BCG X Case Interview: The Coding Test BCG Publishes
BCG publishes a BCG X selection process in one country only: an online coding test, then technical interviews with case studies and live coding. Read closely.
- One BCG page publishes a BCG X selection process, and it is the Australian associate page. It names an online coding test before any interview.
- First round is virtual technical interviews with case studies and live coding in the same sentence. The final round adds case-based and technical interviews.
- A successful BCG X applicant is offered a seat on BCG's Consulting team in Data Science or AI Engineering, in the firm's own words.
- BCG's global interview process page names four steps and never mentions BCG X. Neither page cancels the other.
- No page states the coding test length, its platform, its language, or how many interviews a round holds.
BCG X is inside BCG, and its hiring is partly inside BCG's too. One page breaks the pattern. The Australian associate page for BCG X publishes a selection process of its own, and the first thing in it is an online coding test that no BCG consulting page mentions.
That page is the reason this guide exists. Everything below is either read on a BCG page and quoted, or named as missing. Sources checked 29 September 2026.
What BCG X is, and what a build question rewards
The BCG X careers page opens with a number. It says BCG X "brings together over 3,000 of the best tech, design, and entrepreneurial minds to create businesses, products, and services that will change the world." The job searches it links point at four families: engineering, data science, product, and design.
The titles in its own testimonials are worth reading, because they are the vocabulary your recruiter will use. The page quotes a "Senior Forward Deployed AI Scientist" in London, a "Growth Principal" in London, and a "Principal Forward Deployed AI Engineer" in Boston. Forward deployed is the phrase to notice. It describes someone who builds at the client rather than in a lab.
That shapes what a good answer sounds like here. A generalist case rewards a structure that covers the profit equation. A build question rewards something narrower: what decision does this model change, what data would you need, what would you ship first, and what happens when the data turns out to be worse than promised. A candidate who reaches for a model architecture before naming the decision has answered a question nobody asked.
The page publishes nothing about hiring. No steps, no assessment, no entry title, no calendar. For that you have to go to a country page, and only one of them says anything.
The one BCG X process, ending in a consulting offer
The BCG X associate recruitment page for Australia carries a section headed Selection Process. It is four steps, and here is what each one says.
| Step | What the page says |
|---|---|
| Online application and coding assessment | Eligible candidates "receive an invitation to complete an online coding test" after the team reviews the resume |
| Application review and shortlisting | Resume and coding assessment are reviewed together before shortlisting |
| Interviews, first round | "Virtual technical interviews, including case studies and live coding challenges" |
| Interviews, final round | "Additional case-based and technical interviews to assess problem-solving and analytical thinking" |
| Offers | An offer "to join BCG's Consulting team in Data Science or AI Engineering" |
Three lines in that table change how you prepare.
The coding test comes before a human reads anything beyond your resume. Preparation that starts with case frameworks starts at step three of a four-step process.
Case studies and live coding sit in one sentence, in the same round. The page never says whether they are two exercises or one exercise with two halves. Prepare for both and let the invitation settle it.
The offer lands on BCG's Consulting team. The firm's own wording puts a BCG X associate inside the consulting business, which is the clearest published answer to whether this is a separate employer.
The eligibility rules on that page are unusually hard-edged. It is aimed at people in their final year or within two years of finishing, in fields it lists as "computer science, artificial intelligence, data science, machine learning, engineering, mathematics, statistics, physics, or related technical fields". It asks for a CV, cover letter and transcripts uploaded as one document. It says the firm "can only consider candidates who currently hold Australian citizenship or are Australian permanent residents", and it tells you to rank Melbourne, Perth or Sydney first.
What applies when you are not in Australia
BCG's interview process page publishes four steps for its consulting hiring: application, skills interview, case interview, team interview. It says that "for client facing roles, we'll assess your problem-solving and analytical skills in a case interview." It names five qualities its interviewers look for: integrity, intellectual curiosity, creative thinking, collaborative mindset, drive.
The word BCG X appears nowhere on it. That cuts both ways, and the honest reading is the boring one. BCG has not published that its consulting sequence covers BCG X, and it has not published that BCG X is exempt.
