Chapter 24: Revenue Management: Covers, Seat Turns, RevPASH, and Pricing the Room

"You cannot sell yesterday's seven o'clock." — constructed; the sentence every operator eventually reduces this job to

Overview

Here is the number that should keep you up, and it is not a cost.

Bellwether will manufacture 116,688 seat-hours next year. Sixty-eight chairs, five hours a night, five nights a week, plus two brunch services on the weekend. Nobody orders those seat-hours, nobody counts them, and they do not appear on any report the restaurant will produce. They are made continuously, whether anyone is sitting in them or not, and the plan sells about forty-five percent of them.

The other fifty-five percent are not slow-moving inventory. They do not go on the shelf. They cannot be marked down next Tuesday, frozen, repurposed as a special, or written off against next year. At 6:01 p.m. the six-o'clock hour is simply gone, and the rent on it, the insurance on it, the mortgage-equivalent of the hearth that heats it, and the wages of the crew standing in it were all paid in advance.

That is the whole idea of this chapter, and it is the fourth theme of this book: every seat-hour is inventory you can't store. Airlines and hotels worked this out decades ago and built an entire discipline around it. Restaurants mostly did not, because restaurants count covers — a number that is easy to gather, universally quoted, and, on its own, close to useless.

I want to show you exactly how useless. Bellwether's Saturday brunch does 110 covers. Bellwether's Tuesday dinner does 62. The brunch does seventy-seven percent more people, and produces \$212 less revenue. Nobody in the building will notice, because everyone in the building is looking at the cover count, and the cover count says brunch is nearly twice the service.

Chapter 19 already found half of this problem from the cost side: because labor was modeled as a flat percentage of sales, the plan budgets the least labor for the emptiest nights and the most for the fullest, which is backwards from how a crew actually works. Revenue management is the other half. You cannot fix a Tuesday by cutting a server; there is a floor below which the room stops functioning. You fix a Tuesday by selling more of what you already manufacture.

And there is a hard stop on all of it, which Chapter 14 nailed to the wall: the hearth sustains twenty-eight items an hour, and wave-splitting creates zero capacity. You cannot manage duration past the fire. Any revenue technique in this chapter that requires the kitchen to go faster is not a technique. It is a wish.

In this chapter, you will learn to:

• Count the inventory nobody counts, and state what fraction of it your restaurant actually sells.
• Compute covers, seat turns, capacity utilization, and RevPASH, and decompose RevPASH into the only three levers that move it.
• Read the shape of a week and the shape of a night, and find the hours where the money is not.
• Set party-size demand against table inventory and price the cost of a two-top seated at a four.
• Use menu design and coursing as duration tools — and recognize the point past which they do nothing.
• Apply demand-based pricing where it works, avoid it where it backfires, and think honestly about the fairness problem it creates when the product is dinner.
• Build a no-show policy sized to the actual exposure, and a shoulder-hour program that fills seats you are already paying for.
• Close Bellwether's revenue bridge — the \$139,240 that has been open since Chapter 1.

Learning Paths

🏗️ Opening — §24.4 is the one to do before your millwork order goes in; a table mix is physically expensive to change and free to get right on paper. §24.8 and the Business Plan checkpoint close the revenue gap your plan has been carrying since Chapter 1. 📋 Managing — §24.2 and §24.3 are your instruments. If you take one habit from this chapter, take the weekly RevPASH-by-daypart report; it will change which shifts you argue about. 🍸 Beverage — the bar is the highest-velocity seat inventory in the building and the cheapest place to raise revenue per occupied seat-hour. §24.4 shows why your twelve bar seats relieve a dining-room problem, and §24.8 is where a shoulder-hour bar program lives. 🚚 Small Format — you have no seat-hours to manage, which sounds like an advantage and is not: your perishable capacity is production hours, and every idea here transfers to a service window and a fryer. Read §24.1, §24.5, and §24.6 closely; skim §24.4.

24.1 The perishable-inventory idea, applied to a dining room

Walk into your walk-in and look at a case of romaine. That is inventory. It has a cost, a shelf life, a par level, and a count sheet. If you do not sell it this week you will feel bad about it, and if you do not sell it next week you will throw it away and the loss will show up in Chapter 13's ideal- versus-actual variance.

Now walk out to table nine and look at the chair. That is also inventory, and it is worth considerably more than the romaine, and almost nobody counts it.

What a seat-hour is

A seat-hour is one seat, available to be sold, for one hour. It is the atomic unit of a dining room's capacity, and the reason it is the right unit is that neither half of it works alone. A seat by itself tells you nothing — a seat that is occupied from 5:00 to 10:00 by one party drinking two glasses of wine is a very different asset from a seat that turns three times. An hour by itself tells you nothing either. Multiply them and you have the thing you actually sell.

Bellwether's arithmetic is simple and worth doing by hand, because doing it by hand is what makes the number real:

FIGURE 24.1 — Bellwether's seat-hour inventory, one week          [the Bellwether plan]

  DINNER    68 seats × 5 hours (5:00–10:00) =   340 seat-hours per service
            340 × 5 services (Tue–Sat)      = 1,700 seat-hours

  BRUNCH    68 seats × 4 hours (10:00–2:00) =   272 seat-hours per service
            272 × 2 services (Sat–Sun)      =   544 seat-hours

  ────────────────────────────────────────────────────────────────────────
  WEEK                                        2,244 seat-hours
  YEAR      2,244 × 52                      116,688 seat-hours

  Of which the plan sells, in the sense that somebody is sitting in them:
            about 45%.   The other 64,000-odd are manufactured and discarded.

Sixty-four thousand seat-hours. If you priced them at what an occupied one produces — a number we will compute in §24.2, and it is about \$29 — you would be looking at a number so large it becomes useless, which is why nobody frames it that way. There is no restaurant on earth that sells all its seat-hours, and there never will be, because guests want to eat at 7:30 and not at 5:10 and no amount of management changes that. The prize is not the sixty-four thousand. The prize is three or four points of it, and three or four points of it, at Bellwether's numbers, is somewhere between forty and eighty thousand dollars a year on a business whose plan profit before debt service is around sixteen and a half percent of sales.

Why "perishable" is the exact word and not a metaphor

Three properties define perishable capacity, and a dining room has all three.

