Capgemini Invent Case Interview Examples: 7 Digital and Operations Cases Solved (2026)

A Capgemini Invent case interview is a guided business problem of about 45 minutes in which the interviewer plays the client and you structure, analyze and recommend, according to Capgemini Invent's own students page. Some countries add a group exercise, and Capgemini Invent publishes only one sample problem. This page gives you seven original practice cases, six individual and one group, with every number worked, built on the customer, industry, enterprise and sustainability work Capgemini Invent lists on its own site.

The Capgemini Invent case interview guide owns the process, the group case etiquette and the behavioral questions, and the Capgemini Invent application guide in German covers the German process. This page is the practice, and it does not repeat them.

Capgemini Invent-style practice cases at a glance

How a Capgemini Invent case works: the interviewer presents a client scenario, you decide how to approach it, and the interview is guided and about 45 minutes long. Capgemini Invent does not publish more than that, so the facts below are labelled by source.

• Official, format: a real-time scenario between a client (the interviewer) and a consultant (you), guided, approximately 45 minutes (Capgemini Invent students page).
• Official, what is assessed: structured problem solving, quantitative and qualitative analysis, communication, thinking and presenting under pressure, and thinking holistically, including identifying risks and mitigation tactics, with the note that the list is not exhaustive.
• Reported, other stages: a recruiter screen of about 30 minutes and, in some offices, a group case (CaseBasix; Hacking the Case Interview).
• Official, areas of work: Strategy and transformation, Creative and design, Data, Engineering and Leading-edge tech on the students page, and Customer First by frog, Intelligent Industry and Enterprise Transformation as capability teams on the global careers page.

Start free now: EV Charging Hub Profitability is a profitability case on a Spanish network of EV charging hubs with cafes attached, the nearest library match to the automotive and EV sample Capgemini Invent publishes: 8 questions, 3 with math, 3 exhibits, and a median of 17 minutes across 45 completed voice runs. An AI interviewer runs it out loud, shows each exhibit when you reach it and scores the debrief. Your first full case is free when you create an account, and the worked excerpt further down runs its first exhibits. A free account also gets one full run in every simulator inside the app: one complete run of each of our four McKinsey Solve practice formats (Red Rock, Sea Wolf, SFL Classic and SFL Island), a full Casey-style case, a full CCA-style diagnostic and a complete Bain SOVA-style diagnostic.

Case (style, industry)	Quant difficulty	Qualitative difficulty	The math you run	Median time
EV Charging Hub Profitability (McKinsey, energy and retail). 8 (3 with math), 3 exhibits.energy and retailQuant difficulty mediumQualitative difficulty medium17 min· 45 voice runsTry free	medium	medium	Segment profit bridge, percentage change	17 min45 voice runs17 min, 45 voice runs	Try free
Electro Chargers City Entry (BCG, EV infrastructure). 7 (2 with math), 3 exhibits.EV infrastructureQuant difficulty mediumQualitative difficulty medium23 min· 24 voice runsTry free	medium	medium	Vehicle base projection, station investment	23 min24 voice runs23 min, 24 voice runs	Try free
Cloud Computing Capacity Expansion (BCG, cloud infrastructure). 9 (4 with math), 3 exhibits.cloud infrastructureQuant difficulty hardQualitative difficulty hard30 min· 13 voice runsTry free	hard	hard	Price read, profit uplift, payback, payoffs	30 min13 voice runs30 min, 13 voice runs	Try free
TravelLine Call Center Optimization (BCG, travel customer service). 8 (3 with math), 2 exhibits.travel customer serviceQuant difficulty hardQualitative difficulty mediumFewer than 10 runsTry free	hard	medium	Call volume, agents needed, capacity gain	Fewer than 10 runsFewer than 10 runs	Try free
MetroFresh Distribution Center Operations Turnaround (McKinsey, grocery distribution). 8 (4 with math), 4 exhibits.grocery distributionQuant difficulty easyQualitative difficulty hard23 min· 17 voice runsTry free	easy	hard	Picking capacity gap, EBITDA initiatives	23 min17 voice runs23 min, 17 voice runs	Try free

One is adapted from a consulting club casebook (Wharton); the rest were written for Road to Offer. The library has no case in Capgemini Invent's own style, so these are matched by case type to the automotive, technology, customer service and operations work Capgemini Invent lists, and none is Capgemini Invent's own material. More are in the case library, 92 cases in all. Every debrief scores seven categories: structure (25% of the score), quantitative analysis (20%), business judgment (15%), and problem understanding, exhibit reading, communication and synthesis (10% each), with feedback on every question. Median times come from completed runs between June 25 and September 22, 2026, and count runs, not people.