So there are two documents, and the one that governs you is the invitation email. Until it arrives, prepare the union: a coding exercise you can talk through, a business case you can structure aloud, and a story about something you built.
The case interview preparation page is worth reading whichever round you land in, because it is the only place BCG says what it wants from the conversation. It writes that "there isn't always a single 'right' answer" and that "what matters most is how you approach the problem and the quality of your reasoning." It then names twelve behaviors, six to do and six to avoid, including clarify assumptions, show your thinking, and stay silent as a failure mode. Our BCG guide works through that page in full.
Where BCG's own practice cases live, and what they are not
A reasonable next question is whether BCG hands out anything to practice on. It does, from its own careers pages, and none of it is labeled for BCG X.
The announcement sits on the case interview preparation page under Put Your Skills to the Test, which offers "a free job simulation that provides insights into the challenges we solve". The scale is on the virtual experience page: "With 13 unique modules to choose from" and "Enroll today and join the 700,000+ participants". BCG writes the modules. A third party hosts the files, and they open without an account.
Three published clauses decide what that material is worth to you here.
- The pages carry "No grades · No assessments". So nothing in a module reports back, and nothing in it is a screen.
- BCG never calls a module a case interview, and never says one resembles the exercise in its selection process.
- The firm attaches a restriction in writing: the experience "is not an employment relationship with BCG", and its policy is that you do not list it as employment, work experience or an internship anywhere.
None of the thirteen is flagged as technical, data science or engineering work, which is the thing a BCG X candidate would want. So treat the catalog as business reading rather than as rehearsal for the coding test, for which BCG publishes nothing at all. For an exercise with the arithmetic and a published answer beside it, the inventory of what five firms serve themselves is on case interview cases, and the BCG line there is counted separately for exactly these reasons.
A deployment case, worked in dollars
Here is the kind of arithmetic a BCG X case rewards. A grocery chain runs 900 stores. Fresh produce thrown away averages $1,800 per store per week. The client wants to know whether a demand forecasting model is worth building.
- Size the pool before touching the model. Per store, $1,800 a week is $93,600 a year. Across 900 stores that is $84.24m of waste a year. Nothing about the model matters until that number is on the table.
- Put a number on the lift, and say where it comes from. Suppose the model cuts waste by 12%. Twelve percent of $84.24m is roughly $10.1m a year. State it as an assumption to be tested rather than as a result.
- Subtract what the thing costs to run. A build at $4.5m and a run cost of $1.2m a year leaves net annual benefit of about $8.9m. Payback on the build is $4.5m divided by $8.9m, so a little over six months.
- Attack your own answer with the rollout. If year one covers 300 stores instead of 900, the benefit is a third of $10.1m, about $3.4m, and the run cost barely moves. Net is around $2.2m, and payback stretches past two years. Same model, same lift, four times the wait.
- Say what decides it. "Build it, on the condition that the first wave covers the 300 highest-waste stores rather than a random third, because payback is driven by waste per store and not by store count."
Read step four twice. The trap in analytics cases is a correct percentage applied to a smaller base than the one you sized. Waste per store is the unit that carries the answer, and stores are the unit that carries the cost.
What we score, and the check good candidates miss
BCG publishes no marking scheme, and no firm does. The weights below are ours, applied by our own grader, each out of a hundred. They are read from the code that scores, and the reasoning behind them sits on how an answer is assessed.
- Answer40
- Working25
- Interpretation20
- Units15
| Structuring | Coverage 45 · Separation 25 · Depth 20 · Concision 10 |
|---|---|
| Idea generation | Breadth 30 · Buckets 30 · Specificity 20 · Prioritisation 20 |
| Market sizing | Scope 25 · Assumption chain 30 · Arithmetic 25 · Sanity check 20 |
| Exhibits | Observation 25 · Implication 25 · So what 30 · Next test 20 |
| Synthesis | Answer first 35 · Evidence 25 · Risk 20 · Next step 20 |
Two of them decide most BCG X answers.
