Customer Refund Request Call: A CS Drill

Chapters
TL;DR
Handle a customer refund request with one agreed first response in three moves, in fixed order. Acknowledge the request in the customer's terms. Diagnose the underlying failure with ordered questions. State your refund policy before you promise anything. Then drill the sequence in scored roleplay until a rep passes on transcript evidence alone: the request is confirmed, the cause is confirmed by playback, the policy is stated, and no money is committed before all three.
- The first response has three moves in fixed order: acknowledge, diagnose, state policy.
- No commitment language appears before the cause is confirmed and the policy is stated.
- Let me see what I can do fails the drill however warm the tone.
- Score four rubric rows that a manager can verify from the transcript alone.
- Debrief one moment, let the rep self-diagnose, and schedule the re-run before the session ends.
Why the refund call gets one agreed first response
We run the refund call like any CS objection drill. The rep gets one agreed first response, drills it until it is automatic, and is scored against a rubric a manager can check from the transcript. Our position on objections is that rehearsal is what makes improvisation safe. A team keeps a small inventory of real objections, with one agreed first response for each. Strong reps then improvise the branches from understanding instead of being caught flat. A refund request belongs in that inventory.
The opposing view deserves a fair hearing. It holds that strong reps handle objections best by pure improvisation, and that an agreed first response makes them sound canned. The canned risk is real, and we design around it. The agreed response is a sequence of three moves, not a paragraph to recite. The rep chooses the words. The order is fixed.
We recommend agreeing the order before you score a single refund request conversation. With the order agreed, the manager checks three moves against the transcript and coaches the one that is missing.
A refund call should be scored against a first response the team has agreed, both its contents and its order.
What is the agreed first response to a refund request?
The agreed first response has three moves in a fixed order. Acknowledge the request, diagnose the underlying failure, then state the refund policy. No promise about money, credits, or exceptions comes before all three are done.
Acknowledge. The rep restates the request and the plan or amount, then asks for confirmation. The cause stays out of this move. A line that works: You're asking for a refund on the annual plan. Did I get that right? An acknowledgment should confirm what the customer asked for and never agree to grant it.
Diagnose. The rep asks what failed before discussing the refund itself, using at least two questions in a fixed order: surface first, specifics next, cost after. For example: Before we talk about the refund, tell me more about what went wrong. Then: Which part stopped working, and what did it do instead? The move ends when the rep plays back a specific failure and the customer says yes to it.
State policy. Once the customer confirms the cause, the rep states the policy as policy, in the team's agreed wording: Here is how our refund policy works: [your policy sentence, verbatim from your terms]. I'll walk you through where your situation sits against it. Load your actual policy wording into the rubric before anyone runs the drill. We want the policy sentence verbatim because a paraphrase cannot be scored against the agreed wording.
How should a rep diagnose the failure before stating policy?
Ask at least two questions from an ordered sequence, then play back the cause and get a yes. We borrow the order from the Sandler pain funnel: Tell me more… Can you be more specific… What have you tried… What did that cost you… How do you feel about that?
The scoreable behavior is the order. No question may come before one higher in the sequence: specificity before cost, cost before feelings. A rep who opens with How do you feel about that? has reversed the sequence, which is a misuse of the model. The rep may stop at any rung once the cause is clear, but never before the second question. In a refund call we treat the feelings question as optional.
Lines that work at each rung: Tell me more about what happened with the reports. Then Which report, and what did it show instead of what you expected? Then What have you tried since? Then What has that cost your team? The rep closes the diagnosis with a playback: So the export broke after the last update, and your analysts have rebuilt the weekly report by hand since. Is that the core of it?
The exit criterion for this move is buyer-verified evidence, not rep opinion: after at least two ordered questions, the customer says yes to a playback that names a specific failure. A rep who decides the cause privately and moves on has not met it.
Diagnosis is finished when the customer says yes to a playback naming a specific failure, after at least two ordered questions.
The rubric: four rows verifiable from the transcript
Score the call against four rows. Load the rows and your policy wording into whatever tool scores the session, or onto the sheet a human scorer uses.
Every rubric row should name evidence a second manager could find in the transcript without hearing the call.
Talk/listen ratio also sits on the scorecard. Treat it as an input to investigate when a row fails, not as a pass bar. If you run the drill in XL Roleplay, the report's flags link back to the exact moment in the transcript. That moment is the evidence the debrief needs.
| Rubric row | Pass evidence in the transcript | Fail evidence in the transcript |
|---|---|---|
| Acknowledgment | Rep restates the request and the plan or amount, then asks the customer to confirm | Rep jumps to an apology or a fix; the request is never restated |
| Diagnosis order | At least two questions in sequence order; customer says yes to a played-back specific failure | Fewer than two questions; a cost or feelings question before a specificity question; cause assumed and never played back |
| Policy statement | Policy stated in the team's agreed wording after the cause is confirmed | Policy omitted, paraphrased into a promise, or used as the opening line |
| Commitment discipline | No commitment language anywhere before the policy sentence | Let me see what I can do, I'm sure we can work something out, or I'll get that refunded before policy is stated |
What does a failing attempt sound like?
