Compare Rep Talk-Time Across Call Transcripts

You are a sales analyst trained to compute rep vs. prospect talk dynamics from agency call transcripts. You estimate from text only — never invent durations.
Agency: [AGENCY_NAME] — [SERVICES]
Reps in this batch: [REP_NAMES]
Transcript batch (label each as ===CALL N | rep: NAME | outcome: WON/LOST/OPEN===):
[TRANSCRIPT_BATCH]
For each call, estimate (by word count or speaking turn proportion): (1) rep talk share %, (2) prospect talk share %, (3) longest rep monologue (in words + topic), (4) longest prospect monologue, (5) # of open-ended questions asked by rep. Then compare across reps and tie to outcomes.

- Estimate from text only — never invent timestamps or seconds
- Quote exact lines for the longest monologues
- Flag any rep with 
Per-call row: Call # | Rep | Outcome | Rep talk % | Longest rep monologue (words + topic) | Open-ended Qs | Notes
Then rep-level rollup: avg talk %, avg open-ended Q count, win rate.
Then 2-3 observations + one coaching recommendation per rep (only reps with 3+ calls).

• [AGENCY_NAME] — Your agency name
• [SERVICES] — What you sell
• [REP_NAMES] — Reps included in this batch
• [TRANSCRIPT_BATCH] — Labeled transcripts: ===CALL N | rep: NAME | outcome: WON/LOST/OPEN===

Agency: NorthLoop — SEO + paid retainers
Reps: Maya, Devon, Priya
Batch: 15 calls (Maya 6, Devon 5, Priya 4)
Sample: Call 3 | rep: Maya | outcome: WON — Maya 42% talk, longest monologue 240 words on local SEO methodology

| Call | Rep | Outcome | Rep Talk % | Longest Monologue | Open-ended Qs | Notes |
|---|---|---|---|---|---|---|
| 1 | Maya | WON | 42% | 240w / local-SEO methodology | 11 | balanced |
| 4 | Devon | LOST | 71% | 580w / pitch deck walkthrough | 3 | monologue heavy |
| 7 | Devon | LOST | 68% | 410w / process overview | 4 | same pattern |
| 12 | Priya | OPEN | 38% | 180w / case study | 14 | high-question style |

**Rollup**
- Maya — avg 44% talk, avg 10 open-ended Qs, 4W/1L/1O
- Devon — avg 67% talk, avg 4 open-ended Qs, 0W/4L/1O
- Priya — avg 39% talk, avg 13 open-ended Qs, insufficient sample (4 calls, all OPEN)

Observations: Devon's 67% talk share correlates with 0 wins in 4 closed deals; longest monologues are always the pitch deck walkthrough. Maya's open-ended Q rate is 2.5x Devon's.

Coaching: **Devon** — cut deck walkthrough to half its current length and insert 3 mandatory check-in questions during the process section. **Maya** — keep the current pattern; consider mentoring Devon on discovery-style questioning.

• Always label outcomes — without them, talk-time data is trivia
• Don't fire a rep on this data alone — it's a signal, not a verdict. Pair with discovery-quality audits
• Use the longest-monologue topic to spot which sections of your deck are over-explained