What does a Capgemini Invent case interview look like?

It is a guided, real-time client scenario of approximately 45 minutes in which the interviewer plays the client and you decide how to approach it. That is how the Capgemini Invent students page describes it, and it adds that the client prompts you and provides the information needed to solve the issue.

• Official, the setup: the case is presented as a real-time scenario between a client and a consultant. You then decide the approach, so your structure is the first move even though the interview is called guided.
• Official, the tips Capgemini Invent gives: take your time composing your thoughts, ask for clarification, take notes, summarize specific issues and findings, state your recommendations, and outline next steps and expected results. The cases below end the same way.
• Reported, what comes around it: Hacking the Case Interview describes a recruiter screen, the case, a group case or assessment centre that depends on the office, and a closing interview with a Principal. CaseBasix describes a 30 minute recruiter screen, a 30 to 45 minute case and a final round of several interviews.
• What to do with that: the 45 minutes are one case, so practice one case end to end in that time, then add the group format only if your recruiter confirms it.

Now run one out loud the way a Capgemini Invent interviewer expects: a technology case, asked before any number is on the table.

Run it as a client conversation: restate the decision, ask what success means, and only then structure. The rest of this page gives you the written reps to do before and after.

Does Capgemini Invent publish practice cases?

Yes, one: a sample problem statement on its students and graduates page, with no answer attached. The statement reads as follows in substance: an automotive client wants to understand what it should do as a business with respect to electric and self-driving vehicles, and wants a recommendation on its future in these markets, covering four questions.

1. What is the market size opportunity in North America?
2. How might it position itself against others in the industry?
3. What value will it bring to customers?
4. Who should it target as customers, and how would you find that out?

Read those four questions as a structure the interviewer has already handed you. The first is a sizing, the second a competitive choice, the third a value question you can quantify and the fourth a segmentation and research question. A strong answer takes all four in order, puts a number on the first and the third, and ends with a risk, which is the fifth published skill. Case 1 does exactly that.

The same page lists a few reference links for case practice and says it strongly encourages candidates to practice case interviewing beforehand. No other Capgemini Invent page we checked contains a case prompt, an answer key or a scoring sheet, so everything else on this page is original practice material and says so.

Get your Capgemini Invent prep plan

Build my prep plan

1

Learn the format

15 steps

• Learn how to caseGuided learning · 35 min
• The Case Interview MethodLesson · 8 min
• Profitability & Turnaround: AfrihealSkill drill · 5 min
• Profitability & Turnaround: Berty BankSkill drill · 5 min

Is the Capgemini case interview the same across the whole group?

Not necessarily, and Capgemini Invent's pages are the only ones that describe a case. Capgemini is a group, and Capgemini Invent is its innovation and transformation brand, so a candidate who searches "Capgemini case interview" lands on advice written for different units.

• Official, group recruitment page (US): interviews by phone, video or in person, a possible mix of behavioral, technical and cultural fit interviews and assessments, and the note that the process varies by country. It does not mention a case.
• Official, Capgemini Invent global page: video, phone or in-person interviews covering behavioral, technical and cultural fit, and an assessment centre with group exercises that depends on the country and the role. It does not mention a case interview either, and the page says the process is entirely dependent on the country and role.
• Official, Capgemini Invent students page (US): the guided 45 minute case, with the five skills above.
• Reported, prep sites: PrepLounge's Capgemini guide is written for the group, names behavioral and case interviews without a format and does not mention a group case or Invent. CaseBasix describes three stages with a group case of 4 to 6 candidates and does not mention Invent either. Hacking the Case Interview separates Invent from the technical roles of the wider group and says Invent runs the consulting style interview.

What this means for you: if your posting is for Capgemini Invent strategy, transformation or data consulting, prepare for the guided case on this page. If it is for another unit, ask the recruiter which interviews and assessments to expect, because the official group page lists technical assessments and does not promise a case. The Accenture case interview guide covers the nearest technology-led peer, and types of consulting firms shows where Invent sits.

Which case types do these seven cases cover?

