Reading Exhibits: How to Interpret Charts Under Pressure

Interviewers hand you exhibits to test a specific sequence: can you orient in unfamiliar data, find what matters, quantify it, and connect it to the case—out loud, in about a minute? The chart is rarely complicated. The pressure is the point.

Knowing the test changes the preparation. You're not training chart literacy; you're training a repeatable routine that works when your heart rate is up. Routines survive pressure. Improvisation doesn't.

The first fifteen seconds belong to the frame, not the data: read the title, both axes, the units, and the time period—ideally aloud. 'This shows contribution margin by product line, quarterly, over three years, in percentage points.' Only then look at the shape of the data.

Skipping orientation is the single most common exhibit failure. Candidates who dive at the biggest bar routinely misread units—confusing percent with percentage points, or millions with thousands—and build two minutes of analysis on the misreading.

Every case exhibit contains a deliberate signal: a trend that breaks, a segment that defies the average, two lines that cross, a bar wildly out of family. The interviewer built the chart around that anomaly. Your job is to name it within the first minute.

The reliable search pattern is comparison: against time (what changed?), against peers (who deviates?), and against expectation (what should this look like for a healthy business?). The anomaly lives where one of those comparisons snaps.

Naming the anomaly earns half credit. Sizing it earns the rest: 'Retail margin fell eight points while wholesale held flat—on retail's 60 percent revenue share, that's roughly a five-point drag on total margin, which is most of the decline we're solving for.'

One or two quick calculations per exhibit is the norm. Choose the calculation that connects the anomaly to the case question, not the one that's easiest—the interviewer notices the difference.

Close your exhibit read with implication: what does this mean for the client's decision, and what would you check next? 'This says the margin problem is a retail-channel problem, not a product problem—so I'd next want retail pricing and promotion data.' That sentence is what the exhibit was for.

Structure the whole delivery as headline first: lead with the insight, then support it from the chart. Candidates who narrate their scanning process ('so on the left we have…') bury the signal in travelogue.

A handful of trap patterns recur across firms: truncated axes that inflate small differences, mix effects where every segment improves while the average worsens, cumulative charts disguised as growth, and market share given in percent while the market itself shrinks.

You don't need to detect these from suspicion—you need the habit of asking one denominator question per exhibit: 'percent of what, and is the what changing?' That single question defuses most engineered misreads.

Pull any chart from a business publication, set a ninety-second timer, and run the routine aloud: orient, anomaly, quantify, so-what. Ten charts a week for three weeks makes the routine automatic—at which point exhibit sections switch from threat to scoring opportunity.

Then upgrade the difficulty: practice on charts with deliberate traps, and on pairs of exhibits that must be combined. Multi-exhibit synthesis—linking a cost chart to a volume chart—is where final rounds are decided.