One: it is manufactured continuously and cannot be stored. You do not decide to make seat-hours. They accumulate at sixty-eight an hour the entire time the doors are open, and at 10:00 p.m. the day's production is finished and the unsold portion is destroyed. There is no back stock.

Two: the cost is sunk before the sale. Rent, property insurance, the loan payment on the hearth, the salaried manager, the opening prep cook, the closing dishwasher — Chapter 1 called this the fixed labor floor — are all committed before anyone walks in. Chapter 19 sized the floor. The consequence is the one that matters here: the marginal cost of selling one more seat-hour is almost entirely just the food. At Bellwether's blended cost of goods of 27.8%, an incremental \$46 cover contributes about **\$33.23** toward fixed costs and profit, and on a Tuesday, when the crew is already standing there, it costs essentially nothing in labor to serve.

Three: demand is wildly uneven and largely outside your control. Bellwether's Saturday is roughly double its Tuesday. The 7:00 hour is roughly six times the 5:00 hour. Nothing about the food changes between those hours. The variance is entirely a fact about how people live.

Put those three together and you have precisely the problem an airline has with a departing aircraft or a hotel has with a room-night. The seat leaves at 6:40 whether or not it is sold.

👨‍🍳 On the Line

What sunk capacity feels like at 5:20 on a Tuesday in February.

The chef is here. The sous is here. Two line cooks and a dishwasher are here. The bartender has cut forty limes and is polishing glassware that was already clean. A host stands at a stand with an empty book. Two servers are folding napkins they folded an hour ago. Every single one of those people is being paid, and the room is producing sixty-eight seat-hours of inventory an hour, and nine people are in the dining room.

The manager's instinct in that moment is a cost instinct: cut somebody. And sometimes that is right — Chapter 19 builds the staffing guide that tells you when. But most of the time you are already at the floor. You cannot cut the chef. You cannot cut the dishwasher, because at 9:40 there will be dishes. You cannot cut below one server without abandoning half the room.

So the cost lever is spent, and the room is still empty, and the money is still leaving. That is the moment the revenue lever becomes the only one you have. Everything in §24.8 — shoulder programs, early prix fixe, a bar that draws its own traffic, an event on a night you are otherwise dark — exists because at 5:20 on a Tuesday in February, filling is the only move left.

And here is the part that takes new managers a year to internalize: the tenth guest who walks in at 5:45 that night is worth more to the business than the hundred-and-twentieth guest on Friday. Same check, same food cost, but the Friday guest needs an incremental server hour, an incremental hearth slot the kitchen may not have, and a table that somebody else wanted. The Tuesday guest needs nothing. The cheapest covers to serve are the ones you don't have.

What transfers from airlines and hotels — and what does not

Yield management — the discipline of maximizing revenue from a fixed, perishable capacity by varying price and availability across customer segments — is a real body of practice with a documented history. It emerged in commercial aviation after the Airline Deregulation Act of 1978 gave U.S. carriers freedom to set fares, and it spread to hotels through the 1980s and 1990s. Its translation into restaurants was largely academic work out of Cornell's hotel school, and it gave the industry the metric this chapter is built around. That lineage matters, because it tells you which parts to borrow and which parts to leave.

What transfers cleanly:

• Fixed capacity that perishes on a clock.
• Very high fixed cost against very low marginal cost per unit sold.
• Predictable, repeating demand patterns — the shape of the week is remarkably stable once you have a few months of data.
• Demand that can be segmented by time, which means it can be shifted with price and availability.

What does not transfer, and this is where restaurants that copy hotels get hurt:

• The guest chooses the hour, and the hour is most of the product. An airline passenger has a handful of departure times and will trade one for a lower fare. A couple celebrating an anniversary wants Saturday at 7:30, and 5:15 is not the same product at a discount. It is a different product.
• Duration is variable and the guest controls it. A flight from here to there takes what it takes. A table takes between fifty minutes and three hours, and the operator's ability to influence that is real but modest and easily abused.
• Capacity is two-dimensional. An aircraft's constraint is seats. A restaurant's constraints are seats and kitchen throughput, and they bind at different times. Bellwether's Saturday, as we will see in §24.3, is not seat-constrained at all — the room is 57% full — but it is over the hearth's sustainable rate for two hours. Managing the seat when the fire is the constraint accomplishes nothing.
• Price is read as a statement about worth. Nobody thinks a \$219 fare means the airline respects them more than the \$389 passenger. Guests absolutely read a restaurant's prices as a statement about what they are being offered and what the restaurant thinks of them. §24.6 is about living inside that constraint.

Hold those four differences in mind through the rest of the chapter. Every technique that follows either respects them or fails.

24.2 The metrics: covers, average check, seat turns, and RevPASH

Four numbers. You already have two of them from Chapter 1.

A cover is one guest served. Average check — also per-person average, or PPA — is sales divided by covers. Both are Chapter 1's, and both are necessary. Neither is sufficient, and the reason is that neither one knows anything about time or about the size of the room.

Seat turns

Seat turns is covers divided by seats, for a service. It is the number of times you rented each chair.

$$\text{Seat turns} = \frac{\text{Covers}}{\text{Seats}}$$

Bellwether on a Saturday: $123 \div 68 = 1.81$ turns. On a Tuesday: $62 \div 68 = 0.91$ turns — which means that on Tuesday the restaurant does not, on average, sell each chair even once.

Seat turns is a genuine improvement on the cover count because it normalizes for the size of the room. A hundred covers in a 40-seat room is a triumph; a hundred covers in a 200-seat room is an emergency. Turns tells you which you are looking at, which a cover count never will.

But turns has a blind spot, and it is a big one: turns has no idea how long your service window is. A restaurant doing 1.8 turns over an eleven-hour day and a restaurant doing 1.8 turns over a four-hour dinner are not remotely the same business, and turns rates them identically. Turns also cannot compare a dinner to a brunch, or a Tuesday to a Saturday when one of them runs an hour longer, or your restaurant to the one down the street with a different service pattern.

To fix that you have to put time in the denominator.