Each case sits in an area Capgemini Invent lists on its own pages, so you can start with the practice you applied to. Capgemini Invent does not publish a case type list, so the match is by area of work, not a claimed interview split.

The library cases in the table above are the ones to run out loud next, in this order for an Invent loop: EV Charging Hub Profitability for the exhibits, Electro Chargers City Entry for the sizing, Cloud Computing Capacity Expansion for technology, TravelLine Call Center Optimization for customer service, and MetroFresh for operations. All numbers in Cases 1 to 7 are assumptions made up for practice. State your own assumptions out loud in an interview, exactly as the cases do.

The loop to run on every case

Read only the prompt, write your structure and your questions, then open the solution. The cases below use the same five parts so you can compare each to your own:

1. Prompt: the client, the decision and the constraint.
2. Clarifying questions: what you would ask, and what the interviewer would plausibly say.
3. Structure: two to four branches you can test, not a template.
4. The math: every step substituted, with units.
5. Recommendation: a decision, one number and one risk, the way Capgemini Invent's fifth skill asks.

If you have never run a case, the digital transformation case interview, operations case interview and technology case interview pages teach the frameworks. This page is for reps.

Worked excerpt: an EV charging case, exhibit by exhibit

Capgemini Invent's own sample is about electric vehicles, so here is the first half of the library's EV Charging Hub Profitability, worked from its three exhibits. It is a profitability diagnosis, the other half of what an EV prompt can turn into once the sizing is done. Cover each answer and work the step before you read it.

The prompt. A client runs electric vehicle charging hubs across Spain, each with a cafe and retail space. Profit fell between 2022 and 2023. What caused it, and what should the client do?

A structure that works. Profit is revenue minus cost in each of the two businesses, charging and cafe and retail. Split by business line first, because the two can move in opposite directions, and check electricity prices as the obvious outside cause.

Exhibit 1, electricity prices. Spain's price is €0.13 per kWh in 2022 and €0.13 in 2023, a 0% change. A price that did not move cannot explain a profit fall between those years.

Exhibit 2, profit by business line (€ millions). Profit is revenue minus costs.

Exhibit 3, visits (thousands). Visits that can earn retail revenue are the visits to both businesses plus cafe and retail only visits: 440 + 260 = 700 in 2022 and 250 + 180 = 430 in 2023. That is a 38.6% fall, close to the 40% fall in retail revenue from €120M to €72M.

What a strong answer sounds like. "Profit is down €29M, or 15.6%, and all of it is in cafe and retail, which lost €48M while charging gained €19M. Electricity prices were flat. Visits that include the cafe fell 38.6%, in line with the retail revenue drop, so I would find out why charging customers stopped buying and pilot charging-to-retail offers before scaling. The risk is that I am reading a visit mix change as a cause, so I would check it against the customer data."

Run the EV Charging Hub profitability case

EnergyMedium

Practice this case free

Case 1: Electric and self-driving vehicles for an automotive client

Prompt. This is the problem statement Capgemini Invent publishes, and the numbers below are ours. An automotive client wants a recommendation on its future in electric and self-driving vehicles: the North American market size, how to position, what value to bring to customers and whom to target. It is an original practice solution to a published prompt, not a Capgemini Invent answer.

Clarifying questions. Which year, 2030? Light vehicles only? Does electric mean battery electric only? Does the client sell today, and at what share? Is self-driving a feature on a car or a separate service? You get: 2030, light vehicles, battery electric, the client sells gasoline cars today with a 10% share of new vehicle sales, and self-driving means a paid driver-assistance option on new cars. All figures are practice assumptions.

Structure. The four published questions become the four branches:

The math.

1. Electric units in 2030: 16M new light vehicles x 25% electric share = 4.0M.
2. Electric revenue pool: 4.0M x $50,000 = $200B. At the client's 10% share that is 400,000 units and $20B.
3. Self-driving option pool: 16M x 10% take rate = 1.6M cars, x $5,000 = $8B, so the option is 4% of the electric pool.
4. Annual running cost of a gasoline car: 12,000 miles / 30 miles per gallon = 400 gallons, x $3.60 = $1,440.
5. Annual running cost of an electric car at home: 12,000 x 0.30 kWh = 3,600 kWh, x $0.16 = $576.
6. Saving: 1,440 - 576 = $864 a year. With a $6,000 price premium, payback is 6,000 / 864 = 6.9 years.

