Ch12 Discussion

Discussion Guide

1. "Which one do you want more of?" Open with the trout and the burger before anyone has defended a position in writing. Ask the room to commit, then reveal the contribution margins, then reveal the unit counts.

Listen for: the students who reach for percentage first — that reflex is the thing the chapter is treating, and naming it in the room is more effective than reading about it. Listen also for the student who asks "how often does each one sell?" before answering. That student has already understood the chapter; make them explain why the question mattered.

2. Bellwether has three Stars, one Plowhorse, one Dog, and no Puzzle. Is that a good menu or a suspicious analysis?

Listen for: the recognition that both lines are computed from these five items, so it is arithmetically impossible for everything to be a Star. Push toward the deeper point — the tight $5.01 CM spread means the margin axis is doing almost no work here, so the quadrant labels are separating items that are nearly identical in earning power. A strong answer notices that "three Stars" is partly a statement about how few items are on the menu.

Redirect if: the discussion becomes a debate about whether the food is good. The matrix has no opinion about the food, and saying so is the useful move.

3. The trout misses the popularity line by 1.2 units a week, and a full season of data still cannot place it. So what do you actually do on Monday?

Listen for: students trying to solve it with more data. Let them try, then walk them through §12.8's arithmetic — 3,300 entrées over a season still leaves a band of roughly 12.4% to 14.8% around a 14.0% line. The productive answer is the practitioner's one: gather what you can, recognize when the data has stopped being able to answer, then decide on judgment and write the reasoning down with a review date.

The point to land: writing down the reasoning is what converts a judgment call into experience you can evaluate later. Without it you have made the same guess forever.

4. Cut the ember trout. Now defend the number you just produced.

Split the room and give the two halves different substitution assumptions (60/25/15 and 80/20/0). Let them compute, then compare.

Listen for: the moment a student says "one of us must have made a mistake." That is the entry point. Neither group made an arithmetic error; the assumption produced a $14,869 swing between −$10,273 and +$4,596.

Push further with: "So how would you actually estimate the defection rate?" Good answers get operational fast — ask servers how often a table asks whether there is a fish; count trout going to two-tops where the other guest ordered something else; look at whether trout orders cluster with wine by the glass. That is real research an independent can do in two weeks, and it beats another quarter of POS data.

5. Case Study 2's owner cut six items and the food cost went up two periods later. Whose fault is that — the operator's, the software's, or the framework's?

Listen for: students who blame the software. Push back: the report was arithmetically correct. It answered "which items are below the two averages," and the owner heard "which items should I cut."

The distinction worth extracting: a tool that prints a verdict is more dangerous than one that prints a measurement, because it supplies the confidence to act without supplying the information required to act well. Ask what a responsibly designed report would have printed instead. Answers that mention showing distance-from-line, flagging items within the noise band, and refusing to label without disclosing the threshold are exactly right.

Connect to: Exercise 12.39, on the professional obligation not to present undisclosed-threshold labels to a chef whose dish is about to be cut.

6. Case Study 2 is a labeled composite. Case Study 1 is real companies with the item-level numbers missing. Which teaches you more, and what did each one cost you?

Listen for: the recognition that this is a genuine methodological trade-off rather than a flaw. The composite computes all the way through and can therefore demonstrate a mechanism; it cannot prove that the mechanism occurred. The real case proves the argument exists at scale; it cannot show you the arithmetic, because independent restaurants and franchisees do not publish item-level margins.

Ask directly: "What would you demand of an author before trusting a composite's numbers?" Good answers: internal consistency you can check, an explicit label, and no claims that require the specific figures to be true.

7. Bellwether's plan shows a 28.0% ideal entrée food cost against a 30% target. Is that two points of good news?

Listen for: students who want to spend it. Walk them through the three reasons not to — it is ideal cost and actual always leaks; it is a projected mix with no evidence behind it; and it covers 40.0% of dinner sales and none of brunch or beverage. The two points are worth $22,320 on the plan's food sales, which is exactly why treating them as a cushion rather than a saving is the honest move.

Close the loop: ask what would have to be true, and by when, for the partners to start believing the 28.0%. The answer — four periods of ideal-versus-actual comparison — is Chapter 11 and Chapter 13, and it is a good place to end the session.