RevPASH

RevPASH — revenue per available seat-hour — is total revenue for a period divided by the total seat-hours available in that period.

$$\text{RevPASH} = \frac{\text{Revenue}}{\text{Available seats} \times \text{Hours open}}$$

That is the whole formula. What makes it powerful is the word available: the denominator counts every seat-hour you manufactured, sold or not. RevPASH is therefore the only common restaurant metric that is denominated in the thing you actually have a fixed supply of.

🧮 Run the Numbers

Tuesday against Saturday, honestly.

Bellwether, dinner service, 5:00 to 10:00. Sixty-eight seats. Five hours. So every dinner service manufactures $68 \times 5 = 340$ seat-hours, and that number does not change no matter who shows up.

Tuesday. 62 covers at a \$46 average check = **\$2,852. $\$2{,}852 \div 340 = \$8.39$ RevPASH.**

Saturday. 123 covers at \$46 = **\$5,658. $\$5{,}658 \div 340 = \$16.64$ RevPASH.**

Saturday produces \$8.25 more revenue per available seat-hour than Tuesday — 98% more, on 98% more covers, which is not surprising because the check and the hours are identical. So far RevPASH has told you exactly what the cover count told you.

Now do the comparison a cover count cannot do.

Saturday brunch. 110 covers at a \$24 average check = **\$2,640, over a four-hour window: $68 \times 4 = 272$ seat-hours. $\$2{,}640 \div 272 = \$9.71$ RevPASH.**

Read those two lines next to each other:

	Covers	Revenue	Seat-hours	RevPASH
Tuesday dinner	62	\$2,852 | 340 | \$8.39
Saturday brunch	110	\$2,640 | 272 | \$9.71

Brunch serves seventy-seven percent more people and brings in \$212 less money. It does it in a shorter window, so its RevPASH is \$1.32 better — a real but modest edge, and nothing like the gap the cover counts imply.

If you manage this restaurant by covers you will believe brunch is your second-strongest service and Tuesday is a disaster. If you manage it by RevPASH you will see two weak services separated by \$1.32 an hour, and you will start asking the right question, which is why the brunch check is \$24.

That is the case for the metric in one table. A cover count measures traffic. RevPASH measures the productivity of the asset. They are not the same thing and they diverge exactly where the money is.