Recommendation. Go for the $20B electric pool first, with a premium crossover that justifies its price on software and range, and treat the $8B driver-assistance option as a paid software feature added later. Fuel savings alone do not pay back the premium for 6.9 years, so the value has to come from something else (performance, features or a lower premium). The risk is that the 25% electric share drives everything: at 15% the pool falls to 2.4M units and $120B, and the client's 10% share becomes $12B, so test the share with an owner survey before committing plants.

Case 2: A telecom operator shifting care to digital

Prompt. A mobile operator with 8 million customers handles 30 million customer contacts a year. The CEO wants to know whether moving half of them to the app and chat is worth it. This sits in the customer experience work Capgemini Invent lists under Customer First by frog. All numbers are practice assumptions.

Clarifying questions. What does a contact cost by channel? How many contacts are digital today? Do digital contacts always resolve? What does the program cost? You get: 24 million of the 30 million contacts are calls at $6.00 each, 6 million are digital at $0.80 each, the target is 50% digital, 25% of the contacts moved to digital fail and end in a call anyway, and the build costs $20M once plus $5M a year to run.

Structure.

• Volume: contacts today by channel and by reason.
• Unit cost: call, digital and failed digital that falls back to a call.
• Levers: how many contacts can move, how many stay moved.
• Customer effect: satisfaction and churn for the customers who are moved.

The math.

1. Today's cost: 24M x $6.00 = $144M, plus 6M x $0.80 = $4.8M, so $148.8M.
2. Shift: 50% of 30M is 15M digital, which is 15M - 6M = 9M more.
3. Savings if every move sticks: 9M x ($6.00 - $0.80) = 9M x $5.20 = $46.8M.
4. Fallback calls: 25% x 9M = 2.25M calls x $6.00 = $13.5M, so savings fall to 46.8 - 13.5 = $33.3M.
5. Net of run cost: 33.3 - 5 = $28.3M a year, and payback on the $20M build is 20 / 28.3 = 0.71 years, about 8.5 months.
6. Break-even failure rate: the program only stops covering its $5M run cost when 9M x (5.20 - 6.00 x f) = $5M, so f = 77%.

Recommendation. Do it, because the savings are about $33M a year at a 25% failure rate and the program still pays at failure rates three times higher. The number to manage is the failure rate, not the adoption target: every 10 points of failure on the 9M moved contacts costs about $5.4M a year (9M x 10% x $6.00). The risk is churn among customers who are pushed to digital and fail, so measure satisfaction for moved customers for one quarter and keep a call option for them.

Case 3: A smart plant and the cost of unplanned downtime

Prompt. A manufacturer of industrial pumps runs four plants and wants to put predictive maintenance sensors on critical machines. Should it roll the program out to all four? This is the Intelligent Industry kind of problem, where technology meets the factory floor. All numbers are practice assumptions.

Clarifying questions. How many hours do the plants schedule, and how much of that is unplanned downtime? What does an hour of output earn? Can the extra output be sold? What does the program cost? You get: each plant schedules 6,000 hours a year and produces 2,000 units an hour at a contribution of $18 a unit, unplanned downtime is 8% of scheduled hours, sensors cut it by 30%, the program costs $4.5M per plant once and $1.0M a year to run, and the market can absorb only 60% of any extra output.

Structure.

• Prize: hours lost, value per hour, share the program can recover.
• Constraint: whether recovered output sells.
• Cost: one-time and running cost per plant.
• Risk: whether the 30% holds outside the supplier's own pilots.

The math.

1. Downtime per plant: 6,000 x 8% = 480 hours.
2. Value per hour: 2,000 x $18 = $36,000, so lost contribution is 480 x 36,000 = $17.3M per plant and $69.1M across four.
3. Recoverable: 30% x 480 = 144 hours, x $36,000 = $5.18M per plant.
4. If every hour sells: net of the $1.0M run cost is $4.18M, and payback is 4.5 / 4.18 = 1.1 years.
5. With only 60% sold: 0.6 x 5.18 = $3.11M, net $2.11M, and payback is 4.5 / 2.11 = 2.1 years.
6. Break-even for a three year payback: net benefit needed is 4.5 / 3 + 1.0 = $2.5M, and 2.5M / (36,000 x 0.6) = 116 hours, which is 24% of the 480.

