25 AI Roleplay Scenarios for Frontline Teams: Sales, Service, Safety, and Collections

Quick answer
25 ready-to-use AI roleplay scenarios for sales, customer service, safety, and collections teams - with prompts, scoring criteria, a...
Most frontline skills aren't knowledge problems. Your store associate knows the product spec sheet. Your collection agent knows the RBI guidelines. What they haven't done is said the words out loud to a difficult customer - and the first time they do, it's with a real one.
That's the gap AI roleplay closes. Instead of waiting for a trainer to run practice sessions branch by branch, an AI agent plays the customer - skeptical, angry, confused, or in a hurry - and the worker practices until the response is muscle memory. If you're new to the concept, start with our explainer on what AI roleplay training is, then come back for the scenario library.
Below are 25 scenarios we see work across frontline teams, organized by function. Each can run as a chat roleplay or a voice call. With Leap10x Converse, the AI calls the worker on WhatsApp, runs the roleplay in their own language, and scores comprehension automatically - no trainer scheduling required.
How to use these scenarios
Three rules make roleplay stick:
- One objection per scenario. Don't build a mega-scenario with five twists. Short, repeatable reps beat marathon simulations.
- Score against observable behavior. "Did the agent acknowledge the complaint before responding?" is scoreable. "Was the agent empathetic?" is not.
- Run them in the worker's language. A Tamil-speaking promoter should practice in Tamil, because that's the language the real conversation will happen in.
Sales roleplay scenarios (retail, FMCG, BFSI)
1. The price objection. Customer: "The same AC is ₹3,000 cheaper online." Goal: acknowledge, pivot to installation/warranty/exchange value, close without discounting.
2. The silent browser. Customer answers in one-word replies. Goal: use two open-ended questions to uncover the need.
3. The competitor loyalist. "I've always used the other brand." Goal: validate the choice, ask what they like, position one differentiator.
4. The feature overload trap. Customer asks a simple question; agent must answer in under 30 seconds without dumping the full spec sheet.
5. The upsell moment. Customer commits to base model. Goal: offer one relevant add-on without pushiness.
6. The insurance mis-selling trap (BFSI). Customer asks "guaranteed returns?" Goal: explain honestly and stay IRDAI-compliant. Pairs well with our POSP insurance agent training playbook.
7. The credit-card cross-sell. Bank teller spots an eligible customer. Goal: pitch in two sentences, handle "I already have three cards."
Customer service scenarios (QSR, hospitality, telecom, e-commerce)
8. The wrong-order blow-up. Customer received the wrong dish and is loud about it. Goal: apologize, resolve, don't argue policy.
9. The refund demand outside policy. Goal: hold the line while offering alternatives.
10. The language mismatch. Customer switches to a language the agent half-knows. Goal: slow down, confirm understanding, escalate gracefully.
11. The review threat. "I'll give you one star." Goal: de-escalate without bribing.
12. The VIP guest with an odd request (hotels). Goal: never say a flat "no"; offer the closest yes.
13. The billing dispute. Customer insists they were overcharged. Goal: verify calmly, explain the bill line by line.
14. The delivery delay call. Rider is 40 minutes late; customer is calling. Goal: honest ETA, no invented excuses.
Safety and compliance scenarios (manufacturing, construction, logistics)
15. The shortcut pressure. A senior colleague says "skip the lockout, it takes too long." Goal: refuse safely and report.
16. The missing PPE confrontation. Worker must ask a peer to put on a helmet without triggering a fight.
17. The near-miss conversation. Goal: report a near-miss to a supervisor clearly - what, where, what almost happened.
18. The forklift right-of-way dispute. Pedestrian worker vs. forklift operator. Goal: state the rule, stand ground.
19. The auditor question. External auditor asks "show me how you'd handle a chemical spill." Goal: walk through the SOP verbally.
20. The fatigue admission. Driver must tell a dispatcher they're too tired for another trip. Pairs with our driver safety training guide.
Collections and field-force scenarios (NBFC, MFI, telecom)
21. The genuine hardship case. Borrower lost their job. Goal: stay within RBI conduct rules, offer restructuring path, document the promise.
22. The hostile defaulter. Borrower gets abusive. Goal: stay calm, never threaten, close the call within guidelines.
23. The third-party contact. A family member answers. Goal: protect borrower privacy - say nothing about the loan.
24. The partial-payment negotiation. Goal: secure a dated commitment, not a vague "next week."
25. The field visit refusal. Customer won't open the gate. Goal: leave compliant notice, log the visit, no escalation.
What good AI roleplay scoring looks like
For each scenario, define 3–5 binary checks. Example for scenario 21:
- Did the agent identify themselves and the institution? (Y/N)
- Did they acknowledge the hardship before discussing payment? (Y/N)
- Did they stay within permitted calling conduct? (Y/N)
- Did they secure a specific date and amount? (Y/N)
AI voice agents score these automatically across thousands of calls - something a central QA team would take weeks to sample manually. Indian BPOs are already using this to cut agent ramp time by half.
Rolling this out in a week
- Day 1–2: Pick 5 scenarios that map to your top failure points (mystery-audit fails, escalation logs, lost-sale reasons).
- Day 3: Load them into your roleplay tool with scoring rubrics in the local languages your team speaks.
- Day 4–5: Pilot with one region; workers get a WhatsApp message, tap, and practice.
- Week 2: Compare scores by branch, coach the bottom quartile, add scenarios monthly.
Because Leap10x runs roleplay over WhatsApp - chat or AI voice call - there's no app to install and no lab to book. Workers practice between customers, in 70+ languages, and managers see scores in real time.
Call to Action
Want your team practicing these scenarios by next week? Book a free Leap10x demo - bring one tough customer conversation your team keeps fumbling, and we'll turn it into a live AI roleplay before the call ends. Request a demo →