The full weekly picture

🧾 Read the Numbers

```text FIGURE 24.2 — "The week, in seat-hours" [the Bellwether plan] THE ARTIFACT A one-page weekly RevPASH report by daypart, of the kind a POS can produce nightly once seat counts and service hours are configured. Year-1 plan basis. THE CONTEXT Bellwether has not opened. These are the plan's own figures, restated in seat-hours: 68 seats, dinner 5:00–10:00 Tuesday through Saturday, brunch 10:00–2:00 Saturday and Sunday. Patio closed (the base case). Planned dinner dine time 95 minutes; brunch 70 minutes.

                         Covers  Turns   Revenue  Seat-hrs  RevPASH   Util.
               Tue          62   0.91    $2,852       340    $8.39   28.9%
               Wed          78   1.15    $3,588       340   $10.55   36.3%
               Thu          92   1.35    $4,232       340   $12.45   42.8%
               Fri         120   1.76    $5,520       340   $16.24   55.9%
               Sat         123   1.81    $5,658       340   $16.64   57.3%
               ───────────────────────────────────────────────────────────
               DINNER      475   1.40   $21,850     1,700   $12.85   44.2%
               Sat brunch  110   1.62    $2,640       272    $9.71   47.2%
               Sun brunch  110   1.62    $2,640       272    $9.71   47.2%
               ───────────────────────────────────────────────────────────
               BRUNCH      220   1.62    $5,280       544    $9.71   47.2%
               ═══════════════════════════════════════════════════════════
               WEEK        695   1.46   $27,130     2,244   $12.09   45.0%

               Annualized: 36,140 covers · $1,410,760 · 116,688 seat-hours.

WHAT IT SHOWS Three things the cover column hides. (1) Brunch fills more of its window than dinner fills of its own — 47.2% capacity utilization against dinner's 44.2% — and still returns a lower RevPASH, because a $24 check over 70 minutes produces $20.57 an hour against dinner's $29.05. Brunch is a well-attended, low-yield service, and its cover count says the opposite. (2) The week's revenue is enormously concentrated: Friday and Saturday dinner are 41.2% of weekly sales on 30.3% of the seat-hours. (3) Tuesday and Wednesday consume exactly the same 680 seat-hours as Friday and Saturday and return $6,440 against $11,178 — 57.6 cents on the dollar for identical capacity. WHAT IT DOESN'T It says nothing about profit. A high-RevPASH service can be the least profitable one in the week if it takes disproportionate labor or runs a heavier food cost, and Chapter 32 is where that question gets settled. It also averages away the shape of the night — Saturday's $16.64 is the mean of an hour worth $4.27 and an hour worth $25.18 (Figure 24.5). And it contains no patio, no events, and no takeout, because none of those consume dining- room seat-hours. THE DECISION Adopt this as the standing weekly report from week one, alongside the flash report in Chapter 31, and set the target where the plan can be held to it: a dinner RevPASH floor of $12.85 and a Tuesday floor of $8.39. When a Thursday comes in at $11.10, you have a number to argue about before the month closes. THE LESSON Covers measure traffic. RevPASH measures whether the building is working. A service that doubles its covers and halves its check has done nothing, and only one of these two numbers will tell you so. ```

A word on that "Util." column, because it is the second metric this chapter needs.

Capacity utilization is occupied seat-hours divided by available seat-hours. To get occupied seat-hours you multiply covers by the average dine time in hours. Bellwether's planned dinner dine time is 95 minutes — 1.583 hours — so a Saturday's 123 covers occupy $123 \times 1.583 = 195$ seat-hours out of 340 available, which is 57.3%.

Note what that means. On its single busiest night of the week, with a wait at the door at 7:30, Bellwether's dining room is 43% empty on average. That is not a failure. It is the arithmetic of a peaked arrival curve, and every restaurant lives with a version of it. But it is a useful corrective to the feeling in the room, because the feeling in the room on a Saturday at 7:30 is that the restaurant is completely full, and the building is not full at all. The 7:30 hour is.

The decomposition — the most useful thing in this chapter

RevPASH looks like one number. It is actually two multiplied together, and each of those has its own levers. Watch:

$$\text{RevPASH} = \underbrace{\frac{\text{Occupied seat-hours}}{\text{Available seat-hours}}}_{\text{utilization}} \times \underbrace{\frac{\text{Revenue}}{\text{Occupied seat-hours}}}_{\text{revenue per occupied seat-hour}}$$

And the second term decomposes again, because revenue per occupied seat-hour is nothing more than the average check divided by how long the guest sits:

$$\text{RevPASH} = \text{utilization} \times \frac{\text{average check}}{\text{average dine time (hours)}}$$

Check it against Bellwether's dinner week: $44.2\% \times (\$46.00 \div 1.583) = 0.442 \times \$29.05 = \$12.85$. It resolves exactly.

FIGURE 24.3 — What RevPASH is made of                    [constructed teaching example]

                                 RevPASH
                                    │
                 ┌──────────────────┴───────────────────┐
                 │                                      │
      CAPACITY UTILIZATION                REVENUE PER OCCUPIED SEAT-HOUR
   occupied ÷ available seat-hrs                 ┌──────┴──────┐
                 │                               │             │
     levers:                            AVERAGE CHECK    ÷  DINE TIME
       demand generation                  levers:            levers:
       table mix (§24.4)                    menu design        coursing (§24.5)
       reservation shape                    beverage attach    pacing
       shoulder programs (§24.8)            add-on selling     table reset
       no-show control (§24.7)              pricing (§24.6)    menu length

   Bellwether dinner, on plan:
       44.2%  ×  ( $46.00 ÷ 1.583 hours = $29.05 )  =  $12.85 RevPASH

   THREE LEVERS. That is all there are. Every technique in this chapter is one
   of them, and any proposal that is none of them is not a revenue-management
   proposal — it is a hope.

This decomposition is worth memorizing because it disciplines the conversation. When someone in your building proposes an idea "to grow revenue," the question is always: which of the three does it move, and by how much, and what does it cost in the other two? A prix fixe shortens dine time and lowers the check. A tasting menu raises the check and lengthens dine time. A shoulder-hour program raises utilization at a lower check. Almost nothing moves one lever cleanly, and the operators who get burned are the ones who counted the lever they liked and not the one they paid with.

Now use it on Tuesday. Tuesday's revenue per occupied seat-hour is $\$2{,}852 \div 98.2 = \$29.05$, and Saturday's is $\$5{,}658 \div 194.8 = \$29.05$. Identical.

They are identical by construction in the plan, because the plan assumes one check and one dine time for every dinner service, and you should be suspicious of any conclusion that falls out of an assumption. So test it against reality: in a trading restaurant these two numbers do diverge — a Saturday check usually runs a dollar or two above a Tuesday's on better beverage attachment, and Saturday tables sit a few minutes longer. Call it \$30 against \$28. The spread in revenue per occupied seat-hour between your best and worst night is a few percent. The spread in utilization is 28.9% against 57.3% — a factor of two. That relationship holds in every full-service restaurant I have ever measured, and it is the finding that matters.