Recommendation. Pilot at one plant, the one with the most downtime, before committing $18M across four. The case for rolling out depends on two numbers the interviewer has not proved: whether the 30% reduction holds, and whether 60% of the extra output sells. A measured reduction of 24% or more still gives a three year payback, so set that as the rollout trigger. The risk is the sell-through assumption, because if the extra pumps cannot be sold, the sensors only make the plant look more efficient.

Practice an operations case for Capgemini Invent

ManufacturingMedium

Start the operations case

Case 4: Replacing an ERP system, big bang or phased

Prompt. A retailer runs a 20 year old ERP platform and has two replacement options. The CFO asks which to choose. Enterprise Transformation is a Capgemini Invent capability team, and this case is not about any vendor. All numbers are practice assumptions.

Clarifying questions. What are the options, what do they cost and when do benefits start? What happens if the program fails? Over what horizon do we judge it? You get: Option A, big bang, costs $90M and delivers all benefits from year 3. Option B, phased, costs $105M, delivers 40% of benefits in year 3 and all from year 4. Full benefits are $30M a year ($8M lower run cost and $22M of process savings). A fails with probability 20%, costing an extra $60M and a year of delay. B fails with probability 5%, costing an extra $20M and six months of delay. Horizon is 7 years, undiscounted.

Structure.

• Benefits: run cost and process savings, and when each starts.
• Costs: one-time cost, by year.
• Risk: probability and cost of failure for each option.
• Do nothing: what the old platform costs the client if it waits.

The math.

1. A, benefits: years 3 to 7 are 5 years x $30M = $150M. Net of the $90M cost: $60M.
2. A, failure: a delay of one year costs $30M of benefits, so a failure costs 60 + 30 = $90M. Expected: 20% x 90 = $18M, leaving an expected net of $42M.
3. B, benefits: year 3 gives 40% x 30 = $12M and years 4 to 7 give 4 x 30 = $120M, so $132M. Net of $105M: $27M.
4. B, failure: six months of delay costs $15M, so a failure costs 20 + 15 = $35M. Expected: 5% x 35 = $1.75M, leaving an expected net of $25.25M.
5. A beats B by about $17M of expected value. The failure probability at which they tie is (60 - 25.25) / 90 = 39%.
6. Doing nothing leaves $30M a year of benefits unclaimed, so over the five full years it costs $150M against Option A.

Recommendation. Choose the big bang if the client believes its own failure risk is below about 39%, since it is worth about $17M more in expectation. The risk to name is that A's bad outcome is much worse: if A fails, its net is 60 - 90 = -$30M, against -$8M for B. If the client has a record of large programs that overran, take B. Say that the horizon is undiscounted and that discounting hurts B's later benefits more, then give the recommendation on the numbers, not on a preference for phasing.

Case 5: Decarbonizing a delivery fleet

Prompt. A logistics company with 10,000 delivery vans emits 150,000 tonnes of CO2 a year. The board wants a 40% cut in five years. What should it do, and what will it cost? Sustainability is a service line on Capgemini's own pages, and the ESG and sustainability case interview page covers the framework. All numbers are practice assumptions.

Clarifying questions. Which emissions count, tailpipe only? What levers are available, and what do they save and cost? Can they be combined? You get: tailpipe only, so 15 tonnes per van a year. Three levers: route optimization software cuts 5% of total emissions and saves $40 per tonne; replacing 2,000 urban vans with electric vans cuts 80% on each, at a net cost of $1,500 per van a year after fuel savings; renewable diesel on the other vans cuts 75% on each van it fuels, at $210 per tonne, with enough supply for 5,000 vans.

Structure.

• Target: 40% of 150,000 tonnes.
• Levers: size, cost per tonne and constraints for each.
• Order: cheapest tonne first, until the target is met.
• Overlap: which levers cannot be added without a haircut.

The math.

1. Target: 0.40 x 150,000 = 60,000 tonnes.
2. Route optimization: 5% x 150,000 = 7,500 tonnes at -$40 per tonne, which is -$0.3M a year (a saving).
3. Electric vans: 2,000 x (80% x 15) = 2,000 x 12 = 24,000 tonnes. Cost per tonne is 1,500 / 12 = $125, so $3.0M a year.
4. Running total: 7,500 + 24,000 = 31,500 tonnes, which is 21% of the baseline.
5. Remaining: 60,000 - 31,500 = 28,500 tonnes from renewable diesel, which needs about 28,500 / (15 x 0.75) = 2,533 vans fuelled, well within the 5,000 available. Cost: 28,500 x $210 = $5.99M a year.
6. Total cost: -0.3 + 3.0 + 5.99 = $8.7M a year, about $145 per tonne abated.

