LEK Case Interview 2026: Seven Steps, Two That Collide

The seven case interview steps, in the firm's own words

The interview preparation page is short, and every line of it is instruction.

Read steps one and two together and they buy you something most candidates give away: time. Paraphrasing the question back is the cheapest error check available, and restating the facts forces you to notice which ones you actually have rather than which ones you assumed. Doing both costs about fifteen seconds.

Step five is the one that names a posture rather than an action. Sharing your thought process as information is revealed is hypothesis-led working, out loud. It is a view that moves while the interviewer watches it move, which is a different exercise from delivering a structure and then defending it.

The contradiction, and when each side wins

Step three and step six cannot both be followed at once, and the firm does not reconcile them. That tension is the real one a case puts on you, and it is worth resolving before you are sitting in one.

Stick to your approach when the new information sits inside a branch you already opened. A number that is larger or smaller than you expected does not invalidate the structure that asked for it. Finish the branch, say what the number rules out, and move on. Candidates who rebuild their whole approach every time a figure surprises them read as having had no approach.

Rethink when the new information invalidates a branch, or reveals one you did not open. If you structured a profit decline around cost and the interviewer tells you the whole category lost a fifth of its volume, the structure was wrong at the root, and filling it in will not save it. Say so out loud, name what changed your mind, and restructure.

The distinction is simple. Does the new fact contradict your map, or fill it in? Say which one is happening at the moment you decide, out loud, because the deciding is the part being assessed and none of it is visible otherwise. A silent pivot looks like drift. A narrated pivot looks like judgment.

Compare it with how another firm words the same moment and the difference is real. Bain's case interview page asks you to build on ideas as the conversation evolves, and our Bain guide reads that page in full. L.E.K. asks you to move methodically, point by point, and to rethink against new data. Same interview, different default.

What L.E.K. asks you to do before the case interview

Four suggestions, and two of them are more specific than the usual advice.

It tells you to practice with people who currently work in strategy consulting, and to know the kind of work the firm does so you can anticipate the questions. That is a harder ask than practicing with a friend, and it is the one worth acting on, because a practice partner who has run a real case pushes back where a peer accepts.

It tells you to review the case studies and other content on its own website. Note the wording: those case studies are client work, and none of them is a practice case. The three prompts described above are a separate thing, printed on the preparation page itself, and the firm publishes no solution to any of them. A URL we previously pointed readers to is dead. What the rest of the site carries is published client work, which tells you what its cases are made of.

It tells you to attend case preparation workshops on campus, and to read the business press regularly. Both are ordinary, and both are cheap.

Put together, the four say something consistent: the firm expects familiarity with its work and with the format, and it expects you to have said a case out loud to someone who knows what good sounds like.

What one associate says the process was

The firm publishes a first-hand account from a first-year associate in London. Treat it as what it is: one person, one office, one role, writing on their employer's platform. It is first-hand about his experience and it is not a statement of recruiting policy.

What he describes is multiple rounds, including online testing and face-to-face interviews. The online testing included psychometric and math exams. The in-person interviews were case-based and covered problem solving and fit. He is explicit about the standard: "The key was not necessarily getting the right answer but demonstrating structured problem-solving skills."

The most useful thing in the post is a mistake. Asked to size the wedding market, he built a clean methodology, talked it through clearly, produced a figure, and had the interviewer point out that he had sized as though each person had a wedding rather than two people per wedding. He reports that they laughed about it.

Take two things from that. The sanity check on a sizing answer is a question about units. Does the thing you counted exist in the world the way you counted it? No amount of careful multiplication will tell you. Count households. Count weddings. Then say which unit you chose before you multiply anything by it. And the recovery mattered more than the slip, which is what step six describes.

A preparation order that uses what L.E.K. publishes

No fixed calendar, because the right length depends on which dimension you lose, and anyone selling a calendar has not asked.

1. Diagnose, in twenty minutes. One short drill of each type: math, structuring, brainstorming, exhibits, synthesis, sizing. Read the dimension each rubric names as the first correction. That list is the plan.
2. Rehearse steps one and two until they are automatic. Paraphrase the prompt, state your assumptions, restate the facts. Ten seconds, and it is free insurance the firm has told you to buy.
3. Practice the pivot out loud. Take a case, have someone hand you a fact that breaks a branch, and narrate the decision to restructure. That is step three against step six, and it is the thing you cannot learn silently.
4. Check every sizing answer against the unit. The market sizing drill scores the assumption chain separately from the arithmetic, which is where the wedding mistake lives.
5. Read two pieces of the firm's published client work in the sector you are applying to, and take one specific thing from each into the conversation.
6. Find a practice partner who has actually done this job, because the firm names that specifically and a peer cannot replace it.

