What it does
This skill prepares a rep for an upcoming discovery or expansion call. Given the account, the attendee titles, and a qualification framework (MEDDICC, BANT, SPICED, or your own), it produces a focused prep doc: a refreshed view of the account, a hypothesis about why they took the meeting, the single goal of the call, and a set of tailored discovery questions that each ladder up to a specific field of the framework instead of being a generic list copied from a template. The point is to walk in sounding like you did the homework and to make the call efficient enough that the rep can actually listen instead of reading from a script.
What separates a good prep doc from a bad one is that the questions are tailored and tagged. A bad prep doc is a generic question bank, "what are your goals, what's your timeline, who's involved", that any rep could ask any account, which teaches the buyer nothing and signals no preparation. A good one sequences questions broad-to-specific, ties each to the exact framework field it fills ([Metrics], [Decision Process], [Champion], [Implicate Pain]), and references something concrete about THIS account in the wording. The framework tags do double duty: they tell the rep live which fields they still need to fill, and they make the post-call CRM update fast because every answer maps to a field.
It has two modes, and choosing the right one is most of the value. For a net-new prospect it runs Discovery mode: cold qualification, why-now, economic buyer. For an existing customer it switches to Expansion mode and reorients everything around adoption, new use cases, stakeholders beyond the original champion, and renewal risk, because qualifying an existing customer like a cold prospect is a fast way to annoy them. The same Confirmed-vs-Inferred discipline from good account research applies: web-grounded account facts where available, clearly-labeled inferences otherwise, never fabricated specifics asserted as truth.
Where it beats manual prep: it makes the framework actually get used. Most reps know MEDDICC and fill it out after the fact from memory; this front-loads it into the questions so the call generates the framework data instead of the rep reverse-engineering it later. Where it does not help: it doesn't know what was said on prior calls unless you paste the notes, and it won't invent CRM history. Feed it your prior-touch notes and the hypothesis sharpens; give it nothing and it's working from public signal only and will say so.
Inputs & outputs
Inputs
- Account name / URL and the meeting attendees' titles, required
- Your qualification framework: MEDDICC, BANT, SPICED, or a custom one (defaults to MEDDICC)
- What you sell and the outcome you drive
- Optional: notes from prior touches or emails
- Optional (for existing customers): current usage, product adoption, health score, renewal date
Outputs
- A prep doc with: Account refresh, Why-now hypothesis, Goal of the call, tailored discovery questions tagged to framework fields, Likely objections with responses, and Next-step options
- An explicit Confirmed vs. Inferred split on account claims
- An Expansion-mode variant when the account is flagged as an existing customer
- Questions sequenced broad-to-specific, each one referencing something specific about the account or an attendee's role
How to set it up
Set it up with your framework baked in
Create a Claude Project named "Call Prep" and paste the SKILL.md body below into its custom instructions, then add one line stating your default framework and what you sell: "We run MEDDICC. We sell
Pick the framework explicitly the first time
Tell the skill which qualification framework you run. It will tag every question to a field of that framework, which is the whole point, it means your handwritten call notes map cleanly back into the CRM's MEDDICC (or BANT/SPICED) fields afterward instead of needing translation. If you run a custom framework, paste its fields once and it'll tag against those.
Run it before the call with whatever context you have
Send: "Prep me for a discovery call with
Walk in with three questions memorized, then adapt
Don't read the doc as a script. Memorize the three highest-priority questions and the why-now hypothesis, then let the framework tags tell you live which fields you still haven't filled so you can steer the conversation toward the gaps. After the call, the tagged structure makes the CRM update a two-minute copy job instead of a memory exercise.
The SKILL.md
Save this as SKILL.md in a folder named discovery-call-prep, or paste the body into a Claude Project. Tell it the account, attendees, and your framework.
---
name: discovery-call-prep
description: Use when a rep needs to prep for a discovery or expansion call. Produces an account refresh, a why-now hypothesis, the goal of the call, and tailored discovery questions each mapped to a qualification framework field (MEDDICC/BANT/SPICED/custom). Has an Expansion mode for existing customers. Tags account claims Confirmed vs Inferred and never fabricates account history.
---
# Discovery Call Prep
You prepare a rep for an upcoming call. Output a focused prep doc the rep can scan in two minutes and reference live. The questions are the heart of it: each must be tailored to THIS account and tagged to the framework field it fills.
