What the hiring manager is actually assessing
The person playing the prospect has a scorecard in their head, and "did they explain the product well" is not on it. They already know their product. They want to see if you can sell theirs — or anyone's.
Four things carry almost all the weight. Question quality: do your questions open the prospect up, or close them into yes/no answers? Quantifying pain: when they admit a problem, do you turn it into a number, or nod and move on? Multi-threading: do you notice the person in front of you cannot sign, and go find who can? Next-step control: do you end on a specific calendared commitment, or a vague "I'll send something over"?
Everything else — tone, product knowledge, polish — is a tiebreaker. Candidates who lose this round almost always lose on one of those four, usually the first and the last.
How the round is usually structured
Prompt given in advance. You get a persona sheet a day or two out: fake company, industry, a rough problem, maybe a job title. This is the more common, more forgiving format — they expect you to show up having "researched the account." Do the homework; walking in cold when you were handed the brief is an instant knock.
Cold, no prompt. Some teams hand you nothing and say "we sell a customer-success platform, go." This tests whether you can build a discovery framework on the fly. Rarer, harder, and the bar for your questions is higher because you have no excuse to be generic.
The persona is deliberately vague. Either way, the interviewer plays the prospect guarded, opening with some version of "honestly, things are mostly fine." That is the test. A prospect who hands you the problem gift-wrapped is order-taking, not discovery. Expect friction and treat it as the point.
The prep framework: research, first five questions, the close
- 1
Research the fake company like it is a real one
Build a one-page account plan from the persona sheet. What does a company this size and industry usually struggle with? If they gave you "Director of Customer Success at a 60-person CS org," walk in already knowing CS leaders live and die by net revenue retention, that creeping churn is the nightmare they do not advertise, and that a director usually does not own budget. You are not guessing on the call — you are confirming a hypothesis you already built.
- 2
Plan your first five questions
Write them down before the call. Your first question sets the tone; make it open and specific, not "so tell me about your challenges." Something like "When an account is about to churn, how does your team find out — is that a system flagging it, or someone noticing in a QBR?" Each question should assume the problem exists and dig into the mechanics of it. Layer them: current process, then the gap in that process, then the cost of the gap.
- 3
Plan the close before you dial
Decide, in advance, what a win looks like. In a discovery call it is almost never a sale — it is a booked next meeting with the right people in the room. Know who the "right people" probably are (the economic buyer, maybe IT or procurement) and have the exact ask ready: "Let us get 30 minutes with you and your VP on the calendar this week." You do not want to be inventing your close in the last 90 seconds when your adrenaline is highest.
A full mock discovery call, start to finish
You are the AE. The interviewer plays Priya Nair, Director of Customer Success at Meridian Benefits (a fictional benefits-admin SaaS company, ~60-person CS org). She is polite, guarded, and downplays pain. What she has not said: churn has drifted from 8% to 12% this year, her team works reactively out of spreadsheets, and she does not own budget — the VP of Customer Experience does.
Priya
Thanks for making time. Things are mostly running fine on our side, honestly — but I said I would hear you out.
AE (you)
Appreciate you keeping the slot. I will make it worth it — and if it turns out there is nothing here for you, I will be the first to say so. Before I say a word about us: when a customer is heading toward not renewing, how does your team usually find out it is happening?
Priya
Usually the CSM flags it. They know their accounts.
AE (you)
That makes sense when someone has the context. Where does that break down — is it ever an account nobody flagged, and then the churn shows up at renewal as a surprise?
Priya
I mean, it happens. Not constantly, but yeah, a couple times a quarter something slips that we should have caught.
AE (you)
Got it. And when one slips — what is a typical account worth to you annually? Ballpark is fine.
Priya
Depends, but call it forty to sixty thousand a year for a mid-size one.
AE (you)
So a couple a quarter at fifty thousand is somewhere north of four hundred thousand a year walking out the door — not from bad service, just from not seeing it coming. Is that roughly how you would size it, or am I off?
Priya
That is... probably not far off, honestly. It is more than I like to say out loud.
AE (you)
That is exactly the number worth chasing. Has your overall retention moved this year versus last, even a point or two?
Priya
It has drifted. We were around 8% net churn, we are closer to 12% now. I would rather that not be in a slide deck anywhere.
