknowledged, but what the competitor's offer includes was not asked.
• Communication (15): met.
• Sales structure (10): met.
• Next step (15): not met — ended with "I'll call you later".

In a table like this two lines matter more than the total: objection handling and next step. The next session's topic should be one of them.

Common mistakes

• Criterion names without descriptions.
• Copying the real-call QA form into practice unchanged.
• Fifteen criteria or more.
• No "not applicable" status.
• Changing criteria every month — progress cannot be compared.
• Leaving the weighting to the model's "judgement".

Limitations

Even the best scorecard does not capture all of a conversation's quality: a creative, non-standard but successful approach can score low. AI evaluation can apply a description wrongly, so results need human review and a way for the employee to challenge them. A scorecard score is a signal for practice, not the sole basis for a people decision.

The scorecard in Vexvon AI Training

In Vexvon AI Training the built-in rubric is the six criteria above, but the company can set its own criteria (up to 15), descriptions and weights. Weights come from your configuration, not the model: the model returns a score and a status per criterion, and the total is calculated with your weights. Each evaluation keeps the weights and rubric version it used. In call practice each criterion has a status — completed, partial, missed, not applicable — and "not applicable" drops out of the total. A test call is scored only if the profile has a good call standard written.

Next step

Write a one-sentence observable description for each of your current criteria. A criterion that cannot be described should leave the scorecard. For reading feedback see AI feedback, and for the link to real sales training score vs real sales; more articles are in feedback and coaching. To build it together, get in touch.