An AI Company's Sales-Call Audit

I worked with an AI training and implementation company in 2025. It had a large online audience, a small consulting team and a calendar of booked sales calls. Its booking form already asked prospects to confirm a minimum spend, a line the team had added because too many small buyers were booking.

Buyers couldn't tell what the company sold, which made it hard for them to buy. AI meant something different to every prospect, from people who'd never opened ChatGPT to engineers building their own systems. The team could keep booking calls, but each new client meant delivery work the company had no offer or process to support yet.

I started with the calls. The company's recording tool only exported transcripts on a premium plan, so I wrote a scraper to pull all 29. I scored each call against a rubric of 15 to 25 angles and kept the exact words buyers used. That came to about 1,400 pages of notes and a blueprint of more than 100 pages. Most buyers had the same problems, and three kinds of build covered about 90% of what they asked for: content, research and a tool people could talk to. So I sorted every possible build into five tiers, from an assistant loaded with the company's own material, a couple of hours' work, up to systems that learn from their own mistakes. A buyer and the team could now agree on the size of a problem before anyone quoted it. My plan priced implementation at $10,000 and up, sold in three-, six- or nine-month terms with three milestones each. It moved qualification before the call: a questionnaire that returns a rough estimate, books the buyers who fit and sends the rest to the company's content and community. And I advised against selling to the calls already booked until there was an offer to deliver.

• Custom transcript scraper
• Rubric-based call scoring
• Buyer-language notes across 29 calls
• Five-tier build ladder
• Pre-call qualifying questionnaire

The founder started planning the build around the five tiers, beginning with the first. The sales calendar stayed closed while the team finished a new product, and the founder asked me for the qualifying questions and the copy for the new consulting page. I shared the raw notes with the whole team, so anyone could reread a call beside its scoring.

The team got one scale for sizing any request, from a two-hour assistant to a learning system, and a plan for qualifying buyers before anyone spends an hour on a call.

A company's recorded sales calls already say what its buyers will pay for. Here they said buyers need to know where they're starting before they'll choose a build, so the tiers and the questionnaire came before any new pitch.