How does Roonly score a conversation?
Every scored conversation gets an overall score out of 100 plus a score for each stage, produced by AI reading the transcript against your team's rubric and playbook.
Roonly does not grade you on a checklist of keywords. An AI model reads the whole transcript, with your lines and the prospect's lines labeled, and scores it the way an experienced coach would: against a rubric for your kind of selling and against the playbook your company uploaded.
What goes into the score
- The transcript, with speaker labels. Roonly first works out which speaker is the rep, so your lines are judged as yours.
- The rubric for your team type. Door-to-door reps, in-home inspection techs, in-home service techs and office call reps each have their own rubric with different stages.
- Your company's playbook and profile, if a manager has uploaded one. The scorer is told to weigh how closely you followed it.
- Prospect sentiment across the conversation, which helps the model see whether the mood improved or soured.
The score card
The door-to-door rubric produces an Overall Score out of 100 and a score out of 100 for each stage:
- Opener. First impression, confidence, rapport and the hook.
- Discovery. Whether you asked questions and listened.
- Value. How well you tied the offer to what the prospect cares about.
- Objections. What you did when the prospect pushed back.
- Close. Whether you asked for a decision and how you handled the exit.
Tech rubrics use different stages that match a home visit, such as Arrival and Rapport, Homeowner Education, Findings and Value, Pricing Confidence and Honesty and Compliance. On the mobile app these are shown mapped onto the five familiar bars; the web dashboard shows the full list.
Scores are colored the same everywhere: green at 70 and above, orange from 50 to 69, red below 50.
Short conversations are scored fairly
A door slam is not a failure of your close, so the rubric scores by conversation length. For a quick rejection under 2 minutes, only the parts you controlled are scored, and a graceful exit earns a reasonable close score rather than a zero. Stages that never had a chance to happen are not penalized.
Averages follow the same idea. Any conversation of 30 seconds or less counts for much less in your average score on the Team page, your profile, the leaderboard and contests, so a day of quick rejections does not sink your number.
What is not scored
Some recordings never get a score:
- No speech. The transcript is almost empty.
- Self talk. One voice only, either under a minute or longer with very few words. This is usually you narrating between doors.
- Split. A recording over 2 minutes that contained two or more separate conversations. The parent is set aside and each child conversation is scored on its own.
- Unqualified visits on tech teams. When the AI determines the person was a renter or not the decision maker, the outcome is marked unqualified and the conversation is left out of averages.
None of these appear in your History.
Beyond the number
Alongside the scores, the AI writes a short summary, a list of what went well, what to improve, two or three coaching tips, the objections it heard, and a set of focus areas. It also extracts each objection as its own record with how you responded and how well that response worked. Those feed your manager's Insights page and your own coaching plan.
Outcomes are inferred too. If you did not set one, Roonly reads the transcript and picks Sale, Lead, Callback or No Interest, and it pulls the prospect's name into the contact field. You can correct either on the conversation screen.
Related
Still stuck? Contact support.