The entire difference between an \$8.39 night and a \$16.64 night is the utilization term.

That single finding rules out half the things people propose for a slow Tuesday. Tuesday is not a check problem — the check is fine. Tuesday is not a duration problem — the tables turn at the same rate they do on Saturday, and there is nobody waiting for them anyway. Tuesday is a demand problem, and the only cure for a demand problem is traffic. Discounting the Tuesday check attacks a variable that is not broken, and unless the discount creates traffic that did not exist, it subtracts from the one number that was healthy.

🔍 Check Your Understanding

1. A 90-seat restaurant is open for dinner 5:00–11:00, six nights a week. How many seat-hours does it manufacture in a week? If it does \$34,000 of weekly sales, what is its RevPASH?
2. Bellwether's Thursday does 92 covers and \$12.45 of RevPASH. Its Friday does 120 covers and \$16.24. Both run a \$46 check and a 95-minute dine time. Without computing anything further, what must be the only difference between them, and why?
3. Why can two restaurants with identical seat turns have RevPASH figures that differ by a factor of two?

(1: $90 \times 6 \times 6 = 3{,}240$ seat-hours; $\$34{,}000 \div 3{,}240 = \$10.49$ RevPASH. 2: Capacity utilization — 42.8% against 55.9%. If the check and the dine time are the same, the revenue-per-occupied-seat-hour term is the same, so the entire difference lives in the utilization term. 3: Because turns has no time denominator. A restaurant doing 1.5 turns across a four-hour dinner is selling its seat-hours twice as fast as one doing 1.5 turns across an eight-hour all-day service, and RevPASH sees the difference while turns cannot.)

24.3 Reading the shape of your week and the shape of your night

Every restaurant has a shape of the week — the stable, repeating distribution of demand across its operating days — and a shape of the night inside each of those days. Both are far more stable than new operators expect. After about eight weeks of trading you will be able to forecast a Wednesday within a handful of covers, and you will keep being able to do it for years, interrupted only by weather, holidays, and whatever the neighborhood is doing.

The shape is the single most useful thing you own, and almost nobody uses it for revenue. They use it for ordering (Chapter 13) and for scheduling (Chapter 19), and then they stop.

The shape of the week

FIGURE 24.4 — The shape of Bellwether's week, in RevPASH        [the Bellwether plan]

                $0   $5   $10  $15  $20
                |....|....|....|....|
  Tue dinner    ████████                     $8.39     62 covers   28.9% full
  Wed dinner    ██████████▌                 $10.55     78 covers   36.3% full
  Thu dinner    ████████████▌               $12.45     92 covers   42.8% full
  Fri dinner    ████████████████▎           $16.24    120 covers   55.9% full
  Sat dinner    ████████████████▋           $16.64    123 covers   57.3% full
  Sat brunch    █████████▋                   $9.71    110 covers   47.2% full
  Sun brunch    █████████▋                   $9.71    110 covers   47.2% full
                └─────────────────────
                Week blended: $12.09 across 2,244 available seat-hours.
                One block ≈ $1.00 of revenue per available seat-hour.

  Friday + Saturday dinner  = $11,178 = 41.2% of weekly revenue  ┐ each consumes
  Tuesday + Wednesday       =  $6,440 = 23.7% of weekly revenue  ┘ 680 seat-hours

Look at those last two lines together, because they are the shape of the problem. Two pairs of nights consume exactly the same capacity — 680 available seat-hours each, 30.3% of the week — and one pair returns \$11,178 while the other returns \$6,440. Fifty-eight cents on the dollar, for identical inventory, in the same building, with the same menu and the same crew.

Now here is where this chapter meets Chapter 19, and I want to be precise about it because the interaction is the most expensive thing in Part V.

⚠️ Where the Money Leaks

The percentage-of-sales labor model, seen from the revenue side.

Chapter 19's finding was that Bellwether's plan, which models labor at a flat 32.3% of sales, budgets:

	Revenue	Labor at 32.3%	Covers	Budgeted labor per cover
Tuesday	\$2,852 | \$921	62	\$14.86
Saturday	\$5,658 | \$1,827	123	\$14.85

The model is internally consistent and operationally false. It says a Tuesday crew should cost \$921. But Tuesday needs a chef, a dishwasher, a bartender, a host, and at minimum two servers — Chapter 1's fixed labor floor — and that crew does not cost half of Saturday's crew. It costs something close to Saturday's crew. So the plan under-funds the emptiest night, the operator staffs the floor anyway because the alternative is a broken room, and Tuesday runs a labor percentage in the high thirties or worse.

That is the cost-side story, and Chapter 19 owns the fix: staff to a forecast and a productivity standard, not to a percentage.

The revenue-side story is the same fact wearing a different coat. The floor is a sunk cost on Tuesday. It is being paid regardless. Which means the contribution of an incremental Tuesday cover is not \$33.23 minus a share of labor — it is very close to the full **\$33.23**, because the crew serving it is already on the clock and not busy.

Run that all the way out. Ten more covers on Tuesday and ten more on Wednesday — from 62 and 78 to 72 and 88 — is $20 \times \$46 = \$920$ a week, \$47,840 a year, at a labor cost of approximately zero, because the same crew serves 88 as serves 78. Ten more covers on Friday is the same \$46 each, but it needs an incremental server hour, it lands on a kitchen already over its rate (see below), and it competes for tables somebody is waiting for.

The plan treats every cover as worth \$46. They are not. Slow-night covers are the most profitable covers in the building, and they are the only ones the plan makes no effort to get.

The shape of the night

Now go inside a single service. This is where the seat stops being the constraint and the kitchen takes over.

FIGURE 24.5 — The shape of a Saturday night                       [the Bellwether plan]

  hour          covers   occupied   hearth items   RevPASH
                seated   seat-hrs   fired that hr  that hour
  5:00–6:00        14        10          11         $4.27   ████
  6:00–7:00        26        33          24        $14.08   ██████████████
  7:00–8:00        34        54          32  ◄OVER $23.04   ███████████████████████
  8:00–9:00        31        59          32  ◄OVER $25.18   █████████████████████████
  9:00–10:00       18        39          24        $16.64   ████████████████
  ────────────────────────────────────────────────────────────────────────────
  NIGHT           123       195         123        $16.64   (average of the five)

  Available seat-hours: 68 per hour, 340 for the night. Utilization 57.3%.
  Hearth sustains 28 items/hour (Ch. 14). Firing lags seating by ~20 minutes.
  One block ≈ $1.00 of revenue per available seat-hour in that hour.

Read the columns against each other and the night comes apart in a way the nightly average never shows.

The room is never full. Peak occupancy is 59 of 68 seats in the 8:00 hour — 87% — and the average across the night is 57%. Bellwether does not have a seat shortage on Saturday. It has a seat shortage for about ninety minutes.

The kitchen, however, is over its rate for two full hours. Chapter 14 measured the hearth at twenty-eight items an hour sustained. Between roughly 7:20 and 9:20 the fire is being asked for thirty-two. That is what a fourteen-minute ticket time on a Saturday is made of, and it is why the expo window backs up at 7:45 and not at 6:15.

