About
We built an AI quality control platform for a US outbound sales company with 150+ reps. It transcribes 2,000+ sales calls a day, detects 8 call types and scores each call against the script across 10+ criteria, including lead qualification. Managers get per-rep dashboards and edit rules without developers. The system covers 100% of calls and saves $150K–180K a year versus a QA team.
Challenge, approach, and impact
Evaluation Based on Outcomes
Management judged reps only by the number of meetings booked. There was no way to see at which stage of a conversation a rep went off script, missed key details or qualified a lead incorrectly, and no consistent standard for comparing reps. The company needed objective, comparable evaluation of every call to improve coaching and protect leads and meetings.
How we built
Testimonials
Project Manager @ Stellar
Stellar
“We worked closely with the client’s sales managers to define what a good call looks like for each of the 8 call types, and moved from keyword matching to LLM analysis when the first version couldn’t read context. Seeing every call scored within minutes, after a month of development, was a great result for the whole team.“
Team structure
Client team
Peter
Product owner
The client stakeholders were working closely with the team at Stellar
Agency team
Project Manager
Production