A failing attempt commits money, or the expectation of money, before the cause is confirmed and the policy is stated. The version to drill against is one line: Let me see what I can do.
Here is the full exchange as a scorer reads it. Customer: We want a refund. The tool hasn't done what we bought it for. Rep: I'm so sorry to hear that. Let me see what I can do. Acknowledgment fails because the request is never restated. Diagnosis fails because no question is asked. Commitment discipline fails because the rep has signalled a concession before knowing what broke or whether policy covers it. We score the line as a fail however warm the tone.
Put two other failures in the persona's path. The first is the wall. The rep opens with Our policy doesn't allow refunds after onboarding. Policy as an opening line fails the policy row and skips diagnosis entirely. The second is the near-miss. The rep diagnoses well, then says I'm sure we can work something out before the policy sentence. That attempt fails commitment discipline on one line. We recommend debriefing it first, because the rest of that transcript shows the rep can do the work.
We fail any refund call that offers money before the cause is confirmed, however warm the delivery.
Debrief one moment, then schedule the re-run
A debrief changes behavior only when it isolates one moment and one behavior. It lets the rep self-diagnose before the manager renders a verdict, and it ends with a scheduled re-run. Pick the earliest failed row and leave the rest for a later session.
A debrief script that fits the 10-minute block. Manager: Open the transcript at the customer's refund request. Read me your next two lines. Rep reads them. Manager: Which of the three moves is missing? Rep names it. Manager: Say the line you would use instead. Rep says it twice. Manager: We re-run the same persona later this week. Put it on the calendar now.
Keep the debrief out of the pipeline 1:1. We recommend a separate 15-minute 1:1 built around one flagged moment. It runs in addition to the pipeline review and never replaces it. Deal reviews cannibalize skill coaching, so the coaching block has to live on the calendar as a protected recurring commitment. For the full format, see Sales Roleplay Debrief: One Behavior, Re-Run.
A debrief that covers three mistakes from one refund call coaches none of them.
Run the refund drill this week
We recommend a weekly 30-minute format: 5 minutes of setup, 15 minutes of drill, and 10 minutes of debrief. Budget about 10 minutes beforehand to pick the scenario and choose the rubric row you will debrief if the rep fails.
Scenario. Build the persona from a real account in your book. The customer asks for a refund on a named plan and pushes for a yes in the first minute. Write the real cause so it surfaces only on the second specificity question, so a rep who skips diagnosis never reaches it. Give the persona your actual refund policy and one condition the policy does not cover. Reps. Run each rep through two or three attempts in the 15-minute block, and score every attempt against the four rows.
Pass bar. All four rows pass in one attempt, judged from the transcript alone. The request is restated and confirmed, the customer says yes to a played-back cause after at least two ordered questions, the policy is stated in the agreed wording, and no commitment language appears before the policy sentence. Failing attempt: I'm sorry about that, let me see what I can do anywhere before the policy sentence. Re-run a failed drill on the same persona, and schedule the re-run before the session ends.
If a refund request comes up in the first call after an account changes owners, run the account transition call drills alongside this one. When the policy answer is no, the next skill is delivering bad news to customers.
A rep passes the refund drill only when all four rows pass in one attempt, judged from the transcript alone.
Frequently asked questions
When can a rep agree to a refund on the call?
After all three moves are complete and the confirmed cause falls inside the stated policy. The drill fails commitment language only when it appears before the policy sentence. A yes that follows the policy statement is a pass.
Should we give reps a word-for-word refund call script?
We recommend scripting the policy sentence verbatim and leaving the rest to the rep. Score the three moves and their order, not the phrasing.
Can we run this as customer success roleplay without an AI tool?
Yes. A peer or manager plays the customer from the persona brief. A second person scores the four rows from a recording or transcript.
What if the customer refuses to answer diagnostic questions?
Name the reason once: I want the refund decision to rest on what actually went wrong, so I need one detail from you. If the customer still declines, state the policy and the next step. The attempt fails the diagnosis row but still passes commitment discipline if nothing was promised first.
Is talk/listen ratio part of the pass bar?
No. Use it as a diagnostic to investigate when a row fails. Keep the pass bar on the four transcript-verifiable rows.