PartnerRound- 01Title and unitsWhat is measured, and in what
- 02Big pictureThe main trend or gap
- 03OutlierWhat breaks the pattern
- 04So whatWhat it means for the case
By the end you can
- Read any chart or table with a fixed four-step routine in under a minute
- Say something useful in the first 15 seconds without guessing
- Spot truncated axes, percentage point confusion, mix effects and averages that hide segments
- Turn an observation into a so-what that moves the case forward
On this page
01What exhibits test
Most cases include at least one chart or table, often handed over without a word. The interviewer wants to see how fast you find the one thing that matters, and whether you connect it to the client's question.
Description is not insight. 'Sales went up in 2022 and down in 2023' reads the chart back. 'Sales fell 15% in the year the competitor launched, and only in the north' is something a client can act on.
Exhibits also hide traps on purpose: a footnote that changes the units, a scale that exaggerates a gap, an average that conceals two opposite trends. Candidates who rush past the title walk straight into them.
02The four-step routine
- Title and units. Read the title, axis labels, units, time period and footnotes. Is it $m or $bn, share or absolute, annual or cumulative, nominal or adjusted for inflation?
- Big picture. Find the one main pattern: a trend, a gap between groups, a ranking. State it with a number.
- Outlier. Find what breaks the pattern: the segment moving the other way, the year that jumps, the bar that dwarfs the rest. This is often where the case is.
- So what. Say what it means for the client's question, and what you would look at next.
Spend 20 to 30 seconds on the first three steps, in silence if you need to. Write down the two or three numbers you will quote and do any quick maths before you speak, so your first sentence is right.
Quick maths on exhibits
03The first 15 seconds
The silence after an exhibit lands feels long. Fill it with something useful and safe: what the exhibit shows, then a request for a moment. Never guess at the insight before you have read the axes.
If something is unclear, ask before interpreting. 'Is this nominal or adjusted for inflation?' and 'Are these shares of revenue or of volume?' are sharp questions, not weak ones.
04The traps interviewers set
Two more to watch. Over long periods, compare real figures, because inflation alone can create growth. And on cumulative charts, a line that only ever rises can hide a sharp slowdown.
Summed across segments: shift the shares and the blend moves.
05Worked example: are margins collapsing?
Worked example
Retailer: 'Our margins are collapsing'
The CFO of a retailer says margins are collapsing and wants to know why. The interviewer hands you this table and says nothing.
1. Title and units
Revenue in $m and gross margin in percent, by channel, two years apart. The total row gives the blended margin: 35% falling to 31.5%.
2. Big picture
Revenue grew 25%, from $400m to $500m, but blended margin fell 3.5 percentage points. That is a 10% relative fall (3.5 ÷ 35), which is what has alarmed the CFO.
3. Outlier
Margin rose in both channels: stores from 40% to 41%, online from 20% to 22%. Every segment improved, yet the total fell. That is the signature of a mix effect. Online grew from 25% of sales to 50%, and it earns roughly half the margin of stores.
Quantify it. At 2023 margins, the 2025 mix would give 0.5 × 40% + 0.5 × 20% = 30%. So the shift in mix cost 5 points, and better margins in each channel won back 1.5: 35% - 5 + 1.5 = 31.5%.
In dollars, gross profit rose from $140m ($120m + $20m) to $157.5m ($102.5m + $55m), up 12.5%.
4. So what
Margins are not collapsing. The business is shifting towards a lower-margin channel that is still adding profit. The real questions are whether online margin can keep rising, and whether online sales are new or taken from the stores.
Weak
So store revenue went from $300m to $250m, online went from $100m to $250m, and the margins are 40%, 41%, 20% and 22%. Total margin has gone down from 35% to 31.5%.
Strong
The headline is that blended margin fell 3.5 points, from 35% to 31.5%. But margin rose in both channels, so the whole fall is mix: online, at about half the margin of stores, has grown from a quarter of sales to half. So margins aren't eroding, and gross profit is actually up $17.5m. The question is whether online margin can rise further, so I'd like to look at online cost per order next.
Why: The weak answer reads the table back and stops at the alarming number. The strong one leads with the headline, finds the twist, puts a number on it and turns it into a next step.
06Mistakes and checks
- Missing the units or a footnote, and being out by a factor of 1,000.
- Comparing two lines that sit on different axes.
- Talking for two minutes before reaching the point.
- Treating the exhibit as a fresh puzzle instead of linking it to the step of the case before it.
Check yourself
A client's market share rose from 8% to 10%. Which statement is correct?
Check yourself
Every region's conversion rate rose this year, but the company's overall conversion rate fell. What is the most likely explanation?
Cheat sheet
The lines worth rereading the night before.
- 1Four steps, every time: title and units, big picture, outlier, so what.
- 2First 15 seconds: say what the exhibit shows, then ask for a moment. Never guess.
- 3Lead with the one insight, with a number, then the so-what and the next step.
- 4Percentage points for the gap between two percentages; percent for relative change.
- 5If every segment improves but the total falls, it is mix.
- 6Check where the axis starts before you believe a gap.
- 7Averages hide segments: ask for the split.
Now drill it
Reading builds the model; drills make it automatic. Start with the first topic, easiest drills first.
- Reading charts7 drills · Read a chart fast, pull out the number that matters and avoid the tempting wrong conclusion.
- Tables & exhibits7 drills · Find the insight in a dense table: margins, shares and the segment driving the change.
- Sanity checks7 drills · Catch the implausible number before the client does.
Questions candidates ask
How long should I take to read an exhibit in a case interview?
About 20 to 30 seconds of quiet reading is normal, once you have said what the exhibit shows. Much more than a minute without speaking starts to look stuck, so share your first observation and build from there.
What if I don't understand a chart in a case interview?
Ask. 'Is this nominal or adjusted for inflation?' or 'Are these shares of revenue or of volume?' are sharp questions. Building an argument on a misread chart costs far more than a ten-second question.
Should I describe everything in the exhibit?
No. Lead with the one or two points that matter for the client's question, with numbers. Mention anything else only if it changes the conclusion.
What is the difference between percent and percentage points?
Percentage points measure the gap between two percentages: 10% to 12% is up 2 points. Percent measures relative change: the same move is a 20% increase. Interviewers expect you to keep the two apart.