Practice — Hot · Solution
1–2 · Verdicts
| # | Verdict | Mechanism |
|---|---|---|
| A | Fix (restore zero baseline) | The $1.5M floor turns the true Midwest:Southwest ratio of ~1.6:1 into a visual ~5:1 — the Southwest reads as collapsing. If the small differences genuinely matter, data labels carry them honestly; a broken baseline never does. |
| B | Rebuild | The title claims a trend for one category; a stacked column cannot support it — Lighting floats on other segments, so its apparent motion includes everyone below it. The claim needs a line per category (or Lighting alone). Title and species contradict each other; the species loses. |
| C | Ship. | Right species for a sequence claim, zoomed markers appropriately off at 42 points, and the title makes two claims the line visibly supports (drift up, summer humps). Say so in one sentence and move on — clearing C quickly is what makes your flags on A, B, D, E credible. |
| D | Fix (recolor) | Red-to-green encodes rank, but the theme's semantic colors mean good/bad performance. Decor shows red for being smallest — which is not "bad," it is just fifth — and the board's eye learns that red means nothing. Neutral cards, or rank shown by order. |
| E | Fix (or defensible ship, stated as judgment) | Five slices with labels is inside the pie's honest niche; the L07 share bar is still better (comparable lengths, sums visibly to 100). This one is a preference with reasons, not a violation — the memo should say exactly that, and note that "the CEO likes donuts" is a real input a working analyst weighs. |
3 · Damage ranking
B, A, D, E. B leads because its title asserts a specific false-supportable claim about momentum — a board could act on it. A distorts magnitudes but the labels still carry truth for a careful reader. D wastes semantics without asserting anything false. E is a legibility preference. (Reasoned alternatives exist — B and A can argue — but E first would be wolf-crying and D first mistakes annoyance for damage.)
4 · A memo that would land
"C ships as-is — right chart, and the title's two claims are both visible in the line. Two fixes: A needs its axis back at zero (the current floor shows Southwest at a fifth of Midwest's height when the real ratio is about 1.6 to 1 — labels can carry the fine differences instead); D's red-to-green encodes rank, but our theme reserves those colors for good-versus-bad, and Decor isn't 'bad' for being fifth — go neutral. One rebuild: B's title promises Lighting's trend, and a stacked column can't show one category's trend — Lighting rides on the segments below it. A line for Lighting (or per-category lines) supports the claim; keep the stack only if the claim becomes 'how the total composes.' E is inside the honest-pie niche at five labeled slices; I'd still prefer our share bar, but that's preference, not correction. Net: one rebuild, three ten-minute fixes, one clean ship."
Reflection answers
- Most reviewers report E — because it requires separating "violates a rule" from "differs from my preference," which is the hardest line in reviewing. (C is the runner-up, for the opposite reason: confirming correctness thoroughly enough to sign your name to it takes real reading.)
- Directly: a reviewer who flags all five teaches the team to discount the review, exactly as a checker that cries wolf teaches you to ignore it. Precision is what makes recall valuable — in chart review, code review, and the course's own build tools. Flag what is wrong, clear what is right, and say which is which plainly.