Answer takes 40. A case that ends in a model description with no number has forfeited the largest share. Say the figure, then defend it.
Units take 15, and interpretation takes 20. In the case above, the sentence "12% of waste" means nothing until someone says 12% of $84.24m. Our grader reads the unit and reads whether the sentence after the number says what the number shows. That second check is where good candidates lose marks they never notice.
Work the case math drill until the dimension you lose stops being the same one. A single overall mark tells you an answer was weak without telling you which half to fix.
Where BCG X candidates lose it
- Preparing cases and skipping code. The published Australian sequence puts a coding test first. A perfect case structure cannot save an application that stops at step one.
- Preparing code and skipping cases. The same page names case studies in both interview rounds. Both halves are on the page.
- Leading with the model. Name the decision, the metric and the data first. Architecture is the answer to a later question.
- Sizing the prize after proposing the build. The dollar pool comes first, because it decides whether anything else is worth saying.
- Applying a lift to the wrong base. Twelve percent of a rollout is not twelve percent of an estate.
- Assuming Australia's process is yours. It is the only one BCG publishes for BCG X, and it is explicitly a country page.
- Assuming BCG's four consulting steps are yours. That page never names BCG X either.
Items one through six are avoidable tonight. Item seven is answered by the email that names your format, which is the only document with authority over your process.
McKinsey's analytics arm answers the same question differently: QuantumBlack runs McKinsey's process with one change, and the change is not a coding test.
What BCG does not publish about BCG X
The entry title outside Australia. Every BCG X posting we opened on the careers site had already been filled, so no job description could be read.
The coding test itself. No length, no platform, no language, no number of problems.
Whether the case and the live coding are one exercise. The sentence puts them together and stops.
Whether BCG's own assessments apply. The consulting assessment steps described in our BCG guide are published for consulting hiring. No page says they reach BCG X, and no page says they do not.
The number of interviews per round. The page counts rounds and leaves the rest alone.
Any threshold or scoring scheme. BCG publishes none, for any role.
Everything above about BCG X was read on the BCG page it is credited to, on 29 September 2026. The scoring weights are ours, and the page says so where they appear. If your invitation contradicts anything here, the invitation is right and we would like to hear about it.
Before you go: where do you actually stand?
A short diagnostic: a few questions about where you are, one real case, and a score on the dimensions firms publish. Free, no card.
Sources
Primary sources only: the firm, the test vendor, or the institution concerned. Each carries the date we last read it.
- careers.bcg.com/global/en/locations/australia-and-new-zealand/bcg-x-associate-recruitmentthe four published selection steps, the coding test, the eligibility window, the document bundle, and the wording of the offerchecked 2026-09-29
- careers.bcg.com/global/en/xwhat BCG X says it is, the headcount it publishes, and the job families it nameschecked 2026-09-29
- careers.bcg.com/global/en/interview-processthe four steps BCG publishes for consulting roles and the five qualities its interviewers look forchecked 2026-09-29
- careers.bcg.com/global/en/case-interview-preparationwhat BCG says a case interview is, its twelve named behaviors, and the Put Your Skills to the Test announcement of a free job simulationchecked 2026-09-29
- careers.bcg.com/global/en/virtual-experiencethe count of 13 modules, the 700,000 participant figure, the No grades and No assessments line, and the written policy against listing the experience as work experiencechecked 2026-10-05
Edits the firm and case guides on this site. Reads each claim about a recruiting process back to the firm's own page before it is published, keeps the source ledger dated, and owns the corrections.
How this guide was made
- Written against a published editorial process, not from memory. Every claim about a firm's process is either linked to that firm's own page or absent. We say what we could not verify rather than filling the gap.
- Each page carries a dated source ledger, below. When a firm changes a programme or a deadline, the date tells you how far to trust what you just read.
We look for first-hand candidate accounts for every firm guide. Where we found none we could source, the page says so and carries the firm's own words instead. Nothing here is dressed up as testimony, and if your own experience contradicts what you read, tell us and we will correct it and date the change.
Something wrong or out of date here? Tell us and we will correct it and date the change.
