Verified against ChatGPT · 2026-08-10
Build a discovery call question bank that digs past the first answer instead of checking boxes
Produces a discovery call question set organized by what you actually need to learn, with a built-in follow-up probe for each question so the call surfaces the real underlying problem instead of stopping at a surface-level answer.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Build me a discovery call question bank for an upcoming call. I don't want a generic list of "good discovery questions" — I want questions built around what I specifically need to learn from THIS prospect to know if we're a fit and to build a real business case later. WHAT I'M SELLING A contract review tool that flags risky clauses before legal sign-off, aimed at mid-market procurement teams. WHO I'M TALKING TO Head of Procurement at a 900-person manufacturing company, first call, no prior contact with legal team. WHAT I NEED TO LEARN TO QUALIFY THEM How many contracts they process monthly, whether legal or procurement owns the review step, and their current turnaround time. CALL LENGTH 30 minutes, first call. KNOWN RISK OR OBJECTION GOING IN Suspect they already have an in-house paralegal doing this manually and may see us as redundant. For every question, write the obvious first-layer answer a prospect would likely give, then a specific follow-up question designed to get past that first answer to the real number, constraint, or motivation underneath it — a discovery call that stops at the first answer usually just confirms what the prospect already believes, not what's actually true. Organize the questions into three sections matched to what they should surface: current state and pain, decision process and stakeholders, and success criteria and timeline. Given the known risk or objection, write one question early in the call specifically designed to surface whether that risk is actually present, phrased as an open question, not a leading one that telegraphs the answer I want. Keep the total question count appropriate to the call length — do not hand me forty questions for a 20-minute call. Do not write questions that could be answered from public information already available; every question should require the prospect to tell me something I couldn't have found myself. Format as a numbered list within each section: Question, then "Likely surface answer," then "Follow-up to go deeper."
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The mechanism doing the real work here is the required "likely surface answer, then follow-up" pairing, because it forces the model to reason about the conversation as two layers instead of generating a flat list of nice-sounding open questions — a plain "give me discovery questions" prompt reliably produces questions like "what challenges are you facing?" that a prospect answers with whatever generic pain they'd tell anyone, and the conversation never gets past that first, socially-safe layer. Writing out the anticipated surface answer explicitly is what makes the follow-up land correctly, because a follow-up written without first predicting the likely first response tends to be generic ("can you tell me more?") rather than targeted at the specific gap between what a prospect usually says and what's actually driving the decision. Requiring the known-risk question to be phrased openly rather than leading it matters because a leading question ("you don't already have someone doing this in-house, right?") invites a defensive or performative answer, while an open version phrased around current process gets an honest description the rep can actually diagnose. Capping question count to the call length is a direct check against GPT-5.1's tendency, when asked for a "comprehensive" question bank, to produce an exhaustive list that would take three call lengths to actually ask — a list this long in practice gets skimmed rather than used, and the discipline of matching volume to available minutes is what makes the output something a rep could genuinely run from live on a call.
What you get back
Current state and pain — Q: "Walk me through what happens to a contract from the moment it lands on your desk to signature." Likely surface answer: "Legal reviews it, then it comes back to us for signature." Follow-up: "How long does that legal review step typically take, and has a deal ever slipped because of it?"
Verified against
ChatGPT GPT-5.1 · 2026-08-10
Changelog
- 2026-08-10 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
