Top reps had 35% close rates. New reps had 9%. The playbook lived in their heads.
The team had a 90-page sales playbook in Notion that nobody read past their first week. Reps figured out objection handling and closing through trial and error — fine for the top 3 reps with 18+ months of tenure, painful for the next 22.
The clearest signal was demo conversion: top reps converted at 35%, the median rep at 17%, new reps at 9%. Every deal lost was money on the floor, but more importantly it took ~5 months for new hires to even hit median.
We listened to 40 demo recordings before we touched any code.
Across 40 calls we tagged every objection raised by prospects, every reply given, and the outcome of the deal. The pattern was strikingly clean: 12 objections accounted for 86% of all friction. Top reps had 2–3 strong responses memorized for each. New reps stalled or reached for the wrong angle.
We also noticed reps were rebuilding context every call from scratch — past notes were buried in Salesforce. The CRM was a write-only system in practice.
"It's embarrassing how much of our 'process' was just whatever the rep remembered from last Tuesday's training."
A panel that listens to the call and shows the next move — built in 10 days.
We tapped Zoom's live-transcript stream and ran chunks through a small classifier we trained on the tagged objection set. When an objection fired, the rep saw — in a second-screen panel — the 2 strongest reply scripts from the playbook, plus a one-line CRM context ("this prospect mentioned budget freeze in your last call"). No interruption, no autopilot — the rep stays in charge.
After the call, the same pipeline produced a structured CRM update (next steps, key concerns, sentiment shift) and pushed it to Salesforce. A Slack message went to the rep's manager flagging any deal where strong objections were left unanswered.
4 weeks. Two cohorts. The numbers were obvious by week 2.
We ran the pilot on the 7 newest reps (cohort A) and 8 mid-tenure reps (cohort B). Cohort A's demo conversion went from 9% → 21% in the first 2 weeks and stabilized at 24%. Cohort B saw a more modest +9 points but had 100% adoption — they actively asked for more script variants.
The CRM-update side effect surprised everyone: complete, structured updates went from 41% of calls to 96% of calls. Pipeline review meetings shrank from 90 minutes to 35.
Numbers that survived the pilot.
- Demo → close conversion9% → 21% (new reps)+12pt
- Time to median performance5 months → 7 weeks−65%
- CRM update completeness41% → 96%+55pt
- Pipeline review time90 min → 35 min−61%
- Calls covered by coach—100%