Casey does not fail you for a slightly imperfect framework. It fails you for an early, irreversible choice — usually which dataset to open — and for leaving the last three questions blank because you spent four minutes narrating your structure into a chat box that does not care.
That is the thesis. BCG built a chatbot screen so it could score case skills without putting a partner in the room. The bot will not rescue a vague answer, will not let you rewind, and will not tell you whether the number you typed was close. After reading this, you should know what BCG itself publishes, which details are candidate folklore, how to allocate ~30 minutes across question types, and how I score a Casey-style run when I am the one reading the answers.
What BCG actually says (and what it does not)
On its Mexico careers page, BCG names the product clearly: the Online Case Experience (OCE) is "an interactive exercise where you solve a business case guided by Casey, a chatbot." Preparation, BCG says, is "the same as for a Case Interview," on a personal computer with reliable internet, in English, and it "takes approximately 30 minutes."
Panama's careers page adds the skills list and the closing act: the Online Case Experience tests "quantitative skills, data interpretations, calculations, and business logic," you get 30 minutes, and "at the end, you must record a 1-minute video with the final recommendation."
For U.S. Associate recruiting, BCG's on-campus Associate page says a select group is invited to complete "the online case (~35 minutes)" roughly two weeks after the application deadline, and that candidates who already took it will not be asked again.
What BCG does not publish on those pages: exact question counts, whether you can go back, calculator policy, or a public scoring rubric. Those details live in invitation emails and office-specific candidate guides. Treat Reddit and Glassdoor as folklore, not fact. Your invite wins every dispute.
A 2020 BCG alumni interview with Eugene Goh, who helped create Casey while on BCG's recruiting team, is useful context: the design goal was an automated test that "closely mimicked the actual case interview," using "structured questions, free-text answers, and a video summary," scored consistently on "problem structuring, business judgment, quantitative rigor, [and] succinct communication." That is the skill set. It is not a psychometrics test and it is not a video game.
How Casey sits in the funnel
| Signal in your invite | What it usually is | Prep stance |
|---|---|---|
| Online Case, OCE, Casey, chatbot case | Chatbot business case | This guide |
| Consulting Career Assessment / CCA | Separate multidimensional screen | Different product; do not drill cases for it |
| Live case interview | Partner or consultant interview | Standard BCG casing |
BCG Mexico lists OCE for applicants and CCA for Associates as parallel online assessments before live interviews. The U.S. Associate flow often sends CCA soon after apply, then Casey only to a subset. Panama treats the Online Case Experience as the gate before first-round cases. Offices differ. Match the wording in your email.
Live BCG cases after you pass are still candidate-led and scored in the room. Casey is a filter, not the offer. If you clear it, you still need BCG live-case prep.
Timing: build a clock, not a vibe
Published durations from BCG sit in a narrow band: about 30 minutes (Mexico, Panama) or about 35 minutes (U.S. Associate online case). Candidate reports of "25–30 minutes for 8–10 questions" are common folklore until your invite confirms them. Assume 30 minutes and 9 questions: about 3 minutes 20 seconds per item if you split evenly — which you should not.
| Phase | Share of clock (Ned's estimate) | Job |
|---|---|---|
| Opening structure / which data to open | 10–15% | Name the decision; pick the exhibit that tests it |
| Exhibit reads + short math | 55–65% | Units first, then arithmetic |
| Brainstorm / next analysis | 10–15% | Grouped ideas, one recommended next step |
| Final recommendation (text or 60s video) | 10–15% | Answer first, two reasons, one risk |
If question 2 already took five minutes, you are behind. Blank later items score worse than a slightly rounded number on an early calc. Casey's exact rubric is not public; the pacing logic still holds.
Question types and what a good answer looks like
Expect a mix of formats inside one scenario. Shapes, not a leaked script:
| Type | What the interface wants | What I listen for as a scorer |
|---|---|---|
| Objective / structure | Short taxonomy or priority list | Decision-linked buckets, not textbook MECE theater |
| Dataset selection | Pick which table or chart to open | Evidence that matches the decision metric |
| Exhibit + MCQ | One best option | Title, units, period, then implication |
| Numeric entry | A number (often with on-screen calculator — follow invite) | Correct relationship before the digits |
| Short free text | Two to five crisp lines | Direct claim; no essay |
| Closing recommendation | Text and/or ~60s webcam video | Answer → drivers with numbers → risk |
Dataset selection is the Casey tell. Live cases let you ask for the pickup-density table after you realize revenue is the wrong ranking. Casey often forces the choice up front. Open the prestige metric and you never see the right numbers.
Worked example: Harborline cold-chain routes
Original practice case. Invented company. Reproduce the math on paper.
Prompt. Harborline Logistics runs refrigerated trucks for grocery DCs. Margin fell from 8.2% to 5.1% of revenue in twelve months. Casey asks which analysis to open first to decide whether to cut the three weakest lanes this quarter.
Options.
- A. Total revenue by lane, ranked high to low.
- B. Contribution margin per loaded mile by lane, plus empty-mile share and average dwell at receiving docks.
- C. Competitor rate cards in the same zip codes.
- D. Driver Net Promoter Score by depot.
Pick B. The decision is which lanes destroy contribution after empty miles and dwell. A is a decoy for "big" lanes. C matters later for pricing response. D is a people issue, not this quarter's cut list.
Exhibit math. Three candidate lanes:
| Lane | Loaded miles / week | Contribution / loaded mile | Empty-mile share | Dwell cost / week |
|---|---|---|---|---|
| North | 4,200 | $1.40 | 12% | $1,800 |
| Central | 3,100 | $1.15 | 28% | $3,400 |
| South | 5,000 | $1.55 | 9% | $1,200 |
Model empty miles as diluting contribution: effective contribution ≈ loaded miles × $/mile × (1 − empty share), then subtract dwell.
