Every practice call on Sunnyside ends with a report and a score out of 10. This document explains where that number comes from, what it measures, and why you can trust it.
A score you can defend
Two design principles sit underneath everything:
1. Every claim is backed by evidence from the call. When the report says you surfaced a buying signal, missed a question, or handled an objection well, it cites the exact words from the transcript with a timestamp. Quotes are verified against the recording by the system itself — if a claimed quote doesn't actually appear in the transcript, the claim is discarded. The report can never tell you something happened that didn't.
2. The final number is computed by a fixed formula, not by an AI's opinion. AI does the listening: specialist analyzers read the transcript and identify what happened — which questions you asked, which objections came up, how the buyer responded. But the score itself is calculated from those findings by a published, deterministic formula. Run the same call twice and the same findings produce the same score. You can click into any part of the score and see exactly which moment drove it.
The four pillars
The score blends four things, because a good call is more than textbook technique:
| Pillar | Weight | What it measures |
|---|---|---|
| Outcome | 30% | Did the conversation actually move forward? |
| Engagement | 25% | Did the other person lean in — or check out? |
| Execution | 25% | How well did you run the play for this type of call? |
| Fluency | 20% | Did you use your product and domain knowledge where it counted? |
Outcome is the largest slice on purpose. Calls exist to create movement. The system classifies how the call ended — a committed next step with a time on the calendar scores highest, forward movement without a firm commitment scores well, genuine interest with no movement scores in the middle, and a flat ending scores low.
Engagement reads the other side of the conversation. Buying signals — follow-up questions, requests for specifics, forward-looking language from the prospect — lift the score. Threads they opened that you never came back to pull it down. This rewards reps who create real two-way conversations rather than delivering monologues.
Execution is where the coaching framework lives. Each call type has its own rubric (more on this below), and each rubric dimension is scored 0–10 with evidence and specific teaching attached. The execution pillar is the weighted average of those dimensions.
Fluency measures whether you brought your product knowledge into the room: value propositions, proof points, competitive positioning, and the domain vocabulary your buyer expects. It is computed directly from how many of the expected concepts you actually referenced — verified, as always, against the transcript.
If a pillar doesn't apply to a given call (for example, a conversation where no product concepts were expected), it drops out and the remaining weights spread proportionally. No call is penalized for what it couldn't contain.
Bonuses, not penalties
The score is reward-only. Nothing is ever subtracted from the blend. Two bonuses, worth up to one point combined, sit on top:
- Confident delivery — earned when your vocal delivery (pace, energy, fillers, intonation) shows no confidence concerns and multiple positive signals.
- Clean objection handling — earned when every objection on the call (two or more) was handled with the full acknowledge, anchor, advance pattern.
What the number means
| Score | Band |
|---|---|
| 8–10 | Strong |
| 6–7 | Solid |
| 4–5 | Mixed |
| 0–3 | Needs work |
The right yardstick for the right conversation
A cold call and a discovery call are different skills, so they're scored against different rubrics. Sunnyside also reads the transcript to confirm what kind of conversation actually took place — if a session drifted from its planned type, it's scored against the rubric that fits what really happened.
For sales teams
Each stage of the sales conversation has a purpose-built rubric:
- Cold calls are scored on pattern interrupt, earning permission, delivering concrete value in the first 30 seconds, and booking the next step.
- Discovery calls are scored on the SPIN framework: Situation questions, Problem questions, Implication questions, Need-Payoff questions — plus your listen ratio, because great discovery means the prospect does most of the talking.
- Demos are scored on tailoring to what discovery surfaced, handling objections, checking in, and using customer stories instead of feature lists.
- Negotiation is scored on anchoring, use of silence, multi-option framing, and concession logic.
- Closing is scored on making the direct ask, summarizing value, and securing a concrete next step.
Teams that run MEDDIC can use an enterprise rubric instead, scoring discovery on Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, and Champion.
Teams can also bring their own framework: custom rubrics are scored the same way — each dimension 0–10, with evidence attached, feeding the execution pillar through the same deterministic blend.
Beyond the score
The report goes deeper than the number:
- Delivery and confidence. Acoustic analysis of your actual voice — pace, energy, pitch variation, filler and hedge words, upward intonation — with your lowest-energy moments flagged alongside a stronger alternative line.
- Question quality. Every question you asked, classified as open, closed, leading, or loaded, each with a one-line critique.
- Objection moments. Each objection, how you responded, and which handling pattern you used.
- Skill progression. Every rubric dimension is tracked across your last several calls, so the report can tell you whether a skill is improving, stuck, or slipping — not just how it looked today.
- Targeted drills. Weak dimensions map to specific practice drills, so the report ends with what to do next, not just what went wrong.
The result is a score you can act on: defensible enough to put in front of a team, specific enough to coach from, and consistent enough to track week over week.