AE (you)
Understood, and thank you for trusting me with it. Four points of net churn on your base is a real revenue line, not a rounding error. What is the team using today to spot the at-risk accounts earlier — a health score, or is it living in spreadsheets and the support tool?
Priya
Spreadsheets, mostly. And whatever the CSMs carry in their heads.
AE (you)
That is the gap. Spreadsheets tell you what already happened; they do not warn you. The reason I ask — the teams that pull churn back down usually do it with an early-warning signal, not more headcount. But before I get ahead of myself: if you did decide this was worth fixing, who besides you would need to be in that conversation?
Priya
Realistically the VP of Customer Experience. She owns the retention target and the budget. And IT would have to bless any new system.
AE (you)
That is helpful. Here is what I would suggest, and tell me if it is too fast: rather than me sending a deck that gets forwarded and dies, let us get 30 minutes with you and your VP, where I bring a simple model of what pulling churn from 12 back toward 8 is worth on your base. If the number is not compelling, we all walk. Would you be open to that?
Priya
That is reasonable. She is slammed, but a churn number will get her attention.
AE (you)
Then let us make it easy to say yes to. I am open Thursday afternoon or Friday morning — do either of those tend to work for her, and should I send the invite through you? I will also keep it tight enough that IT can be looped in on the follow-up rather than the first call.
Priya
Send me Thursday at 2. I will get her on it.
AE (you)
Done. I will send an invite for Thursday at 2 with a one-line agenda — size the churn number, decide if it is worth a deeper look. Thanks, Priya. This was a good use of the time.
Coaching note: Notice the shape: not a single feature named until the close, and even then only in passing. The AE assumed the problem, layered questions to surface the real churn number, converted "a couple slip" into a dollar figure, uncovered that Priya cannot sign, and booked a specific meeting with the economic buyer. That is a passing call.
Run this exact scenario against an AI buyer
Practice the guarded-CS-director discovery call above, live, and get a scorecard on your discovery depth, quantification, multi-threading, and close.
Practice this scenarioHow the call gets scored
| What they weigh | Passing looks like | Failing looks like |
|---|---|---|
| Discovery depth | Layered questions that get under "things are fine" | Accepts the first vague answer and moves on |
| Quantifying pain | Turns the churn into a dollar figure the prospect confirms | Leaves the problem as a feeling, never a number |
| Multi-threading | Finds the economic buyer and IT, asks to bring them in | Sells only to the person on the phone |
| Objection handling | Redirects the "just send a deck" brush-off into a question | Agrees to send a deck and ends the thread |
| Next-step control | Books a specific meeting with the right people | Ends on "I will follow up" with no date |
The three ways candidates blow it
Pitching too early. The prospect says "we run on spreadsheets" and the candidate lights up: "Oh, our platform has automated health scoring and..." Dead. Every feature you name before you have quantified pain is a feature they can now say no to. Hold your product until the prospect has practically asked for it.
Interrogation-style questioning. Ten questions in a row with no reaction to the answers feels like a deposition, not a conversation. React instead — reflect the number back, connect it to their last point, show you were listening. "A couple a quarter at fifty grand — so north of four hundred thousand a year, is that how you see it?" That is a question that also proves you heard them.
Ending without a concrete next step. The most common single reason to fail. The call goes well, rapport is good, and the candidate closes with "This was great, I will put together some materials and follow up." No date, no attendees, no ask. In the notes that reads as: cannot control a deal. Always leave with a calendar hold and the right names on it.
The debrief: "how do you think that went?"
Almost every interviewer ends by asking you to self-assess. This is not small talk — it is part of the score. They want to see whether you can coach yourself, because a rep who cannot see their own gaps cannot improve on the job.
Do not say "it went great," and do not trash yourself either. Give a specific read: one thing you ran well, one thing you would change, and why. "The discovery held up — I got the churn number quantified. If I ran it again I would have named the VP earlier instead of at the close, because I left multi-threading a little late." That tells them you understand what the round measured. If it fits, ask one question back — "Where would you have pushed harder?" — which signals you are coachable, the exact trait they want to hire.
Prep the behavioral round too
Most AE loops pair the mock call with a behavioral round. Practice structured "Tell me about a time..." answers and get scored on the STAR framework.
Practice the behavioral round