A commercial due diligence case, worked through

L.E.K. leans commercial: due diligence, growth strategy, market assessment. Cases sit close to that work, so here is one in that shape. The case is ours. The firm prints three prompts and no answers, so nothing below reproduces anything of its own, and the point here is to rehearse the seven steps end to end.

1. Play back the objective. "A private equity client is considering buying a chain of veterinary practices and wants to know whether the market supports the price. Success is a view on whether the growth assumptions hold. Over what hold period?"
2. Ask two or three things that change the structure. Is growth coming from more pet-owning households, more visits per household, or higher prices per visit? How much of the chain's growth so far is acquisitions rather than like-for-like? Those two answers reshape everything.
3. Structure from the arithmetic, and name the unit. "Market value is active households times visits a year times spend per visit. The chain's share of that is practices times revenue per practice. I'll count households rather than animals, because the bill goes to a household. I'd test whether market growth or share gain is doing the work, then whether either is repeatable."
4. Run the numbers aloud. Say 12 million pet-owning households, 60% taking an animal to a vet in a year, so 7.2 million active households. Two visits a year at $90 is $1.3bn. If the chain has 180 practices at $1.4m, that is $252m, or roughly 19% share. To double revenue in five years at flat market growth, share would have to reach nearly 40%. In a fragmented market with local competition, that is a different proposition from the plan's headline.
5. Recommend, first sentence. "The growth case depends on continued acquisition rather than like-for-like growth, so the price should be tested against deal supply rather than against market growth. Evidence: doubling revenue at flat market growth implies moving from roughly a fifth to two fifths of the market. The risk is that acquisition multiples rise as the chain scales, so I'd size the remaining independent practices before anything else."

That is the shape of a diligence answer: what would have to be true, tested, with the number that makes it implausible stated aloud. Notice that the unit was fixed in step three and never changed afterwards. Switch from households to animals halfway and the share figure moves without anything in the business moving. Commercial due diligence goes further into the frame.

The entry title, and the seats around it

The roles page puts graduate entry at Associate and the post-MBA level at Consultant. McKinsey's MBA page sends its own students to Associate openings, so the same word marks two different rungs at the two firms. Our McKinsey guide carries that firm's titles. Read the posting rather than general advice written about the Associate case.

Both levels have a summer route. The firm describes each as eight to ten weeks of real casework, opening with a week of training and a mentor team that stays with you. Summer Associates are undergraduates. Summer Consultants are MBA candidates, and the firm says their work streams match the full-time version of the same seat.

One line on that page describes the work rather than the interview, and it shapes what the cases feel like. The firm says it gives early management responsibility and assigns shorter cases. Shorter engagements mean more of them, which means more first weeks, more scoping and more answers delivered before a question has been fully explored. That is the rhythm a commercial due diligence practice runs at.

L.E.K. publishes no standard by level, so what follows is our reading. At graduate entry you are read on potential: a structure that holds under time, arithmetic you run unprompted, and a recommendation you defend when pushed. Sizing turns up constantly in this kind of work, and sizing is scored on its own rubric where the assumption chain carries more than the multiplication.

What a spoken calculation is scored on

The account above puts the standard on structured problem solving rather than on the right answer, which is a claim about process. No firm publishes a marking scheme for it. The weights below are ours, measured by our own grader on a math answer, each out of a hundred.

Where candidates lose the LEK case interview

1. Skipping the paraphrase. Step one is printed, it costs ten seconds, and it catches a misread objective while that is still free.
2. Rebuilding the structure at every surprise. Step three says move methodically. A number that lands inside a branch you opened does not invalidate the branch.
3. Refusing to rebuild when the map is wrong. Step six says rethink against new data. Holding a structure the evidence has broken is not discipline.
4. Pivoting silently. Both of the two above are assessed through the narration, and a decision nobody hears did not happen.
5. Sizing the wrong unit. Counting people when the unit is households, or participants when the unit is weddings. The firm's own blog carries the example.
6. Stopping at the number. Interpretation is 20 points, and it separates a calculation from a finding.
7. Practicing only with peers. The firm names practice with current strategy consultants specifically, and it is the hardest of its four suggestions to act on.

