Analyze talk ratio and question quality from a call transcript
Measures what conversation intelligence tools measure — talk ratio, question count — but adds the layer they skip: classifying every question by type and quality, and pinpointing the exact moments a rep should have dug deeper instead of moving on. Works from any transcript, no Gong seat required.
You are a conversation analyst who has studied thousands of B2B sales calls. You know the research: top reps listen more than they talk on discovery calls, but talk ratio alone is a vanity metric — a rep can hit 40% talk time while asking nothing but checkbox questions. Quality of questions beats quantity of silence. I'm going to paste a call transcript with speaker labels. Produce: 1. TALK RATIO — estimate each speaker's share of words spoken, as percentages. Note the longest rep monologue (quote its first line and estimate its length). 2. QUESTION INVENTORY — list every question the rep asked, verbatim, classified as: discovery (opens new ground), clarifying (deepens an answer), checkbox (fills a form), or leading (fishing for a yes). 3. QUESTION QUALITY SCORE — 1–10, with a two-sentence justification. Penalize stacked questions (asking two at once), answered-own-question moments, and long setups before questions. 4. MISSED FOLLOW-UPS — the three biggest moments the prospect said something worth digging into and the rep moved on. Quote the prospect's line each time. 5. ONE CHANGE — the single highest-leverage adjustment for the next call, stated in one blunt sentence. Do not pad this with generic advice about active listening. Every claim must point to a specific line in the transcript. Before you analyze anything, interview me. Ask me these questions ONE AT A TIME, waiting for my answer each time: 1. Paste the full transcript, with speaker labels. 2. What kind of call was this (cold call, discovery, demo, follow-up), and how long was it? 3. What did the rep want to walk away with from this call? Once you have my answers, run the analysis. If the transcript has no speaker labels, ask me to identify the rep's first line so you can infer the rest.
How to use it
- 1
Copy the prompt into Claude, ChatGPT, or Gemini.
- 2
Paste a transcript with speaker labels — talk-ratio estimates need to know who's who.
- 3
Tell it the call type; a 70% talk ratio is a problem on discovery and normal on a demo.
- 4
Spot-check the question inventory against the transcript, then focus coaching on the missed follow-ups section.
- 5
Re-run monthly on the same rep to track whether question quality actually improves.
Best practices
Word-share is an estimate, not a stopwatch — treat the number as directional and the question analysis as the real product.
The 'missed follow-ups' section is the best coaching material in the output; role-play those three moments in your next 1:1.
Compare a rep's best call and worst call in the same chat and ask what differed — the contrast teaches more than either alone.
For demos, ask the model to also flag stretches over 3 minutes with zero prospect words.
Example: what this looks like in practice
A founder doing founder-led sales at a 6-person devtools startup suspects he talks too much. He pastes a 40-minute discovery transcript from Fireflies. The result: 71% talk share, a 4-minute monologue about the product roadmap, and 14 questions of which 9 were checkbox and 2 were leading. The missed follow-ups sting — the prospect mentioned 'we tried building this internally last quarter' and he moved straight to a feature tour. The one change: 'When they mention a past attempt, your next three questions live there.' His next call runs 55/45 and surfaces a budget owner he'd have missed.
Best fit
This prompt is one gear in a bigger machine. We orchestrate 20+ tools into outbound systems our clients own — and guarantee the results.
Apply for a Pilot Spot → →Frequently asked questions
Common benchmarks put top discovery reps around 40–45% talk time, but it varies by call type — demos legitimately run higher. The more useful signal is monologue length and question quality: a rep at 50% asking sharp follow-ups beats a rep at 40% running through a checkbox script. This prompt measures both.
More call transcript analysis & coaching prompts
Extract every objection from a call transcript and grade how it was handled
Pulls every objection out of a transcript — including the soft, disguised ones reps miss — with verbatim quotes, a diagnosis of the real concern, a letter grade on the rep's response, and a better line for next time. It's a full objection review in two minutes instead of re-listening to the call.
Score a sales call against MEDDIC or SPICED with a coaching scorecard
Turns any transcript into a framework scorecard with a 0–3 score per element, verbatim evidence behind every score, and the exact question that would close each gap. Stage-aware, so a first discovery call isn't punished for missing decision criteria. It's the deal-review prep that used to take a manager 45 minutes per call.
Annotate a great call to onboard new reps
Converts your best call recording into annotated game film for onboarding: 10–15 key moments with the move named, why it worked, and what a rookie would have done instead — plus the invisible restraint moves, five lines worth stealing verbatim, and one honest imperfection so new reps learn judgment, not imitation.