Call prep and discovery with AI

0:00 Call prep and discovery

Most lost deals are lost in discovery. The rep pitched too early, missed the real problem, spoke to the wrong person, or never learned how the buyer decides. In this lesson, you'll learn how to use AI to walk into every discovery call with a sharp hypothesis, better questions and a plan, and how to practise with an AI role-play partner, while remembering that the conversation itself is still yours.

0:30 The doctor analogy

Here's an analogy. A good doctor doesn't walk in and prescribe. They read your file, form a few hypotheses, ask careful questions, examine, and only then recommend treatment. Discovery is your examination. AI can read the file for you and suggest hypotheses and questions. But you're still the one in the room, listening for what the patient isn't saying.

0:56 Why it matters

Why does this matter? Because buyers now arrive at first calls having already researched you, often with AI. They expect you to understand their business and ask sharper questions than a website can. A rep who shows up unprepared, or pitches in minute three, loses the buyer's attention quickly. Preparation and practice, both faster with AI, are how you earn the right to the next meeting.

1:25 The pre-call plan

Start with a ten-minute pre-call plan built on your account brief. Add the attendees, their roles and what they've said publicly about the problem, professional information only. Add two or three hypotheses, each tied to a verified fact. Add discovery questions tailored to those hypotheses and your qualification framework. List likely objections and a question to explore each. And decide the outcome and next step you want from this call.

1:55 The call-plan prompt

The prompt in the lesson text does this in one go. You paste the brief, the attendees, your offer and your framework, like MEDDICC. It returns hypotheses tied to cited facts, ten open questions grouped by framework element, three likely objections with a question to explore each rather than a rebuttal, and a thirty-minute agenda. It also flags anything that looks unverified. That last line saves you from quoting a wrong fact in the first minute.

2:28 Frameworks

Quick refresher on frameworks. BANT, budget, authority, need and timeline, is simple and suits transactional sales. MEDDICC is deeper and fits complex B2B deals with committees: metrics, economic buyer, decision criteria, decision process, identified pain, champion and competition. SPIN, situation, problem, implication and need-payoff, is a questioning approach that helps buyers put the impact into words. Pick one and use it consistently, because AI can then check your notes against it.

2:59 Simple example

Let's look at a simple example first. You sell scheduling software, and your brief says a clinic group just opened its fifth branch. Hypothesis: booking across branches is getting messy. A good question: walk me through how a patient books today if their usual branch is full. A weak question: you'd agree cross-branch booking is a pain, right? The first invites a story. The second invites a polite yes that tells you nothing.

3:31 Six questions that work

Here are questions that work in almost any B2B discovery. Walk me through how you handle this today. What happens when it goes wrong, and who feels it first? If nothing changes in the next twelve months, what does that cost you? How have you tried to solve this before, and what got in the way? Who else would be involved in a decision like this? And what would success look like six months after a change? Then stop talking and listen.

4:07 AI role-play

Now practise. AI makes a tireless, honest practice partner. Give it a realistic persona, like Fatima, operations director at a seven-branch clinic group in Riyadh, busy, sceptical of vendors, worried about no-shows and staff turnover, and only sharing information when asked good open questions. Tell it about her budget cycle and who else must be involved. Then run the first few minutes of your call. When you type end, ask for feedback on which questions uncovered pain, which were leading, and what you missed.

4:44 Realistic example: UK retail chain

Now a realistic business scenario. A rep at an HR software company in Manchester prepared for a call with a UK retail chain that had announced twenty new stores, a verified fact. Hypotheses: hiring volume would strain manual onboarding, and seasonal turnover was high. On the call, the rep opened broadly, followed the pain around onboarding delays, and discovered the real decision-maker was a newly appointed COO. The agreed next step was a workshop with the COO's team. After the call, AI mapped the notes to MEDDICC and flagged that competition hadn't been explored.

5:25 Common mistakes

Common mistakes. Using the brief to show off your research instead of asking questions; nobody enjoys hearing their own company history recited. Reading AI-generated questions like a script, which kills the flow. Mentioning unverified or personal information, which damages trust instantly. And pitching before you understand the problem. Measure discovery-to-opportunity conversion, qualification completeness in your CRM, how often calls end with an agreed next step, and win rates on deals with complete discovery.

5:57 Using the plan live

Here's a practical tip on using the plan during the call. Don't read it. Glance at your hypotheses before you start, keep your top five questions visible, and let the conversation flow. When the buyer says something important, like we lost two big clients last quarter because of this, follow it with a deeper question rather than moving to your next planned one. The plan is a map, not a script. The best discovery calls often go somewhere the plan didn't predict.

6:33 Prepare in the call's language

One more practical point: prepare for the call in the language it will happen in. If you'll run the call in Arabic or Urdu, ask the AI to produce your questions in that language too, and practise the role-play in it. Phrasing matters: a question that sounds natural in English can feel abrupt in translation. Where you can, have a native-speaking colleague check your key questions. Buyers notice when you've made the effort to speak their business language as well as their spoken one.

7:10 Recap

Let's recap. Discovery decides deals. Build a ten-minute plan with cited hypotheses, open questions, likely objections and a target next step. Use a consistent framework so AI can flag gaps. Rehearse with a realistic role-play, and on the call, listen more than you talk. Try this now: pick your next discovery call, run the call-plan prompt, and rehearse the first five minutes with an AI persona. Next lesson: conversation intelligence.