Forty seconds in a hallway
It's 7:40 on a Tuesday in the second week of fieldwork. A first-year associate has spent the afternoon in the journal entry population and found something she can't explain: a cluster of round-dollar entries posted after the close date, all by the same user ID, all to the same reserve account, none with attached support. It could be a lazy cleanup process. It could also be exactly what it looks like.
The engagement partner is walking past the client conference room toward the elevator, coat over his arm. He glances in and says, "Anything I need to know before tomorrow?"
She has about forty seconds. What she says in them determines whether this becomes a documented, resolved procedure or a line in a deposition three years from now.
Almost every accounting program teaches her how to find those entries. Almost none of them teach her how to spend those forty seconds.
The failure point isn't detection
Read enough enforcement actions, inspection findings, and post-mortems and a pattern gets uncomfortable: in a striking number of them, somebody on the engagement team saw it. The anomaly was noticed. It was mentioned. And then it dissolved — into a hedged sentence at the end of a status update, into a note in a workpaper nobody escalated, into a shrug at 7:41 in a hallway.
The most dangerous sentence in any audit file is "I mentioned it."
This is consistent with what the profession's own fraud research keeps finding: tips and human observation, not analytics, surface most occupational fraud. Detection is very often a human being noticing something and deciding to say so. Which makes the saying-so part of the mechanism, not an afterthought to it — and makes audit escalation training a technical competency, not a communications elective.
Here is the uncomfortable version of the problem for accounting educators. We assess whether students can identify a red flag. We do not assess whether they can make a partner care about one. The first skill without the second produces an auditor who is technically correct and functionally silent.
Why escalating upward is genuinely hard
It's worth being precise about what makes this difficult, because "just speak up" is not advice — it's a restatement of the problem. Three forces work against a junior auditor in that hallway.
The status gap. The partner has thirty years, a title, and an economic stake in the engagement's efficiency. The associate has eleven weeks. Every social instinct a well-socialized twenty-two-year-old has developed says: do not create work for the most senior person in the room over something you're not sure about.
The certainty tax. The associate does not have proof. She has an indicator. And the internal bar she sets for herself — am I sure enough to say this out loud? — is almost always calibrated far too high, because she is imagining the embarrassment of being wrong and not imagining the cost of being right and quiet. Teaching students that an indicator is a legitimate thing to escalate, with the confidence level stated explicitly, is one of the highest-value corrections available in an auditing course.
Time pressure as a social signal. The partner is leaving. The coat is on. Everything about the moment communicates make this quick or make this go away. Students read that correctly and then respond to it incorrectly, by compressing the substance instead of the framing.
None of these are knowledge deficits. All three are behavioral, and all three are trainable — but only through practice, which is the thing the traditional curriculum has never been able to supply at scale.
What a good forty seconds actually sounds like
The most useful thing a fraud detection classroom exercise can do is give students a structure they can execute under pressure. A strong escalation has four parts, in this order:
1. The assertion, first. "I found something in the journal entries I think we need to look at before we sign off." Not a preamble. Not "so I was going through the JE population and, um, this might be nothing, but —" by which point the partner has decided it's nothing, because she told him it was.
2. The evidence, specifically. "Forty-one entries, all round dollar, all posted after the close date, all by one user ID, none with support attached." Concrete, countable, checkable. This is the part students are actually well-trained for, and it's the part they tend to bury under the hedging.
3. Calibrated confidence. "I'm not saying this is fraud. I'm saying I can't get comfortable with it from what's in the system, and I don't think we should conclude on the reserve until it's explained." This is the move that almost nobody executes without practice, and it is the one that makes escalation safe to do. It separates what I observed from what it means — which protects the student from over-claiming and protects the audit from under-reacting.
4. A specific ask. "I'd like an hour tomorrow to pull the support and talk to whoever owns that user ID." Escalation without an ask puts the entire decision burden on the listener, which is exactly how issues die in a busy person's inbox.
Four parts, roughly forty seconds, and a student can be taught to deliver them cleanly. But not by reading them in a chapter. They have to be spoken, badly, several times, to someone who reacts — before they can be spoken well to someone who matters.
Why the usual auditing course activities don't get there
Most programs already teach fraud well on the analytical side: the fraud triangle, red flag inventories, analytical procedures, an Enron-era case or two. That content is solid. It just stops one step short of the behavior.
Written cases have no listener. A student writing "I would escalate this to the engagement partner" in a memo has demonstrated that they know escalation is the right answer. They have demonstrated nothing about whether they could do it. The two are barely correlated — the students who write the most confident memos are frequently the ones who go quiet in the room.
