Case Interview HQ
Case interviews

Case Interview Charts: How to Read an Exhibit in 60 Seconds

Short answer

To read a case interview chart, check the title and source, then the axes and units, then find the one number that stands out, and finish with what it means for the client's question. Say each step out loud. Done in about 60 seconds, this routine works for live exhibits, McKinsey Solve, BCG's Casey and the table and graph sections of tier-2 tests.

Key takeaways

  • Interviewers hand you a chart to see whether you find the insight that matters for the client, so always end with the so-what.
  • Read the units, the base year and the footnotes before any number; most chart traps sit there.
  • Keep percent and percentage points apart: a share falling from 25% to 22.5% is a 2.5-point drop and a 10% relative decline.
  • Stacked bars, dual axes and truncated axes are the three formats most likely to mislead under time pressure.
  • The same reading routine scores in live cases, gamified screens and timed numerical tests.

Why do interviewers hand you a chart?

Interviewers hand you a chart to test whether you can find the one fact that moves the client's decision, quickly and without being told where to look. BCG's official case preparation page lists analyzing data and quick calculations as core parts of every case. A chart combines both: you must read precisely, compute something, and connect the result to the question.

Most exhibits contain more data than you need. That is deliberate. The interviewer watches whether you filter, and whether you end with a conclusion or a description. In virtual interviews, StrategyCase reports that exhibits arrive by screen share or chat, so practice on a screen as well as on paper.

A four-step reading routine

Use the same four steps on every exhibit, and say them out loud.

  1. Title and source. Establish what the chart claims to show, for which company, market and period. Read the footnotes now, since they often redefine a category or change the base year.
  2. Axes and units. Units (EUR, units, %), scale (thousands, millions), and whether an axis starts at zero. Note if there is a second axis.
  3. The outlier. Find the one figure that breaks the pattern: the fastest growth, the biggest gap, the reversal. Calculate it if needed.
  4. So-what. Link the outlier to the client's question in one sentence, then propose what to examine next.

Steps 1 and 2 take about 15 seconds together, step 3 about 30, and step 4 about 15. The arithmetic in step 3 is usually simple, but it must be done without a calculator; the case interview math guide covers the shortcuts.

Six chart types and the trap in each

Chart type Typical use in a case The trap
Bar chart Comparing segments or competitors An axis that starts above zero makes small gaps look large
Line chart Trends over time Two lines on two different axes invite false comparisons
Stacked bar Composition over time Only the bottom segment shares a baseline; the others are hard to compare by eye
Pie or 100% bar Shares of a total Shares hide the size of the total, which may have shrunk
Waterfall Bridges from one value to another Signs and the order of steps; each middle bar is a change from the previous total
Scatter or bubble Relationships between two variables Correlation read as causation; bubble area misjudged as diameter

Percentages versus percentage points

Percentage points measure the absolute gap between two percentages. Percent measures the relative change. A margin that falls from 20% to 15% has dropped 5 percentage points, which is a 25% relative decline (5 / 20). Mixing the two up is one of the fastest ways to lose an interviewer's confidence, because the two numbers can lead to different recommendations. Say "points" for gaps between shares and margins, and "percent" for relative changes.

Worked exhibit 1: a market share table

The client, Velora (a fictional Dutch e-bike brand), has seen unit sales fall and asks why. All data below is invented for practice.

Exhibit 1: Dutch e-bike sales by brand, thousands of units

Brand 2023 units 2023 share 2025 units 2025 share
Velora (client) 100 25.0% 81 22.5%
Brand B 80 20.0% 90 25.0%
Brand C 60 15.0% 54 15.0%
Brand D 40 10.0% 36 10.0%
Others 120 30.0% 99 27.5%
Total 400 100% 360 100%

Weak reading: "Velora went from 100 to 81, Brand B from 80 to 90, Brand C from 60 to 54, and the total went down too. So Velora is losing."

Strong reading: "This shows unit sales and shares by brand for 2023 and 2025. The market shrank 10%, from 400,000 to 360,000 units. Velora fell 19%, so about half of its decline comes from the market and half from losing share: its share dropped 2.5 points, from 25% to 22.5%. Brand B is the only brand that grew, gaining 5 points. I would like to understand what Brand B changed, for example price, range or dealer network."