The first four come straight from the published steps, which makes them the cheapest to fix: the instructions are on a page anyone can read. The rest recur at every firm, which is why a drill diagnosis is worth more than firm-specific study once the format is known.

The three case interview prompts L.E.K. prints, none a market entry

Read on 30 September 2026 on the same preparation page as the seven steps. Each one arrives with a prompt, a picture, and a short list of questions the firm wants you to put to yourself before answering. No solution is published for any of them.

A Brainteaser. Planes leave JFK and LAX every hour on the hour, flying the same five-hour route in opposite directions. You board in New York at noon Eastern. The question is how many planes on the opposite route you pass on the way. The firm asks you what logical traps you want to avoid and what information really matters against what is superfluous, which tells you where it expects the mistake to happen.

Brewery Profits. Two competitors, Cheapo Beer Company and Bourgeois Beer, Inc., report similar revenues growing steadily over five years. Cheapo's profits grow at a moderate rate. Bourgeois's fluctuate. You are asked why, and what you would do as CEO of Bourgeois to damp the swings. The firm asks how you would organize your response and what data you would want.

Large Pharma Co. The client has a daily pill that regrows hair to the thickness you had at fifteen within three months. You are asked to size the market and describe how you would price it. This is the only one of the three that names two tasks in one prompt, and the firm's own list asks what assumptions you would need to make and why.

The set tells you something the seven steps do not. One brainteaser, one profitability diagnosis, one sizing and pricing question: three formats the firm chose to show, and not one of them is a market entry question.

What L.E.K. does not publish

The number of rounds, or what each contains. Nothing on the careers pages. One consultant writes that there were multiple rounds including online testing, and that is testimony from one office rather than a published process.

Whether there is an online assessment for you. The same account describes psychometric and math testing in London. One associate's experience does not generalize into a rule, and the careers pages describe no assessment at all.

Any interview length, threshold or weighting. Nothing published, and we will not invent one.

Any solution to the three cases it prints. The firm publishes the prompts and the questions to ask yourself. It publishes no answer, no exhibit and no model structure for any of the three.

How the process differs by office or by level. The one account we have is London, associate. The firm recruits across five regions and says nothing about variation.

What the firm does publish about itself is counted on our page on L.E.K. Consulting: the 1983 founding, the 27 offices across 16 countries, the 200 partners against 2,300 people, and the fact that it never expands its own initials.

Every statement above about L.E.K. is a quotation from one of its pages, a quotation from the associate's post labeled as testimony, or an absence we name as one. The comparisons with Bain and McKinsey carry those firms' own pages as sources. The weights and the worked case are ours, and the page says so where they appear. If your own experience of this process differs, write to us with what you saw and when, and we will read it back against the firm's pages and date whatever we change.

What we corrected on this L.E.K. page

Every entry carries its date.

29 September 2026. The split first round. An earlier version of this page said the firm describes a first round made of a strategy interview and a quantitative one, and built its title, its tables and much of its advice on that. The preparation page runs to 1,137 characters and describes no rounds at all. The claim was ours and it has gone, along with everything resting on it. The page now leads on the seven numbered steps, which is what the firm does print.

29 September 2026. The online testing. The page said we had found no description of an online assessment, while citing an associate's post that describes psychometric and math exams in London. The post is quoted properly above and labeled as one person's testimony.

1 October 2026. Two passages that had not caught up. The correction below was applied to the sourcing register and to the open-questions section, and two paragraphs in the body were left saying the opposite. A page that contradicts itself is worse than a page that is wrong once, and this one did it for a day.

30 September 2026. The cases exist after all. The firm now prints three case prompts on its preparation page: A Brainteaser, Brewery Profits and Large Pharma Co. On 29 September we read that page in full and there were none, and we said so. They are described above with the date we read them.

29 September 2026. A dead link. A URL we offered as the firm's worked interview case returns a 404. It has been removed rather than softened, and on that date the page said no worked case was found on the site.

29 September 2026. Two claims about the blog. The page said the firm expects candidates to have read its publications, and described a rehearsal method of pausing a recorded case at each question. Neither appears in the post we cited. Both are gone.