Recommendation. Do all three in order of cost per tonne: route software now because it pays for itself, electric vans in the dense urban routes, and renewable diesel for the remainder. The plan costs about $8.7M a year to remove 60,000 tonnes. The risk is the price of renewable diesel, which carries nearly half of the abatement: at $273 per tonne (30% higher) the plan costs about $10.5M a year, so secure supply and price before the board commits to the target. The lever sizes treat each lever as independent, which slightly overstates the total, so haircut route optimization on the electrified vans.

Practice a sustainability case for Capgemini Invent

TechnologyHard

Start the sustainability case

Case 6: An insurer using a model to catch fraud

Prompt. An insurer handles 600,000 claims a year and suspects it pays out on fraud. A data science team has built a model that scores every claim. Should the insurer use it, and how many claims should it flag? This is the Data path on the Capgemini Invent students page, with a business decision attached. All numbers are practice assumptions.

Clarifying questions. What share of claims are fraudulent, and what does a fraudulent claim cost? What catches fraud today? What does an investigation cost? You get: 2% of claims are fraudulent at an average payout of $8,000, manual review catches 30% of them and costs $4M a year, an investigation costs $150, the model costs $6M a year, and flagging the top 10% of claims catches 60% of fraud while the top 5% catches 45%.

Structure.

• Prize: fraud in the book, and what is caught today.
• Model: how much more fraud it catches at each flag threshold.
• Cost: investigations, the model and the manual review it replaces.
• Customer effect: honest customers who are investigated.

The math.

1. Fraud: 2% x 600,000 = 12,000 claims, x $8,000 = $96M a year.
2. Caught today: 30% x 12,000 = 3,600, x $8,000 = $28.8M.
3. Top 10% flagged: 60,000 claims, catching 60% x 12,000 = 7,200, worth $57.6M. Investigations cost 60,000 x $150 = $9M.
4. Net gain at 10%: (57.6 - 28.8) - (9 + 6 - 4) = 28.8 - 11 = $17.8M a year.
5. Top 5% flagged: 30,000 claims, catching 45% x 12,000 = 5,400, worth $43.2M. Net gain: (43.2 - 28.8) - (4.5 + 6 - 4) = 14.4 - 6.5 = $7.9M a year.
6. The extra 30,000 flags between 5% and 10% catch 1,800 more cases, a 6% hit rate. An investigation breaks even at 150 / 8,000 = 1.9%, so the extra flags are worth it.
7. Precision at 10%: 7,200 / 60,000 = 12%, so 52,800 honest claims are investigated.

Recommendation. Use the model and flag the top 10%, for about $17.8M a year of net gain. Keep flagging while the marginal hit rate stays above 1.9%, and ask the data team for the lift curve beyond 10% before going further. The risk is the 88% of flagged claims that are honest, because a slow or intrusive investigation costs customers and complaints that the $150 does not capture, so cap the investigation time and track the repeat claim rate of investigated honest customers.

Case 7: A worked group case on digital ticketing

Prompt. The group case is the format Capgemini Invent's global page mentions only as a possible assessment centre exercise and prep sites describe in more detail. CaseBasix reports 4 to 6 candidates, often 10 to 15 pages of material, about 10 minutes of individual reading and 30 minutes of group discussion, with a short presentation to the interviewers afterward. This case is original and sized to that shape: a regional transit authority asks your group whether to introduce digital ticketing. Public sector is an industry Capgemini lists. All numbers are practice assumptions, and the pack is condensed to the five pages that matter.

The pack.

Your 10 minutes of reading. Skim all five pages once, then work the two option pages and the survey, because they decide the answer. Write down one idea you will offer in the first two minutes: that the group splits the problem into money (what each option earns), riders (who it leaves behind) and delivery (how fast it can be live). Note the one number you will own, such as the payback of each option.

The 30 minutes of discussion.

