The Customer Success Role-Play: Three Scripts, Annotated

The role-play is the round that most reveals a CSM and the one candidates prepare least, because it can't be prepared the way a question list can. You don't know what the interviewer will say.

But you do know the shapes. There are roughly three, they recur across companies, and each is built to catch a specific instinct. Below is the dialogue for all three, with the reasoning underneath each line — not a script to memorise, but a demonstration of what the good version sounds like when it's happening in real time.

Before the role-play starts

Two things to do while the interviewer is still setting the scene, both of which are being observed.

Ask what you know. "Before we start — am I the CSM who has owned this account, or have I just picked it up? And have we spoken before?" This is not stalling. A CSM who inherits an account behaves completely differently from one who has a two-year relationship, and asking tells the panel you know that.

Ask what success looks like for the call. "Is this a call they booked, or am I calling them?" Who initiated changes the opening line entirely.

Then stay in character. The most common self-inflicted wound is breaking frame to explain your thinking — "so at this point I'd probably empathise and then..." The panel wants to hear you do it, not describe it. Save the commentary for the debrief at the end, where it's actually welcome.

Scenario one: the angry escalation

Set-up: "You're the CSM. Your customer's ops lead has just been told that a bug caused three days of incorrect reports, which they've already sent to their own client. They booked this call an hour ago. Go."

Them: "I don't really know what to say. We sent those numbers to our biggest client. Do you understand what that means for us?"

You: "I do. You've had to go back to your own client with a correction, and that's on us. Before anything else — has that conversation already happened, or is it still ahead of you?"

The instinct being tested is whether you rush to apology-plus-solution. Naming the consequence in their terms — "you've had to go back to your own client" — does more than "I'm so sorry". And the question is not sympathy, it's triage: if the conversation hasn't happened yet, there's something you can still do.

Them: "It's tomorrow morning."

You: "Then here's what I'd like to get you before tomorrow morning. A corrected set of the three days, and a short written note from us stating what the error was and that it originated with us — something you can forward rather than having to explain in your own words. Would that help, or would it make it worse?"

Two things happening. The written note transfers the blame to where it belongs and saves them from defending someone else's mistake. And "or would it make it worse?" is the line that separates experienced CSMs — sometimes a vendor letter escalates things internally, and assuming it helps is presumptuous.

Them: "…That would help. But I also need to know this won't happen again."

You: "I can't promise you that today, and I don't want to promise it badly. What I can do is get you the answer to what actually caused it by Thursday, with what's changing. If the honest answer turns out to be that this class of thing can recur, I'd rather tell you that and talk about how you'd catch it, than give you a reassurance that fails in three months."

Scenario two: the account that's quietly shopping

Set-up: "Routine quarterly check-in. Nothing is wrong. Twenty minutes in, they mention in passing that their new VP has 'asked them to look at the market'."

This one is harder than the angry call, because nothing is obviously wrong and most candidates keep the meeting on its agenda.

Them: "...anyway, it's probably nothing. The new VP just wants to look at the market. Standard stuff when someone joins."

You: "That's fair, it usually is. Can I ask two things about it, and then we can get back to the roadmap? First — is the VP looking at everything, or specifically at us? And second, do you know what they'd be measuring us against?"

Notice the permission and the promise to return. You've signalled that you're not going to derail the meeting into a save attempt, which is exactly what they'd expect and exactly what would make them regret mentioning it.

Them: "Everything, I think. Cost mostly — they've come in with a budget target."

You: "That's useful, and it changes what would be helpful from me. If this ends up being a cost conversation, the thing that usually decides it isn't price, it's whether anyone can show what the spend produces. Right now I don't think we could — we've never put together a view of what your team gets out of this in numbers you'd recognise. I'd like to build that, with you, before the VP asks for it rather than after. Would you be up for half an hour on it?"

The move: treat the signal as a deadline rather than a threat. And say what you don't currently have — the admission that no value case exists is credible in a way that a rehearsed pitch never is.

Them: "I mean, I'd be happy to look at that."

You: "Good. One more thing and I'll leave it — if it does get to a comparison, I'd rather be in the room than find out afterwards. Can I ask you to tell me when it gets serious, even if the answer's going badly for us?"

Asking explicitly for bad news early is a CSM instinct that panels look for and candidates rarely demonstrate.

Scenario three: saying no

Set-up: "Your customer wants a feature. It's on no roadmap and it won't be. They've asked three times. Tell them."

Them: "So — any news on the bulk export? It's been a few months."

You: "I've got an answer and it isn't the one you want. It's not going to be built. Not this year, and I don't want to tell you 'not yet' again, because that's what I've been doing and it isn't true."

Say it in the first ten seconds. The commonest failure here is a long preamble about roadmap prioritisation that delays a no everyone can feel coming, which reads as cowardice.

Them: "That's frustrating. We've been waiting on it to justify the renewal internally."

You: "Then I've made that harder and I'm sorry. Can I ask what the export was going to let you do? Because if the job is getting the data into your reporting stack, there may be another way to get there — and if the job is something else, I'd rather know that than guess."

The no is final; the underlying need is still open. Separating the two is the whole skill. Note that you don't promise a workaround before knowing whether one exists.

Them: "It's mostly for the monthly board pack. We rebuild it by hand."

You: "Okay. That's a different problem from an export and probably a smaller one. I'd like to look at whether the scheduled reports could produce that pack directly — I don't know yet, and I'll tell you by Friday either way. If it can't, then you're right that this is a real gap and I'd rather help you make that case internally than pretend it isn't one."

What the panel scores across all three

None of those are about product knowledge, and that's the point. The role-play exists because a question list can't distinguish between a candidate who knows these answers and one who can produce them while someone is unhappy with them.

Practising alone

You don't need a partner for the useful part. Take each scenario, and write the first two sentences you'd say — that's where every one of these is won or lost, and it's short enough to rehearse properly.

Then say them out loud and time the silence afterwards. Every script above has a moment where you ask something and stop. If you can't hold four seconds of quiet after "would that help, or would it make it worse?", the question turns into a suggestion and the scenario changes shape: why silence sounds professional.

For the analytical round — where they hand you an at-risk account and ask what you'd do — the worked version is here: Customer Success Manager Interview Case Study. And the full question set for the loop is here: Customer Success Manager Interview Questions.