After a roleplay session comes the question: "How did it go?" And then one of two things typically happens. Either someone says "Sounds good!" — which improves nothing. Or you get a vague "You should ask more questions," which may be true but provides no clear direction.
The real problem isn't a lack of feedback. It's a lack of structured feedback. As long as coaching relies on gut feeling, it remains subjective, non-repeatable, and nearly impossible to scale. Anyone who takes sales training seriously — whether with AI simulation or traditional coaching — needs a rubric: a set of observable criteria that show where a rep stands and what should be practised next.
A feedback rubric makes coaching objective, repeatable, and scalable. The crucial point: it describes observable behaviour, not personality traits. Not "Be more empathetic," but "Ask at least one question about business impact before presenting the solution."
Why gut-feeling feedback doesn't scale
In small teams, informal coaching often works surprisingly well. The experienced sales leader listens to a conversation, offers a concrete suggestion, and the rep implements it. The problem starts when the team grows or when coaching needs to come from more than one person.
Without a rubric, here's what happens: every coach evaluates by their own standards. What one person considers "good discovery" the other calls "too shallow." Feedback becomes unpredictable — and reps quickly learn that the assessment depends more on the assessor than on their performance. That destroys trust in the system faster than any other mistake.
With AI-powered training, the problem intensifies. When an AI simulator provides feedback, it must be based on transparent criteria — otherwise it's perceived as a black box that nobody trusts. The rubric isn't just a nice framework — it's the prerequisite for the entire training system to be accepted.
The three layers of a sales rubric
A good rubric doesn't evaluate everything at once. It separates three layers that build on each other:
Layer 1: Conversation structure. Does the conversation have a recognisable flow? Is there a clear objective, an agenda, and does it end with a concrete next step? Structure is the foundation — without it, even brilliant content falls flat because the customer can't follow the thread.
Layer 2: Content quality. Was the need actually understood? Was the value framed from the customer's perspective — not as a feature list, but as an answer to a real problem? Were objections recognised and addressed, rather than ignored or argued away?
Layer 3: Communication quality. How were questions asked? Open or closed? Was active listening evident — recognisable by the rep picking up on the customer's statements? Was the tone appropriate — neither too passive nor too aggressive?
Separating these three layers serves a practical purpose: a rep can be structurally strong but weak on content. Another asks brilliant questions but loses the thread. Only when the rubric separates these dimensions can feedback be targeted.
Example rubric: Discovery call
Instead of an abstract description, here's a concrete rubric for a discovery call. Each criterion is scored on a scale of 0 to 3:
Problem understanding. 0 = No attempt to understand the customer's problem. 1 = Superficial question about the problem with no follow-up. 2 = Asks about context and impact of the problem. 3 = Understands the problem, context, impact, and has led the customer to articulate the urgency themselves.
Stakeholder mapping. 0 = No question about the decision process or other stakeholders. 1 = Asks who else is involved. 2 = Asks about roles, decision criteria, and timeline. 3 = Has a clear picture of the buying committee and knows who might block the deal.
Success criteria. 0 = Not addressed. 1 = Asks generally what matters. 2 = Defines success criteria together with the customer. 3 = Has identified measurable success criteria and potential deal-breakers.
Next step. 0 = No next step agreed. 1 = Vague reference to "We'll be in touch." 2 = Concrete next step with a scheduled date. 3 = Mutually confirmed next step with a clear objective and preparation.
These four criteria are enough to evaluate a discovery call thoroughly. Nothing more is needed — and nothing more should be used, because too many criteria dilute the feedback.
Example rubric: Objection handling
For objection scenarios, the criteria shift:
Recognise and mirror the objection. Did the rep perceive the objection as such? Did they paraphrase it before responding? Mirroring signals to the customer that they've been heard — and gives the rep time to think.
Understand the objection type. Is it a price objection, a timing objection, a risk concern, or a genuine deal-breaker? The right strategy depends on the motivation behind the objection.
Reframe + evidence. Did the rep reframe the objection and support it with concrete evidence? Not "Our ROI is better," but "A customer in a similar situation reduced ramp-up time from 5 to 3 months — at your team size, that's roughly €120,000."
Close the loop. Did the rep check after the reframe whether the objection was actually resolved? "Does that make sense in your context, or do you see it differently?" — this check prevents objections from being swept under the rug rather than genuinely addressed.
If you want to apply this rubric in concrete scenarios, the article Practising objection handling with AI offers seven training scenarios with matching objectives.
Two focus areas instead of twenty scores
The most common mistake when introducing a rubric: evaluating too much at once. When a rep receives twelve scores after a roleplay, they don't know where to start. The feedback feels overwhelming — and gets ignored.
The better strategy: two focus areas per week. On Monday, the manager (or the rep themselves) decides which two criteria are the focus for the week. Every drill, every feedback session, every repetition concentrates on exactly those two points. Everything else is deliberately set aside.
After a week, the focus areas rotate. Over a month, the team covers eight different criteria — thoroughly, rather than superficially. And because the focus areas rotate, a spaced-repetition dynamic emerges naturally: what was trained in week 1 is revisited in week 5 — and the rep can see whether the improvement has stuck.
DACH governance: Rubric without surveillance
In the DACH region, any evaluation system is a question of trust. As soon as scores are linked to names and visible to managers, suspicion arises: Is this coaching or monitoring?
The solution lies in three principles:
Personal scores stay private. Each rep sees their own rubric results. The manager sees aggregated team data — averages, trends, distributions — but no individual scores. This protects the safe space and is compatible with most works council agreements.
Transparency about purpose and access. The rubric criteria are open. Everyone knows what's being evaluated and why. There are no hidden metrics, no surprises. Ideally, the team was involved in defining the criteria — or at least reviewed them.
Voluntary by default. Coaching sessions and drill results are learning moments, not exams. Anyone who wants to share their results with the coach or manager can do so. Anyone who doesn't, doesn't have to. This opt-in principle removes the fear — and paradoxically leads to more sharing, not less.
Conclusion
A feedback rubric isn't bureaucratic overhead. It's the tool that makes the difference between "We do roleplays" and "We develop skills systematically." It gives feedback direction, makes progress visible, and creates the foundation for coaching that doesn't rely on one person's gut feeling.
Implementation doesn't have to be complex: three layers, four to six criteria per scenario, two focus areas per week. That's enough to turn training into a genuine development programme.
sales-coach.ai evaluates roleplays against configurable rubric criteria — transparent, repeatable, and with concrete feedback per session. Personal scores stay private; team data shows trends. GDPR-compliant, works-council-compatible. Request a scorecard demo →