The AI wrote 15 sales questions. None asked about the $76,800 lease.

"Is $3,150 a month firm?"

The property manager asked that first. Our cleaning bid was $250 a month above her vendor. A generic discovery guide gave us 15 polite questions. None showed that the price gap was the smallest number in the room.

AI Prompts, Reviewed takes one prompt or workflow from How I AI, Claire Vo's show where guests share their screens and walk through how they really use AI. We run it on a small-business job, score it, and tell you what we'd change.

Credit: Based on "How to Turn a Sales Playbook into an Interactive Script Generator with ChatGPT," a workflow from How I AI (hosted by Claire Vo, written up by the ChatPRD team), with Hiten Shah (co-founder of Crazy Egg and KISSmetrics). Full episode.

What the original does

Hiten Shah loads Winning by Design's sales-framework documents into a ChatGPT Project and asks for a SPICED discovery guide: Situation, Pain, Impact, Critical Event, Decision. Three open questions per section, probes, no pitching. Step two adds a product and customer brief and regenerates the same guide. In one line: "Teach the framework first, then add product and customer context to tailor the same deliverable."

The test

Composite (made-up) bid for a four-person commercial cleaning crew. Test clock: Monday, Oct 5, 2026. The prospect is a property manager with four small office buildings.

• Current vendor: $2,900/mo. Our bid: $3,150/mo. Gap: $3,150 − $2,900 = $250/mo, or $250 × 12 = $3,000/yr.
• Pain: 7 tenant complaints in September and 2 missed nights.
• Contract: the current deal ends Nov 30, 2026 and needs 30 days' written notice, so notice is due Oct 31. Days left: Oct 31 − Oct 5 = 26.

Step one, run in the spirit of the original (no framework PDFs, since most owners don't have them), gave exactly what it asks for: 5 sections × 3 questions = 15 questions. They read like a survey. "What challenges are you facing with your current provider?" "How is this affecting your business?" The Decision section asked who else is involved. It never asked how much she can sign for on her own.

Step two, with our crew brief, got sharper. It still produced no number to write down. So we forced one number per section. Three answers changed the deal:

1. Impact: "Has a tenant brought up cleaning near a renewal?" Yes. Her biggest tenant pays $6,400/mo, or $6,400 × 12 = $76,800/yr, and renews Nov 12. Restrooms were "the reason we're touring."
2. Critical Event: the Oct 31 notice date, 26 days out. Miss it and the old vendor rolls into December.
3. Decision: she can sign up to $3,000/mo alone. Our bid is $3,150 − $3,000 = $150 over, so the building owner signs.

A $250 price gap is small next to a $76,800 lease.

The lease is $76,800 ÷ $3,000 = 25.6 times the yearly gap. We stopped defending the $250 and wrote the owner a one-page note: the lease, Nov 12, Oct 31.

Verdict: 7/10

Strengths: framework first, context second. Discovery stays separate from the pitch. The runbook fixes are honest and useful.

Weakness: it leans on framework documents most owners don't have, and 15 questions is a lot for a 20-minute walkthrough. Nothing forces a dollar figure, a date, or a signing limit out of the call.

Founder tweak: paste your own deal facts. Cap it at 10 questions. Make every section end with one number to write down: the cost of the pain, the deadline, the signer's dollar limit. Then compare your price gap to the impact before you discount.

Copy the MCAFax version

This is our own rewrite, not the original prompt. Copy the original from Hiten Shah on How I AI.

You are my sales prep partner for a small service business. I will paste: what I sell, my price, who I'm meeting, their current vendor and price if known, and anything they've told me. Write a first-call discovery guide using SPICED (Situation, Pain, Impact, Critical Event, Decision).

Rules:

1. Max 10 questions total, 2 per section. Open questions only. No pitching, no features, no price defense.

2. Each section ends with "Write down:" and ONE thing to capture:
   - Situation: what they use now, what it costs per month.
   - Pain: how many problems in the last 30 days (count it).
   - Impact: the dollar cost of the pain. Ask about customers, tenants, or revenue that is at risk, and what that is worth per year.
   - Critical Event: the date that forces a choice (contract end, notice deadline, renewal, inspection). Ask about notice periods.
   - Decision: who signs, and the dollar limit the person I'm meeting can approve alone.

3. Add one follow-up probe per question for a vague answer like "it's fine" or "it's annoying."

4. After the guide, give me a blank price-check line: (my price − their current price) × 12 = yearly gap. Compare the yearly gap to the Impact number. If Impact is bigger, tell me not to discount and to write a one-page note for the signer.

5. Keep it short enough to read on my phone during the call.

Paste into ChatGPT, Claude, Gemini, or Copilot. Leave out customer names you don't need.

Before your next bid call

1. Paste your price, their price, and what you know into the MCAFax version.
2. Fill in every "Write down" line during the call.
3. Do the price-check math with a calculator, not the AI.
4. If the signer isn't on the call, send a one-page note with the impact and the deadline.
5. Keep the deal moving: lead follow-up automation (more Sales prompts).

SBA on marketing and sales for a small business.