## Inputs
- Account name/URL and attendee titles (required).
- Qualification framework: MEDDICC, BANT, SPICED, or custom (default MEDDICC if unspecified).
- What the user sells and the outcome it drives.
- Optional: prior-touch notes; for existing customers, usage/adoption/health/renewal context.
## Mode selection (decide first)
- If the account is a NET-NEW prospect, run Discovery mode: cold qualification, why-now, economic buyer, decision process.
- If it is an EXISTING customer, run Expansion mode: orient questions around current adoption, new use cases, stakeholders beyond the original champion, expansion triggers, and renewal risk. Do NOT cold-qualify an existing customer.
## Method
1. Refresh the account: if web access exists, read the homepage and one relevant page (product, pricing, or news); otherwise reason only from the context given. Tag each fact Confirmed (read) or Inferred (reasoned). Never fabricate specific facts or prior-call history.
2. Form a why-now hypothesis: why did they take THIS meeting / what changed? One or two sentences, marked Inferred unless grounded in a fact.
3. State the single goal of the call in one sentence, the one field or commitment most needed to advance the deal.
4. Write 8-12 discovery questions, sequenced broad -> specific. Tag each with the framework field it fills, e.g. [Metrics], [Economic Buyer], [Decision Process], [Champion], [Implicate Pain], [Competition], [Paper Process] for MEDDICC. Each question must reference something specific about this account or an attendee's role, not a generic template question.
5. List 2-4 likely objections, each with a one-line response.
6. Offer 2 concrete next-step options the rep can land at the end of the call.
## Hard rules
- No invented facts about the account, its tooling, its prior conversations, or its internal structure. Mark every inference.
- Generic questions are banned. If a question could be asked of any account in the segment unchanged, rewrite it to reference this account.
- Keep to one page.
## Output template
**Call Prep: <Account>, <today's date>, [Discovery | Expansion]**
Attendees: <titles>
Sources: <pages read, if any, or "none">
**Account refresh** (3-4 bullets, each tagged Confirmed/Inferred)
- <...>
**Why-now hypothesis** (Inferred)
<1-2 sentences.>
**Goal of the call**
<one sentence.>
**Discovery questions** (broad -> specific, each tagged to a framework field)
1. <question> [Field]
2. <question> [Field]
...
**Likely objections**
- "<objection>" -> <one-line response>
**Next-step options**
- <option 1>
- <option 2>
Run it from the terminal
Example run
An AE preps for a MEDDICC discovery call with two stakeholders at a net-new account.
Prompt: discovery-call-prep. Account: acme-observe.com. Attendees: VP Sales, Head of RevOps. Framework: MEDDICC. We sell an AI SDR that books meetings while reps ramp.
Output:
Call Prep: Acme Observe, 2026-06-10, Discovery
Attendees: VP Sales, Head of RevOps
Sources: homepage, /careers
Account refresh
- API observability SaaS, sales-led (Confirmed, homepage)
- 5 BDR roles open this quarter (Confirmed, /careers)
- Likely scaling outbound ahead of pipeline (Inferred)
- New AI tracing product suggests a new-segment push (Inferred)
Why-now hypothesis (Inferred)
They're hiring five BDRs at once, which implies a coverage gap forming during ramp, they took the meeting because that gap is top of mind.
Goal of the call
Confirm the metric the VP is judged on (pipeline coverage ratio) and identify who controls the budget for an SDR-workflow tool.
Discovery questions
1. How are you thinking about pipeline coverage as the five new BDRs ramp? [Metrics]
2. What does the number look like if ramp slips a quarter? [Implicate Pain]
3. Walk me through how a tool that touches the SDR workflow gets approved internally. [Decision Process]
4. Who owns the coverage number besides your team, RevOps, finance? [Economic Buyer]
5. Who else inside would feel it first if coverage dipped? [Champion]
6. Are you evaluating any other ways to close the ramp gap right now? [Competition]
Likely objections
- "We just hired for this." -> Reframe: the AI SDR feeds those new hires pipeline so they ramp faster; it doesn't replace them.
- "We have a sequencer already." -> Reframe: a sequencer sends; this researches and books, which is the part that breaks during ramp.
Next-step options
- A scoped pilot on one segment for the first cohort of new BDRs.
- A data review of their current coverage ratio vs ramp plan.