1. Minutes 0 to 3, structure. Propose the three branches and ask the group to agree or change them. Offer to keep time.
2. Minutes 3 to 12, split the work. One pair takes Option A, another takes Option B, one person takes the rider survey. Each reports a number back, not a summary.
3. Minutes 12 to 24, compare and decide. Put the two options side by side, test the weakest assumption and bring in anyone who has been quiet.
4. Minutes 24 to 30, synthesize. One person states the recommendation, the number behind it and the risk, and the group checks it in one sentence each.

The math.

1. Cash revenue: 40% x $192M = $76.8M. Handling cost: 7% x 76.8 = $5.38M a year.
2. Option A benefits: evasion 50% x 11 = 5.50; cash handling 60% x 5.38 = 3.23; ridership 3% x 80M = 2.4M trips x $2.40 = 5.76; upkeep 4.00. Total $18.49M a year, less $8M to run, so $10.49M net.
3. Option A payback: 60 / 10.49 = 5.7 years.
4. Option B benefits: evasion 30% x 11 = 3.30; cash handling 40% x 5.38 = 2.15; ridership 1% x 80M = 0.8M trips x $2.40 = 1.92; upkeep 1.50. Total $8.87M, less $3M to run, so $5.87M net.
5. Option B payback: 25 / 5.87 = 4.3 years.
6. Riders left behind by Option B: 22% of riders have no bank card, and they still need a way to pay.

The group's recommendation. Start with Option B, which pays back in 4.3 years against 5.7 and can be live sooner, and add cash top-up points (retail agents and ticket machines) so the 22% of riders without a bank card are not left behind. Decide on Option A's account system once B's first year of results is in. The risk is equity: if cash riders cannot top up easily, ridership among them falls and the authority loses the revenue it set out to protect. Present the recommendation to the interviewers in three sentences: the choice, the payback numbers, and the risk and mitigation.

What the observers watch. In the reports we read, interviewers watch how you work in the group, not only the answer. Open with structure, bring in quiet voices, keep time and synthesize at the end, and do not try to talk the most. The group case interview guide and the Capgemini Invent guide cover those behaviors in detail.

How do you turn these cases into a prep plan?

Run each case once cold, then repeat only the step that broke. Cases 1 and 5 test sizing and ranking, Cases 2, 3 and 4 test unit economics and payback, Case 6 tests reading data, and Case 7 tests the same skills with other people in the room. The sequence below uses all seven in about two weeks.

If the math is what slows you down, the case interview math practice page gives you the formulas, and the cost-benefit analysis framework page covers the payback logic in Cases 2 to 4. For the setting, automotive case interviews teach the industry, and the case interview examples hub collects worked cases from other firms. For the fit side of the interview, use the behavioral interview for consulting guide, and for the Capgemini Invent process, the Capgemini Invent interview prep page.

Turn this guide into a Capgemini Invent plan

Build my prep plan

1

Learn the format

15 steps

• Learn how to caseGuided learning · 35 min
• The Case Interview MethodLesson · 8 min
• Profitability & Turnaround: AfrihealSkill drill · 5 min
• Profitability & Turnaround: Berty BankSkill drill · 5 min

Sources and further reading (checked November 12, 2026)

• Capgemini Invent: students and graduates (US) (what a case interview is, guided and approximately 45 minutes, five assessed skills, the sample automotive problem statement, tips, career paths)
• Capgemini Invent: careers (global) (recruitment depends on country and role, assessment centre with group exercises possible, capability teams, industries, no case interview described)
• Capgemini Invent: Bewerbungsprozess (Germany) (virtual interview with a possible case study, in-person round ending in a subject-related case study, no group exercise mentioned)
• Capgemini: recruitment process (US, group) (phone, video or in-person interviews, behavioral, technical and cultural fit, no case named)
• Capgemini: interview tips (US, group) (general interview preparation, no case described; the site navigation lists Sustainable business under services and Public sector and Automotive under industries)
• Capgemini case interview guide, CaseBasix (reported three stages, group case of 4 to 6 candidates, 10 to 15 pages, 10 minutes of reading and 30 of discussion, example prompts; a prep site, not Capgemini, and it does not mention Invent)
• Capgemini Invent interview guide, Hacking the Case Interview (reported four stages, group case in some European offices, Invent versus the technical roles of the wider group; last updated June 18, 2026; a prep site, not Capgemini)
• Capgemini interview guide, PrepLounge (group-level overview, behavioral and case interviews without a format, no group case or Invent distinction; updated December 15, 2025; a prep site, not Capgemini)