AI at work

How to Practice Sales Calls with ChatGPT Role-Play

Set up a realistic ChatGPT sales role-play with a customer profile, one question at a time, objections and structured feedback.

Prompt Ready English interface preview

What this guide will help you do

  • Define the customer role, company situation and buying stage.
  • Ask for one question at a time.
  • Tell ChatGPT to challenge vague or unsupported answers.
  • Set feedback criteria before the conversation begins.

Ready-to-use prompt

Describe the buyer role, deal stage and scenario without real customer secrets. Use fictional or anonymised data, and keep product claims grounded in approved material.

Ready-to-use prompt
Role-play a first sales meeting with me.
Customer role: [job title and responsibilities]
Company context: [industry, size and current situation]
Problem: [business problem]
Buying stage: [early research, shortlist, renewal, etc.]
Difficulty: [basic, realistic or challenging]

Ask one question at a time and wait for my answer. If my answer is vague, ask a follow-up question. Do not help me sell during the role-play. After [10] exchanges, end the meeting and evaluate:
1. Discovery quality
2. Clarity of value
3. Evidence and examples
4. Objection handling
5. Next-step agreement
Give one specific improvement for each area.

Example: Practising a first meeting for a workflow tool

The customer is an operations manager at a 200-person company. They are curious but concerned about adoption, security review and implementation effort.

Customer: We already have several tools. What problem would this solve that our current process cannot?

After the response, the customer should ask a follow-up based on what was actually said, rather than moving through a fixed script. At the end, feedback should cite specific answers from the role-play.

The example below keeps the AI in the buyer role so the rep, not the model, has to answer the objection.

Ways to adapt the prompt

Change the buyer persona, objection and deal stage, but keep the rule that the model asks one question at a time and does not coach mid-role-play.

Before the meeting

Ask ChatGPT to generate likely questions and the evidence needed to answer them.

After a real meeting

Provide anonymised notes and ask for unclear answers, missed discovery questions and next-step risks.

Change the difficulty

Basic customers explain their needs; realistic customers require clarification; challenging customers question assumptions and evidence.

Checklist before using the result

Before treating role-play feedback as guidance, separate conversational coaching from real product, legal or pricing facts.

  • The customer profile resembles the real target account.
  • Questions are asked one at a time and follow the conversation.
  • Feedback refers to specific answers rather than generic sales advice.
  • Confidential account information is removed or anonymised.
  • The exercise ends with one or two behaviours to practise next.

What we checked hands-on

Prompt Ready

In Prompt Ready we tested role-play requests with different buyer roles, objections and conversation rules and reviewed the request before use.

  • Check dates, amounts and product claims against the source material
  • Do not paste confidential or personal information into an external AI service unless your rules allow it
  • Review the draft for the intended reader before sending or publishing it

Official sources and further reading

Frequently asked questions

Can ChatGPT simulate a specific person?

Use a role and business context rather than asking it to imitate a real individual. This keeps the exercise focused on the situation.

How long should a role-play be?

Ten to fifteen exchanges are often enough for one skill. Shorter sessions with a clear focus are easier to repeat.

Can it score my performance?

Yes, but define the criteria first and ask it to quote or reference the answers that support each score.

Prompt Ready

View the product

Use Prompt Ready to turn a task and a few constraints into a request you can review before pasting it into your AI service.

View the product