20 AI Prompts for Discovery Calls

Each prompt generates questions tagged with the signal they surface — and each signal feeds a numbered row of the scorecard in section 04. Ask with a label, score with the same label.

1. No. 06

   The pain excavation set

   Use this when prospects tell you things are "fine" and you know they are not.

   • AE
   • Discovery
   • Question set

   uncovers: pain→ feeds scorecard row 1

   Generate 8 discovery questions that surface PAIN for a [PERSONA] at a company in [INDUSTRY], relevant to the problems [YOUR PRODUCT] solves. Label every question with the layer of pain it uncovers: SURFACE (the annoyance they'll admit immediately), OPERATIONAL (the workflow it breaks), or PERSONAL (what it costs them — hours, credibility, sleep). At least two questions per layer.
   
   Format each as: the question → the label → a one-line follow-up to use if the answer is vague.
   
   Constraints: no question may contain my product's name or category; no "what keeps you up at night"; every question must be answerable by someone who has never heard of me; plain spoken English a rep can say out loud without sounding like a survey. 230 words maximum.

   Copy promptView full prompt details →

2. No. 07

   The cost-of-inaction probes

   Use this when the pain is admitted but nobody has done the math on it.

   • AE
   • Discovery
   • Question set

   uncovers: stakes→ feeds scorecard row 2

   Write 7 discovery questions that quantify STAKES — what the problem costs [PROSPECT COMPANY] if nothing changes. Context: I sell [YOUR PRODUCT]; the pain we've likely surfaced is [SUSPECTED PAIN].
   
   Label each question with the cost type it uncovers: TIME (hours burned), MONEY (spend or lost revenue), RISK (what breaks or gets missed), or MOMENTUM (what they can't start because of it). Cover all four types. For each question, add the arithmetic I should do out loud with their answer — "12 hours a week × 4 reps × …" — so the cost lands in the room, not in my follow-up email.
   
   Constraints: questions must feel like curiosity, not a trap; never ask them to guess a number they'd need finance for — give an easier proxy instead; no more than 20 words per question. Output as a labeled list, 220 words maximum.

   Copy promptView full prompt details →

3. No. 08

   The current-process walkthrough

   Use this when you need the workflow map they would never draw for you on a slide.

   • AE · SE
   • Discovery
   • Question set

   uncovers: process→ feeds scorecard row 3

   the workaround IS the deal

   Build a walkthrough script that maps PROCESS — how [PROSPECT COMPANY] handles [WORKFLOW OR TASK] today, end to end. Give me 7 questions in walking order, starting from the trigger ("what kicks this off?") to the finish line ("how do you know it worked?").
   
   Label each question with what it exposes: TOOLS (what they use), HANDOFFS (where it changes hands), WORKAROUNDS (the duct tape), or OWNERS (who touches it). Include at least one WORKAROUNDS question — that is where deals live.
   
   After the questions, add two listening cues: phrases that signal a workaround they've stopped noticing, like "and then we just export it to a spreadsheet…"
   
   Constraints: conversational phrasing, no audit tone; each question should build on an imagined previous answer so the set flows as a conversation, not a checklist. 220 words maximum.

   Copy promptView full prompt details →

4. No. 09

   The buying-committee mapper

   Use this when "I'd have to check with my team" keeps ending your deals.

   • AE
   • Discovery
   • Question set

   uncovers: people→ feeds scorecard row 4

   Write 7 questions that map PEOPLE — the buying committee at [PROSPECT COMPANY] — without ever using the words "decision maker," "budget holder," or "buying process."
   
   Label each question with what it uncovers: AUTHORITY (who signs), INFLUENCE (whose opinion moves it), VETO (who can kill it quietly), or HISTORY (how the last similar purchase actually went). Cover all four labels. Each question must be askable of a [PERSONA] on a first call without making them feel small if the answer is "not me." For each, add a one-line read on what a hesitant answer usually means.
   
   Constraints: no procurement jargon; phrase questions around how work gets done and how the last tool got bought, not around org charts. Output as a labeled list, 220 words maximum.

   Copy promptView full prompt details →

5. No. 10

   The money conversation openers

   Use this when the money question sits in your throat for the whole second half of the call.

   • AE
   • Discovery
   • Question set

   uncovers: money→ feeds scorecard row 5

   Give me 6 ways to open the MONEY conversation on a discovery call with a [PERSONA] at [PROSPECT COMPANY] — none of which is "what's your budget?"
   
   Label each with what it uncovers: SPEND (what they pay today for the problem — tools, headcount, agencies), APPETITE (what a fix would be worth to them), or PATH (how money for something like this actually gets approved). At least two per label. For each: the exact phrasing, when in the call to use it, and the graceful exit line if they stiffen.
   
   Constraints: phrasing must let them answer in ranges or comparisons, never demand a number; no apologizing before the question — "I hate to ask, but…" is banned; each question 15–25 words. 210 words maximum.

   Copy promptView full prompt details →

6. No. 11

   The timeline reality check

   Use this when every deal is "a priority this quarter" and half of them never move.

   • AE
   • Discovery
   • Question set

   uncovers: timing→ feeds scorecard row 6

   Write 6 questions that test TIMING — whether [PROSPECT COMPANY] will actually move on this problem or is just browsing.
   
   Label each with what it uncovers: FORCING EVENT (a date that makes waiting expensive — a renewal, a launch, an audit, a hiring plan), SEQUENCE (what has to happen before they could buy anything), or DRIFT (evidence this has been "a priority" for three quarters without movement). For each question, add what a real answer sounds like versus a polite answer, one line each — I need to hear the difference live.
   