The classroom inverts the very dynamic being taught. When a student raises a concern to a professor, they are speaking to someone whose institutional role is to be receptive, encouraging, and glad they asked. That is the opposite of a partner with his coat on. We are, without meaning to, giving students reps in an environment engineered to remove the exact pressure the skill is designed to withstand.
Live role play doesn't scale, and peer role play removes the stakes. A practitioner guest who plays the partner for a few volunteers gives thirty-five other students a spectator experience. A classmate playing the partner has neither the status nor the incentive to create pressure — and pressure is the entire pedagogical payload.
So the behavior gets covered, discussed, agreed with, and never practiced. Students graduate knowing they should speak up, which is not the same as being able to.
What AI role play changes
AI role play for higher education removes the constraint that made this impossible: you can now give every student a partner who is senior, busy, and unimpressed, on demand, as many times as it takes.
The variation is what makes it pedagogically serious. The same fact pattern can be run against:
- The distracted partner, who is half-listening and will let a hedged escalation slide right past.
- The time-boxed partner, who says “you’ve got thirty seconds” — forcing assertion-first structure.
- The skeptical partner, who pushes back: “You’re saying the CFO’s team is cooking the reserve? On forty-one entries?”
- The soft deflector, who is warm and agreeable and gently reframes the issue as immaterial — the hardest version, because there’s nothing to push against.
That last one deserves attention. Students almost universally find the friendly deflection harder than the hostile challenge, and almost universally have never been warned that it exists. Being able to hold a concern in the face of someone pleasant who wants it to go away is a specific, learnable skill, and simulation is the only practical way to give students repetitions at it.
What you get on the other side is a record: a transcript and recording of every attempt, so a student can watch themselves bury the lede, and an evaluation scored against the same rubric for everyone, rather than against whichever practitioner volunteered that semester.
Designing the module: three passes
For faculty building this into an auditing or forensic accounting course, three escalating passes work well.
Pass one — the hallway (early in the term). Forty seconds, receptive partner. The only objective is structure: assertion, evidence, calibrated confidence, ask. Score whether all four appear and in what order. Most students fail on ordering, and seeing that in a transcript is more persuasive than being told.
Pass two — the pushback (mid-term). The partner challenges the interpretation and applies mild time pressure. The student must hold the observation while conceding the interpretation is open. Score whether the factual claim survives intact and whether the student over-claims fraud under pressure — both failure modes matter, in opposite directions.
Pass three — the partner declines to act (late term). The hardest scenario, and the one closest to real enforcement histories. The partner hears it and decides it doesn't warrant additional work. The student must now document the concern accurately in the file and know what the firm's consultation and escalation channels are. Score the documentation and whether the student can articulate the next step without either dropping it or making an accusation.
That third pass is where ethics training for accounting students stops being abstract. It is one thing to agree in a seminar that you would escalate. It is another to have actually done it, three times, to someone who didn't want to hear it, and to have a recording of yourself doing it well.
The program-level argument
For a chair or dean weighing this against everything else competing for curricular space:
It produces direct evidence for assurance of learning. Ethical reasoning and communication sit in nearly every program's stated learning goals and are almost always measured indirectly — through written work or composite course grades. A scored behavioral simulation yields a direct, rubric-anchored measure of an observed behavior, collected consistently across sections. For AACSB assurance of learning purposes, that is a substantially stronger artifact than a reflection paper.
It is experiential learning without the coordination cost. The constraint on experiential learning business school initiatives is never enthusiasm; it's logistics — partners, scheduling, supervision, quality variance across sites. Simulation delivers repeated practice with none of that, plus the thing no real placement can offer: permission to fail and immediately try again.
It addresses the gap firms are already paying to close. Firms run onboarding programs on exactly this behavior. A program that can show recruiters its graduates have documented, scored reps at upward escalation is making a claim about employability that competitors phrase as an aspiration.
The asset works twice. Scenario libraries built for the auditing course transfer directly into executive education and firm-sponsored training, where the same gap is being addressed at meaningfully higher price points.
The short version
We have built an entire curriculum around teaching students to notice things, and almost none of it around what happens in the forty seconds after they do. The professional consequences of that gap are visible in every enforcement action where somebody knew.
Detection is a skill. Escalation is a separate skill. Until recently, only one of them could be practiced at scale. That's no longer true — and the question for accounting programs is whether their graduates will get those reps in a classroom or in a hallway with the elevator door closing.
Foretell AI lets faculty build conversational simulations — including staff-auditor-to-partner escalation scenarios like the one above — with configurable counterparties, transcripts, recordings, and rubric-based evaluation. If you're mapping ethics and communication outcomes to assurance-of-learning goals in an accounting program, we're happy to walk through how other departments have structured it.