The case study round in a customer success interview is not really about the account. It's about what you do in the first ninety seconds, before you know enough to have a plan.
Most candidates start solving. They read the brief, spot the obvious problem, and propose a save plan — which is exactly what the case is built to catch, because the obvious problem is almost never the one that's actually going to lose the renewal.
Below is a realistic case with the data you'd be handed, a full worked answer, and the three wrong turns that come up most often.
The brief
You've just inherited this account. Renewal is in 90 days. Walk us through what you'd do.
Northwind Logistics — $48,000 ARR, customer for two years, renewed once.
Signal Detail Usage Weekly active users down from 34 to 22 over two quarters Champion Left the company four months ago. No formal replacement Support Three open tickets. One is six weeks old, tagged "waiting on customer" Survey Last NPS response was a 6, from a user we don't otherwise know Exec sponsor Named in the original deal. Has not attended a call in a year Commercial Their procurement team asked for a copy of the contract last month Product They're on a legacy plan. Two features they've asked for are on the roadmap
That's roughly what a real brief looks like: seven signals, some of them noise, one of them urgent, and no indication of which is which.
What most candidates do
Three wrong turns, in order of how often they show up.
They fix the tickets. It's concrete, it feels productive, and it's the least important thing on the page. A six-week-old ticket tagged "waiting on customer" usually means nobody cares about it enough to reply — that's a symptom, not the disease.
They book a call with the exec sponsor. Reasonable-sounding, and it burns the one piece of leverage you have on someone who hasn't engaged in a year and has no reason to take the meeting.
They build a 90-day plan on slide one. Confident, thorough, and built on air. The panel is watching for whether you'll commit to a plan before establishing what's actually happening, because that's the behaviour that loses real renewals.
The signal that matters most on that page is the procurement request. It's one line, it's easy to skim past, and it means the renewal is already being processed by someone — possibly as a cancellation, possibly as a competitive review. Everything else can wait a week. That can't.
The worked answer
Here's the whole thing at interview pace. It's about four minutes spoken.
Start with what I don't know.
"Before any plan, there are three things I'd want inside 48 hours. First, what the procurement request actually is — renewal paperwork, a competitive review, or a budget exercise. That single fact changes everything else I'd do. Second, who is making the renewal decision now that the champion has gone, because right now I don't know if that person even knows we exist. Third, what the 22 remaining users are actually doing — whether we lost a whole team or lost the light users."
Then read the signals against each other.
"The usage drop and the champion leaving are almost certainly the same event, and the timing fits — the champion left four months ago, the decline runs over two quarters. So my working theory is that we lost the team the champion ran, and the 22 who remain are a different group who use us for something narrower. If that's true, this isn't a satisfaction problem. It's an account that quietly shrank to a smaller use case and nobody re-sold it."
Then say what would change the theory.
"What would tell me I'm wrong: if the 22 remaining users are the same people who were always here and we lost the light users instead, then usage is fine and this is a pricing conversation, not a value one. I'd check that first because it flips the plan."
Then the plan, with the right first move.
Week 1. Establish the procurement request through whoever we still talk to. Find the new decision-maker and get a meeting on any pretext — a "new CSM introducing myself" call is a legitimate reason and costs them nothing. Pull the usage breakdown by team.
Weeks 2–4. If the theory holds: build the case around what the remaining 22 actually get, in their numbers, not ours. Find out whether the team we lost was replaced by a competitor or by nothing — those are different conversations. Close or kill the three tickets so nothing embarrassing is open during a renewal review.
Weeks 5–8. Get the exec sponsor back in the room, but with a reason — a result, or a decision they need to make — not a check-in. If there's a competitor in the account, put the comparison on the table myself rather than waiting for it.
Weeks 9–12. Renewal conversation with a realistic shape. If the account has genuinely shrunk, I'd rather propose a smaller renewal that matches the real usage than defend a number we can't support.
Then the part almost nobody says.
"And I'd tell my manager in week one that I don't think this renews at $48,000. Right now I'd forecast it as at-risk, probably closing lower. I'd rather be wrong in that direction than have the number move in the last two weeks of the quarter."
That last paragraph is what the case is really for. Everything before it can be learned from a playbook. Volunteering a bad forecast early — before you're forced to — is the behaviour that separates someone who manages a book of business from someone who manages accounts. Every panel member has been surprised by a renewal that collapsed in the final fortnight, and they are listening for whoever will never do that to them.
What each part was scoring
| What you did | What it told them |
|---|---|
| Named what you don't know, first | You won't build plans on assumptions |
| Connected two signals into one theory | You read accounts, not dashboards |
| Said what would disprove the theory | You hold it as a hypothesis, not a conclusion |
| Prioritised procurement over the tickets | You can tell urgent from visible |
| Proposed a smaller renewal as an option | You're commercially realistic, not just loyal |
| Forecast it down in week one | You'd rather be early than right |
Everything, immediately
“I'd reach out to all remaining users, run a health check, clear the support backlog, book a business review with the exec sponsor, and put together a value summary for the renewal.”
Nothing here is wrong, and it's what a candidate says when they haven't decided what matters. With ninety days and one person, this plan does six things badly. The panel hears the absence of a first move.
One thing, then a fork
“Week one is one question: what is procurement doing. Everything after that depends on the answer, and I'd rather tell you two versions of the plan than one that assumes.”
Fewer actions, more judgement. And offering the fork explicitly invites the interviewer to pick which branch they want to hear, which turns a monologue into the conversation they were hoping for.
If they hand you a role-play instead
Some loops turn the case into a live conversation halfway through — the interviewer becomes the new decision-maker and you have to open the call. The switch is deliberate and it catches people who prepared an analysis but not a first sentence.
The opening that works is short and gives them an easy exit: "I've just picked up your account and I wanted to introduce myself before anything commercial comes up. I've got about fifteen minutes of questions rather than a pitch — is now still okay?" Then ask, and let the silence sit after each question — that pause is the hardest part to hold and the most useful: why silence sounds more professional than you think.
The wider set of conversations these loops draw from — the angry customer, the quiet account, the renewal, saying no — is covered here: Customer Success Manager Interview Questions.
Preparing your own version
Bring one real at-risk account you've handled, with the numbers. Not a success story — an at-risk one, ideally one that ended badly, because the panel will ask and a candidate who only has wins has either been lucky or is editing.
For that story you need four things: what the data said, what you thought was happening, what it actually turned out to be, and what you'd forecast differently now.
Practical target: read the brief above once, then talk for ninety seconds out loud without proposing a single action — only what you'd want to know and what you think is happening. If that's uncomfortable and you find yourself reaching for a plan, that's the exact instinct the case is built to expose, and it's much cheaper to find out here.