- North: 4,200 × 1.40 × 0.88 − 1,800 = $3,374
- Central: 3,100 × 1.15 × 0.72 − 3,400 = −$833
- South: 5,000 × 1.55 × 0.91 − 1,200 = $5,853
Central is negative on this definition. Cut Central first. Annualizing: about $833 × 52 ≈ $43k per year on this simplified weekly view — enough for a regional P&L, not enough to pretend you modeled fuel hedges.
60-second video (structure, not a script). "Cut Central this quarter. On the exhibit, Central is about −$830 per week after empty miles and dwell, while North and South stay positive. Keep North and South; renegotiate Central dwell with the DC before any rate war. Risk: empty-mile share may be seasonal — I would confirm two more weeks of GPS data before a permanent exit."
That sequence — decide what data matters, compute cleanly, close in under a minute — is the whole assessment in miniature.
Ned's rule. Treat every early Casey question as a commitment that buys or burns later exhibits. If you cannot name the decision metric in one sentence before you click a dataset, you are guessing.
Scoring: a heuristic you can use in practice
BCG does not publish Casey's rubric. When I score a Casey-style run for practice, I use a simple sheet candidates can reproduce:
| Dimension | Weight (practice only) | Pass signal |
|---|---|---|
| Decision-linked structure / data choice | 25% | Picked the metric that tests the ask |
| Exhibit reading | 20% | Units and period correct before claims |
| Quantitative accuracy | 25% | Right setup; arithmetic within rounding |
| Judgment / next steps | 15% | Prioritized, not laundry-listed |
| Communication (esp. video) | 15% | Answer first; numbers named; one risk |
I do not award "effort points" for long chat answers. Four lines that name the cut, the weekly dollars, and the risk beat a paragraph that restates the prompt. On video, I stop listening when the first ten seconds are throat-clearing.
Integrity is part of the screen. Do not screenshot, do not paste the prompt into an LLM during the live assessment, and do not buy "real Casey questions." Use original practice only.
Mistakes that are Casey-specific
- Preparing only live candidate-led cases. Live casing trains you to ask for exhibits. Casey often chooses for you or forces a one-shot pick. You need chatbot reps where you cannot negotiate the next chart.
- Writing novels into the chat. The bot wants a pick, a number, or a short rationale. Verbosity burns clock and rarely helps.
- Ignoring the invite's calculator and camera rules. If the email forbids external tools, do not open one. If it wants camera on, treat the minute as a client update, not a casual Zoom.
- Confusing Casey with CCA. CCA is a separate assessment BCG describes as multidimensional insight beyond the resume. Casey is a case. Wrong prep wastes the week you have.
- Video that recaps instead of recommends. Recommendation, two case numbers, one risk. Sit still. Look at the lens.
Drill this before your invite
One concrete assignment, not a vague "practice more."
- Read your invite (or a sample office page) and write down duration, video yes/no, and calculator rule on a sticky note.
- Run CoachNed's Casey-style case at /casey/urbanbrew-coffee-margin-case under a hard timer matching your invite. Debrief only the first data choice and any blank items.
- Record a 60-second recommendation of that case on your phone. Watch it once. Cut anything before the recommendation sentence.
- Spend 20 minutes on chart and percent drills — Casey is exhibit math under a clock:
Format overview and free practice also live at /bcg-casey-chatbot. Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month. After Casey, keep drilling candidate-led BCG cases and fit — the chatbot is a gate.
CoachNed is independent practice material and is not affiliated with Boston Consulting Group.
Frequently asked questions
How long is the BCG Casey online case?
BCG publishes roughly 30 minutes for the Online Case Experience on its Mexico and Panama careers pages, and about 35 minutes for the U.S. Associate online case. Your invitation controls the exact clock.
Does Casey always end with a video recommendation?
Not on every public office page. Panama explicitly requires a 1-minute video recommendation; Mexico's FAQ emphasizes the chatbot case without spelling out video on that page. Follow your invite.
Is Casey the same as the BCG CCA?
No. Casey (OCE) is a chatbot business case. The Consulting Career Assessment is a separate screen BCG describes as multidimensional insight beyond the resume. Some processes use one, some both.
Can I go back and change answers on Casey?
BCG's public careers FAQs do not state a universal back-navigation rule. Candidate guides and invites often say submissions are final. Assume you cannot rewind unless your materials say otherwise.
How should I prepare for Casey if I already case well live?
Keep the business judgment, but add timed chatbot reps, dataset-selection drills, and one-minute video closes. Live cases forgive fishing for data; Casey often does not.
Sources
- BCG Mexico careers — Online Case Experience FAQ — Casey named as OCE chatbot; ~30 minutes; English; prepare like a case interview. Checked 2026-09-24.
- BCG Panama careers — Online Case — 30 minutes; quant, data, calculations, business logic; 1-minute video recommendation. Checked 2026-09-24.
- BCG U.S. Associate job opportunities — select candidates; online case ~35 minutes; prior takers skipped. Checked 2026-09-24.
- BCG case interview preparation — official stance that reasoning quality matters over a single right answer. Checked 2026-09-24.
- BCG alumni interview: Eugene Goh on Casey — design intent: structured questions, free text, video; skills assessed. Checked 2026-09-24.
- OneDay@BCG virtual experience — official free job simulations for case-style practice outside the live assessment. Checked 2026-09-24.