Overview
Sales training doesn't scale past a handful of reps. This Custom GPT plays the role of a specific ICP (e.g. VP of Ops at a 200-person e-commerce brand), lets a rep run a mock discovery call, then grades the conversation against MEDDIC or your custom qualification framework.
How it works
- GPT builder with a Knowledge file of your ICPs and qualification framework.
- System prompt: 'You are {persona}. Respond only in-character. When the rep says /debrief, drop character and grade against the framework.'
- Rep runs the conversation in one thread.
- Debrief produces a structured scorecard + recommended next questions.
Benefits
- Every rep gets unlimited practice reps without a coach.
- Consistent scoring rubric across the team.
- New hires ramp in weeks, not quarters.
Use cases
- Sales enablement teams onboarding new AEs.
- SDR teams practicing objection handling.
- Founders selling before they hire a sales lead.
Step-by-step guide
Step 1: Author 3-5 persona files
Each persona: role, company profile, current tools, top pains, budget signals, common objections.
Step 2: Define the framework
MEDDIC, BANT, or your custom fields. Include the grading rubric (1-5 per pillar).
Step 3: Write the meta-prompt
Emphasise staying in character until the /debrief trigger. Give one paragraph on tone (curt CFO vs curious VP of Ops).
Step 4: Test with an internal rep
First-pass: does the persona feel real? Second-pass: is the scoring specific?
Step 5: Publish to team
Share within workspace. Add link to onboarding docs and weekly coaching agendas.
Step 6: Rotate personas monthly
Add new personas from real deal debriefs. Retire ones that no longer reflect your market.
Example
Rep: 'How are you handling returns today?'
Persona: 'Manually. It's the reason my ops lead threatened to quit last quarter.'
/debrief → Metrics: 4/5 (probed pain), Economic Buyer: 2/5 (didn't ask), Decision Criteria: 3/5.