The strong reading separates market decline from share loss, uses points correctly, and ends with a hypothesis that tells the interviewer where to go next.

Worked exhibit 2: a cost waterfall

The client, a fictional bakery chain, sells a loaf at EUR 1.60 and saw its margin halve in one year.

Exhibit 2: Cost per loaf, 2024 to 2025 (EUR)

Step Change Running total
Cost per loaf, 2024 1.20
Flour +0.12 1.32
Energy +0.06 1.38
Labor +0.04 1.42
Packaging -0.02 1.40
Cost per loaf, 2025 1.40

Weak reading: "Costs went up because of flour, energy and labor, and packaging went down a bit."

Strong reading: "Cost per loaf rose 20 cents, from EUR 1.20 to EUR 1.40. With the price flat at EUR 1.60, the margin fell from 40 cents to 20 cents, which is 25% to 12.5% of price: a 12.5-point drop, or a 50% relative decline. Flour explains 12 of the 20 cents, or 60% of the increase. So the first question is whether flour prices are temporary or structural, and whether the client can pass part of the increase on in price."

How chart reading shows up in online tests

The same skill is scored in nearly every consulting screen, with a fixed question and a timer added.

  • Kearney: StrategyCase reports two separate sections, reading a table and reading a graph, with six questions each (candidate-reported).
  • L.E.K.: data interpretation from tables and charts is one of six topic groups, without a calculator (candidate-reported).
  • McKinsey Solve: Road to Offer notes that Redrock Study tests data interpretation with similar chart types and footnote-level detail.

In a timed test, reverse the first step: read the question, then the chart. Try this one.

Exhibit 3: Revenue of a fictional retailer by channel, EUR millions

Channel 2022 2025
Stores 170 180
Online 30 60
Total 200 240

Question: calculate the percentage change in store revenue from 2022 to 2025. Drawn as a 100% stacked bar, the store segment shrinks from 85% to 75% of revenue, which tempts a "decline" answer. In absolute terms, stores grew from 170 to 180, or about 5.9% (10 / 170). The answer is growth of about 6%.

Stephan's note: The written test I took, McKinsey's PST, was built almost entirely on exhibits like these, and it rewarded the question-first habit above everything else. The Key to the PST gives you 26 timed questions on dense exhibits with a full answer guide. The test it was built for is retired, but the skill is the same one Solve, Casey and the tier-2 tests score.

For more practice, work through the PST sample questions, follow the 4-week practice plan, and see how each firm's screen uses exhibits in the consulting online assessment overview and the McKinsey test guide.

Frequently asked questions

How long should I spend reading a chart in a case interview?

Take 30-60 seconds of silent reading for a typical exhibit, and tell the interviewer you are doing so. Dense tables with many rows can justify a little more.

What do I say first when I get an exhibit?

Describe what the chart shows in one sentence (the metric, the units, the period), then state the most important observation. Skip reading every data point aloud.

What if I do not understand the chart?

Ask a precise question about the element you are unsure of, such as the unit or what a category includes. Interviewers expect clarifying questions and prefer them to a wrong assumption.

Do I need to calculate something from every exhibit?

Usually yes. Most case exhibits hide one calculation, such as a growth rate, a share change or a gap to a benchmark, that turns the observation into an insight.

Are online test charts different from live interview charts?

The chart types are the same, but online tests attach a fixed question and a timer. Reading the question before the chart matters even more there.

Sources

  1. BCG: Case interview preparation
  2. StrategyCase: Kearney recruitment test
  3. StrategyCase: L.E.K. numerical reasoning test
  4. Road to Offer: Case interview data interpretation
  5. StrategyCase: Online case interview

Stephan Spijkers

Stephan Spijkers founded Case Interview HQ in 2011 after working through the strategy consulting selection process himself: written test, case rounds and partner interviews. He has interviewed candidates from the other side of the table, coached students one-on-one and in group sessions, and wrote Cracking the Case and the two Key to the PST practice books. Today he runs PIMvendors.com, a product information management consultancy.