And the first hour is nearly dead. Fourteen covers seated between five and six, ten occupied seat-hours out of sixty-eight, \$4.27 of RevPASH against a night average of \$16.64. On the busiest night of the week, the first hour of service returns about a quarter of the night's rate.

Those three observations, taken together, produce the single most valuable move available to Bellwether on a Friday or Saturday, and it is not "do more covers." It is move covers out of the 7:00–9:00 window into 5:00–6:00 and 9:00–10:00.

Do the arithmetic. If six covers move out of the 7:00 hour and three out of the 8:00 hour into the shoulders, the hearth's hourly load becomes 11 → 17, 24 → 25, 32 → 28, 32 → 28, 24 → 27. Nothing is over rate. And now there is headroom: the 5:00 and 9:00 hours can each absorb more covers before the hearth is troubled — potentially taking Saturday from 123 covers to something near 137 without the fire ever exceeding twenty-eight items in any hour, and while staying under Chapter 14's nightly ceiling of about 144.

Be careful how you state this, because it is one word away from something Chapter 14 explicitly disproved. Chapter 14 found that wave-splitting creates zero capacity: chopping the same covers into a 6:00 seating and an 8:00 seating does not make the hearth burn faster, and any plan that assumes it does is wrong. That is correct and it stays correct. What reshaping the arrival curve does is different: it does not add capacity, it uses more of the capacity you already own, by moving load out of the hours where the hearth is over rate and into the hours where it is idling at eleven items. The ceiling is unchanged. The area under the curve gets larger.

And it is genuinely hard to do. You cannot make people want a 5:15 table by wanting them to. What you can do — and this is the whole of §24.8 — is make 5:15 worth choosing.

🤝 Hospitality

The difference between pacing a room and pushing a guest.

Everything above describes moving people around a clock, and it is very easy to read that as treating guests as freight. Some restaurants do exactly that, and you can feel it the moment you sit down: the greeting is fast and slightly clipped, the water is poured before you settle, the server appears at ninety seconds, and there is a bill on the table before you have decided about dessert. Guests do not always name it, but they register it, and it costs you the second visit — which Chapter 23 established is where the business actually lives.

The distinction that keeps you honest is this. Pacing the room is a set of decisions the restaurant makes about itself: how many tables to seat in a given ten minutes, which section takes the next party, when to fire a course, how fast to reset. Pushing a guest is a decision the restaurant makes about them. The first is invisible and worth a great deal of money. The second is visible and costs more than it makes.

Almost everything valuable in duration management lives on the invisible side: seating a five-top at 5:40 rather than 7:40 because the host offered them 5:40 first and made it sound good; firing the entrées when the appetizer plates clear rather than four minutes later; a table reset that takes three minutes instead of nine. Nobody at the table experiences any of that as pressure. All of it moves the clock.

24.4 Table mix: two-tops, four-tops, and the cost of seating two at a four

Table mix is the composition of your seating inventory by table size. It is decided once — at build-out, when the millwork and furniture are ordered — and then it constrains every service you will ever run, which makes it one of the highest-leverage and least-examined decisions in opening a restaurant.

Bellwether's dining room is 56 seats in 17 tables: eight two-tops, seven four-tops, and two six-tops. The bar carries the other 12 seats.

The question nobody asks before signing the furniture order is: does that inventory match the parties who will actually walk in?

Party-size demand against table inventory

Party-size distribution is one of the most stable things about a restaurant, and for a neighborhood chef-driven American room it looks broadly like this. These are Bellwether's planning assumptions, constructed but typical:

Twenty-nine parties averaging just under 2.8 guests fills 80 of the plan's 95 nightly covers; the other 15 land at the bar. Now set that demand against the inventory. Seventeen tables absorbing twenty-nine seatings is $29 \div 17 = 1.71$ seatings per table per night.

FIGURE 24.6 — Party size against table inventory, average dinner   [the Bellwether plan]

  PARTIES WANTING           TABLE INVENTORY (17 tables × 1.71 seatings)      GAP
  ┌───────────────────┬──────────────────────────────────────────┬─────────────┐
  │ 17 seatings for   │  8 two-tops  × 1.71 = 13.7 seatings      │ SHORT  3.3  │
  │   1–2 guests      │                                          │             │
  ├───────────────────┼──────────────────────────────────────────┼─────────────┤
  │ 10 seatings for   │  7 four-tops × 1.71 = 12.0 seatings      │ LONG   2.0  │
  │   3–4 guests      │                                          │             │
  ├───────────────────┼──────────────────────────────────────────┼─────────────┤
  │  2 seatings for   │  2 six-tops  × 1.71 =  3.4 seatings      │ LONG   1.4  │
  │   5–6 guests      │                                          │             │
  └───────────────────┴──────────────────────────────────────────┴─────────────┘
   29 seatings a night in the dining room = 80 covers.
   Plus 15 covers at the 12-seat bar = 95, the plan's nightly figure.

The room is short two-tops and long everything else, which is the near-universal condition of American full-service dining rooms, because rooms get designed around the parties operators imagine — celebratory fours and sixes — and filled by the parties who actually go out to dinner, which are twos.

Three-point-three unmet deuce seatings a night have to go somewhere. Three places, in descending order of intelligence: the bar, a pair of combined two-tops, or a four-top.

What the four-top actually costs

⚠️ Where the Money Leaks

Two people at a four-top, priced.

A four-top seated with two guests sells two seats and kills two. For the length of the table cycle — 95 minutes of dining plus a 10-minute reset, so 1.75 hours — two seats are physically unsellable.

$$2 \text{ seats} \times 1.75 \text{ hours} = 3.5 \text{ destroyed seat-hours}$$

What is that worth? It depends entirely on whether anyone wanted them, and this is the honest part that most treatments of table mix skip.

• Tuesday, 7:10 p.m. The room is 29% full and there is no wait. The two dead seats had no buyer. Cost: zero. Seat the deuce at the four-top, give them the nice one by the window, and don't think about it again.
• Saturday, 7:10 p.m. There is a wait and a quoted time. Those two seats had a buyer. Cost: 2 covers × \$46 = \$92 of revenue that could not be sold, and \$66.46 of contribution at Bellwether's 27.8% blended cost of goods.

Now annualize the version that costs money. Assume three deuces a night are seated at four-tops, and treat only Friday and Saturday as nights where the seats had a buyer:

$$3 \text{ mis-seats} \times 2 \text{ seats} \times \$46 = \$276 \text{ per night}$$ $$\$276 \times 2 \text{ nights} \times 52 \text{ weeks} = \mathbf{\$28{,}704 \text{ a year}}$$

That is a bigger number than the takeout line in Bellwether's revenue plan, produced entirely by a host stand making a reasonable-looking decision six times a weekend. And it is invisible: it never appears on a P&L, because revenue you did not earn is not a line item anywhere.

Four countermeasures, in order of cost.