   End with one summary line: which single label predicts close dates best for a product like [YOUR PRODUCT], and why.
   
   Constraints: no pressure phrasing and no invented time pressure; curiosity tone throughout. 220 words maximum.

   Copy promptView full prompt details →

7. No. 12

   The status-quo autopsy

   Use this when your real competitor is a spreadsheet and nobody will say so.

   • AE
   • Discovery
   • Question set

   uncovers: rivals→ feeds scorecard row 7

   Generate 7 questions that expose RIVALS — everything competing with me for this deal at [PROSPECT COMPANY]: named competitors, the in-house spreadsheet, the intern who "handles it," and doing nothing at all.
   
   Label each question with the rival it uncovers: VENDOR (a competitor is in the building), HOMEGROWN (they built or cobbled something), or INERTIA (nobody owns the problem, so nothing moves). Cover all three. For each, add the follow-up I ask when the answer reveals that rival, and one line on how it changes my next call.
   
   Constraints: never ask "are you looking at competitors?" — every question comes at it sideways, through their workflow and their history; no disparaging any rival, including the spreadsheet. Output as a labeled list, 230 words maximum.

   Copy promptView full prompt details →

Four prompts that eat the transcript. The seven signals you asked about in section 02 come back here as seven scorecard rows — the red string below shows the wiring.

1. No. 17

   The seven-signal scorecard

   Use this when the call felt great and you want to know if it actually was.

   • AE · Manager
   • Post-call
   • Long context

   pulls → all seven signals from [CALL TRANSCRIPT]

   Score this discovery call against seven signals: PAIN, STAKES, PROCESS, PEOPLE, MONEY, TIMING, RIVALS. Transcript: [CALL TRANSCRIPT].
   
   For each signal, output one row: a score of 0 (never surfaced), 1 (mentioned, not explored), or 2 (explored down to a specific fact) — then the single strongest supporting quote from the transcript, verbatim and attributed, or the word SILENT if nothing qualifies.
   
   After the seven rows: TOTAL out of 14; the one signal whose absence most threatens this deal; and the single question that would have raised the score most, written so I can ask it in the follow-up.
   
   Constraints: quotes must be verbatim from the transcript — paraphrase never counts as evidence; do not award a 2 because the rep asked, only because the prospect answered with a specific; keep the whole output under 250 words.

   Copy promptView full prompt details →

2. No. 18

   The gap auditor

   Use this when you replay the call in the shower and keep finding questions you didn't ask.

   • AE
   • Post-call
   • Long context

   pulls → the dropped threads behind rows 1–7

   Audit this discovery call for what I failed to ask. Transcript: [CALL TRANSCRIPT]. I sell [YOUR PRODUCT].
   
   Step 1: list the threads the prospect opened that I never pulled — a verbatim quote fragment for each, with a one-line read on what was probably behind it.
   Step 2: mark which of the seven signals (PAIN, STAKES, PROCESS, PEOPLE, MONEY, TIMING, RIVALS) each dropped thread would have fed.
   Step 3: for the three most expensive gaps, write the recovery question I can ask in a follow-up email without it feeling like an interrogation — framed as a thought I had after the call.
   
   Constraints: only flag threads that actually appear in the transcript; if I genuinely covered everything, say so instead of inventing gaps; recovery questions 25 words maximum each; 230 words maximum overall.

   Copy promptView full prompt details →

3. No. 19

   The evidence-only follow-up email

   Use this when the follow-up email deserves better than "Great chatting today."

   • AE
   • Post-call

   pulls → transcript facts only, nothing invented

   Write my follow-up email using only what is in this transcript: [CALL TRANSCRIPT]. Prospect: a [PERSONA] at [PROSPECT COMPANY].
   
   Structure: a subject line referencing something they said, six words maximum, no "great chatting"; one opening line that proves I listened — their words, not mine; a three-bullet recap where every bullet pairs something they told me with what we will do about it; the agreed next step with its date; and one open question from the call, planted to keep the thread alive.
   
   Constraints: every factual claim must trace to the transcript — if I promise a case study, the transcript must show we discussed one; 120 words maximum for the whole email; no exclamation marks; no "just following up"; write like a sharp colleague, not a vendor. If the transcript shows we never agreed on a next step, tell me that first — then write the email that fixes it.

   Copy promptView full prompt details →

4. No. 20

   The coach's margin notes

   Use this when you want the manager's margin notes without waiting for call review.

   • Manager · AE
   • Post-call
   • Long context

   pulls → every line of the transcript, including yours

   Be my discovery-call coach. Mark up this transcript the way a manager who listens to hundreds of calls would: [CALL TRANSCRIPT].
   
   Output:
   1) TALK RATIO — estimate my share of airtime and flag any monologue over 60 seconds with its position in the call.
   2) QUESTION AUDIT — count my questions; mark each open or closed; flag every question I answered myself before they could.
   3) THE THREE MOMENTS — the best question I asked and why it worked; the worst moment and the line that would have saved it; the dropped thread that cost the most.
   4) ONE DRILL — a single habit to practice on my next call, phrased as an instruction I can tape to my monitor, ten words maximum.
   
   Constraints: quote the transcript verbatim in every observation; no generic coaching — "build rapport" is banned; every note must point at a specific line. 240 words maximum.

   Copy promptView full prompt details →