One: specify combinable two-tops, at the furniture order. Square 30-inch two-tops at a single uniform height, with aisle clearance planned so that two of them push together into a four. This costs nothing extra at build-out (Chapter 7) and is nearly impossible to retrofit into a room with mismatched pedestal bases and banquettes. Eight combinable two-tops give you eight deuces or four fours or any mix, which is the single largest flexibility gain available to a small dining room.

Two: use the bar as deuce and single inventory. Bellwether's twelve bar seats carry 15 covers a night — 1.25 turns, which is soft. Bar seats turn faster than tables because bar guests order differently, sit shorter, and often did not plan to be there. Getting the bar from 15 to 20 covers on Friday and Saturday alone is $5 \times 2 \times 52 = 520$ covers a year, \$23,920 at the plan's check, and it simultaneously relieves the two-top shortage at exactly the hours the shortage bites. That is a Chapter 15 and 16 conversation about what the bar program is for, but the capacity argument belongs here.

Three: give the host a release rule with a time on it. "Hold four-tops for parties of three or more until 7:45; after that, seat them." A rule without a clock is not a rule, it is a negotiation that the busiest person in the building has to conduct every eleven minutes.

Four: do not overcorrect. A room built entirely of two-tops has no six-top business at all, and a six-top is 6 covers × \$46 = **\$276 in a single seating** against a two-top's \$92, usually with a better beverage attachment because larger parties order bottles. Bellwether's two six-tops are under-utilized on an average Tuesday and are the most valuable tables in the building on a Saturday and every December. Keep them.

👨‍🍳 On the Line

The host stand at 7:10 on a Saturday, which is the busiest job in the building.

Here is what is actually happening in the seven-ten minute. Table 6 (a four-top) has finished dessert and is looking at the check. Table 11 (a two-top) has just been dropped entrées, so it is gone until at least 8:05. There is a party of two at the podium who booked three weeks ago, a walk-in party of four who have been quoted forty minutes and are twenty-five minutes into it, and a party of three who just walked in with no reservation and a hopeful expression. Two bar seats are open. The 7:15 book has a four-top and a deuce arriving.

The host has about eleven seconds and imperfect information: table 6 might pay in three minutes or might order coffee and sit for twenty.

The wrong move — and it is the natural move, because it makes the person standing in front of you happy right now — is to put the booked deuce on table 6 the moment it clears, because they are here, they are on time, and it feels rude to make them wait while you hold a table. Cost: \$92, and the party of four is still standing there when the 7:15 four-top walks in, at which point you have two four-parties and no four-top.

The right move is nearly always: deuces to the bar or to a two-top, and if neither exists, ask. "I have a table for you right now at the bar, or about twelve minutes for a table in the room — what would you like?" Guests choose the bar far more often than hosts expect, particularly if the offer is made as an offer and not as a consolation.

The failure mode of this discipline is worth naming too. A host who is holding four-tops aggressively at 9:15, when the wait is gone and there is no four-party coming, is destroying seat-hours to protect a rule. That is why the release rule needs a clock on it. Table mix discipline is a peak-hour tool. After the peak it is just stubbornness.

24.5 Duration management: menu design, coursing, and pacing as revenue tools

Return to the decomposition:

$$\text{RevPASH} = \text{utilization} \times \frac{\text{average check}}{\text{average dine time}}$$

Dine time sits in a denominator, which makes it look like the easiest lever in the equation. Shave ten minutes off a ninety-five minute dinner and the revenue per occupied seat-hour rises from \$29.05 to \$32.47 — an 11.8% gain, apparently for free.

$$\$46.00 \div (85 \div 60) = \$46.00 \div 1.4167 = \$32.47$$

It is not free, and at Bellwether it is mostly not even real. But the tools are worth knowing precisely, because there are restaurants and hours where they are worth a great deal.

Where the time actually goes

A ninety-five minute dinner is not one block. It is roughly:

FIGURE 24.7 — Anatomy of a 95-minute dinner, and where minutes hide  [constructed]

  seated → greeted            2 min   ░░                controllable, cheap
  greeted → drink order       4 min   ░░░░              controllable, cheap
  drink order → drinks        6 min   ░░░░░░            bar throughput
  drinks → food order        11 min   ░░░░░░░░░░░       MENU LENGTH lives here
  order → appetizer          12 min   ░░░░░░░░░░░░      kitchen
  appetizer eaten            14 min   ░░░░░░░░░░░░░░    guest
  cleared → entrée fired      3 min   ░░░                THE HANDSHAKE (Ch. 22)
  entrée fired → delivered   13 min   ░░░░░░░░░░░░░     HEARTH — hard floor
  entrée eaten               18 min   ░░░░░░░░░░░░░░░░░░ guest
  cleared → dessert decision  5 min   ░░░░░
  dessert / coffee           —        (about half of tables)
  check dropped → paid        7 min   ░░░░░░░           controllable, cheap
  ───────────────────────────────────────────────────────
  ≈ 95 minutes seated  +  10 minutes reset  =  105-minute table cycle

  Guest-controlled time (eating, deciding, talking):     ~44 min
  Kitchen-controlled time:                               ~25 min
  Service-controlled time (greet, order, fire, check):   ~26 min

Now you can see which minutes are actually available.

The forty-four guest minutes are not yours. You may not shorten them, and any attempt to is felt immediately. This is where restaurants destroy their own hospitality trying to make a number.

The twenty-five kitchen minutes are the hearth's, and Chapter 14 has already told you the answer. The fire takes what it takes. You can reduce ticket time by simplifying the menu or moving items off the hearth — genuinely available, and a menu-engineering conversation (Chapter 12) — but you cannot speed the fire up.

The twenty-six service minutes are yours, and they are where nearly all honest duration savings live. Two minutes off the greet. Three minutes off the drink-order-to-drinks leg because the bartender has a well-designed station. Three minutes off the fire handshake because the server clears appetizers and rings "fire" in the same trip instead of two trips (Chapter 22's FOH/BOH handshake). Four minutes off the pay cycle because the check goes down with the coffee and the payment happens at the table instead of in two round trips to a station. That is a plausible twelve minutes, none of which the guest experiences as being rushed, and all of it is training and floor design rather than pressure.

And the reset is yours. Ten minutes to reset a table is not physics; it is whether a bus tub, clean linen, and a set-up caddy are within fifteen feet of the table. Cutting reset from ten minutes to five is five minutes of table cycle on every seating, which at 29 seatings a night is 145 table-minutes — the equivalent of roughly one and a half additional table turns you did not have to sell.

Menu design as a duration tool

The eleven minutes between drinks and the food order are the most under-managed minutes in a restaurant, and they are a menu length problem more than a service problem. A twelve-item entrée list is decided faster than a twenty-six-item list, and it is decided far faster than a twenty-six-item list with four inserts and a specials recitation. Bellwether's short seasonal menu is usually justified on food-cost and freshness grounds. It is also worth minutes, and minutes are worth money.

Related tools, honestly rated:

Note the tasting-menu row. Duration management is not a synonym for shorter. A restaurant whose constraint is check average and not seats should be lengthening dine time, not compressing it — a tasting menu at \$135 across two and a half hours produces $\$135 \div 2.5 = \$54$ per occupied seat-hour against Bellwether's \$29.05, and it does it with one seating a night. The lever is the ratio, not the clock.

The limit that matters at Bellwether: you cannot manage duration past the fire

Here is the part that decides whether any of this is worth doing.

Duration savings only convert into revenue where the seat is the binding constraint — that is, where there is somebody waiting for the table you just freed. Everywhere else, a shorter dine time produces an emptier room, which is worth nothing.

Ask where Bellwether's binding constraint actually is:

There is no service at Bellwether where the seat is the binding constraint. On three nights the constraint is demand and a faster table just empties the room sooner. On two nights the constraint is the hearth, and freeing a table at 7:50 hands you a party the kitchen cannot cook for. Chapter 14 was blunt about this and it deserves repeating in revenue terms: wave-splitting creates zero capacity. Neither does a faster reset, if the reason the next party cannot be seated is that there are already thirty-two items on a fire rated for twenty-eight.

So what is duration management worth at Bellwether? Two specific things, and not the thing it looks like it is worth.

First, it is worth the shoulder hours. Cutting twelve minutes off the service legs makes a 5:15 seating genuinely finishable by 6:40, which makes it sellable to a guest who wants to be somewhere at seven — and that seating is in an hour where the hearth is at eleven items and idling. Duration savings are worth money precisely where capacity is free, which is the opposite of where operators usually try to spend them.

Second, it is worth guest experience. Four minutes shaved off the check-drop-to-door leg is not revenue; it is the last ninety seconds of the visit (Chapter 22), and Chapter 23 already established what the last ninety seconds is worth in second visits.

🧮 Run the Numbers

The prix fixe, costed against the Hearth Chicken.

Bellwether's frozen cost card: the Hearth Chicken is half a 3.5 lb air-chilled bird at \$3.20/lb (\$5.60), roasted roots \$0.95, salsa verde \$1.05, butter and aromatics \$0.42, oil and salt \$0.18, garnish \$0.15 — \$8.35**, plus 2% waste = **\$8.52, on the menu at \$29.00. A 29.4% food cost and a \$20.48 contribution margin. It is the book's Star.

Now build an early prix fixe on it. Three courses at \$34: a seasonal salad, the Hearth Chicken at a quarter bird instead of a half, and a dessert. The change to the cost card is explicit:

Component	Baseline (½ bird)	Prix fixe (¼ bird)
Chicken	\$5.60 | **\$2.80** (0.875 lb × \$3.20)
Roasted roots	\$0.95 | \$0.95
Salsa verde	\$1.05 | \$1.05
Butter, aromatics	\$0.42 | \$0.42
Oil, salt, misc	\$0.18 | \$0.18
Garnish	\$0.15 | \$0.15
Subtotal	\$8.35 | **\$5.55**
+2% waste	\$8.52 | **\$5.66**

Add a salad at \$1.85 and a dessert at \$1.30: plate cost \$8.81 for three courses. At \$34 that is a 25.9% food cost — better than the à la carte chicken.

Now compare the two covers properly, which means including beverage, because the \$46 check is a check and not a food price. At the plan's 72/28 mix and 30%/22% cost targets, a \$46 à la carte cover carries \$12.77 of cost of goods (\$33.12 food at 30% = \$9.94; \$12.88 beverage at 22% = \$2.83) — 27.8% blended, exactly the plan — and contributes **\$33.23**.

The prix fixe guest is eating at 5:15 and drinks less. Assume \$6 of beverage:

	À la carte cover	Early prix fixe cover
Food	\$33.12 | \$34.00
Beverage	\$12.88 | \$6.00
Total check	\$46.00** | **\$40.00
Cost of goods	\$12.77 | \$10.13
Contribution	\$33.23** | **\$29.87
Dine time	95 min (1.583 hr)	75 min (1.25 hr)
Contribution per occupied seat-hour	\$20.99** | **\$23.90

The prix fixe cover is worth \$3.36 less in dollars and \$2.91 more per hour. That is the entire economics of a prix fixe in one row, and it explains both why it is a real revenue-management instrument and why it is dangerous.

It is an instrument because at 5:15 on a Tuesday, the seat-hour is otherwise worth zero, and \$29.87 of contribution against zero is not a close call.

It is dangerous because at 7:30 on a Saturday the comparison is \$23.90 against \$20.99 — the prix fixe still wins on the hourly measure — and if you leave it available, your Saturday guests will order it, your check average will fall from \$46 to \$40, and the twenty minutes you saved will convert into exactly nothing, because the hearth is already firing thirty-two items an hour and cannot cook for the party you just made room for. You will have traded \$6.00 of check for a turn you cannot use.

Which is why the next section is about fences.

24.6 Demand-based pricing: what works, what backfires, and the fairness problem

Demand-based pricing is charging different prices for the same or similar product depending on when it is bought, by whom, or through which channel, in order to shift demand toward capacity you have and away from capacity you don't.

Restaurants have always done this. Happy hour is demand-based pricing. The early-bird special is demand-based pricing. Brunch at \$24 and dinner at \$46 in the same room with much of the same equipment is demand-based pricing. Restaurant Week is demand-based pricing with a marketing budget. Nobody objects to any of it. Which should tell you that the technique is not the problem — the framing is.

The one asymmetry that decides everything

Consider two policies with identical economics:

FIGURE 24.8 — The same $12 spread, framed two ways        [constructed teaching example]

  POLICY A — "PEAK SURCHARGE"          POLICY B — "EARLY MENU"
  Fri/Sat 6:30–8:30 seatings           Tue/Wed 5:00–6:15 seatings
  pay $12 more per cover               pay $12 less per cover

     effective check:                     effective check:
       peak      $58                        peak      $46
       off-peak  $46                        off-peak  $34
     spread      $12                      spread      $12

  Financially: the same instrument with the base price moved.
  In the dining room: one of these is a gift and the other one is a news story.

This is not a subtlety. It is the most reliably observed fact about restaurant pricing, and it has a public record behind it. Firms across food service that have publicly floated raising prices at peak times have generally met sharp consumer reaction and walked the framing back; firms that discount slow periods have done so for decades with no reaction whatsoever. The mechanism is loss aversion, and it does not care that your two policies are algebraically the same.

The operating rule that follows: discount the trough, never surcharge the peak. Set your menu price at what you intend the peak to pay, and build every demand-shifting instrument as a reduction from it. You give up nothing financially. You avoid the entire fight.

What works in restaurants

Rated honestly, from most to least reliable:

Fencing — the actual technical idea

A fence is a restriction that prevents a guest who would have paid the full price from accessing the discounted price. It is the difference between revenue management and a price cut, and it is the thing most restaurants get wrong.

Airlines fence with advance purchase and change penalties. Hotels fence with non-refundable rates and minimum stays. Restaurants have three fences and only three:

1. Time fence. The price exists only for seatings within a stated window — 5:00 to 6:15, or after 9:15. The guest who wants 7:30 cannot have it.
2. Day fence. The price exists only on stated days — Tuesday and Wednesday. The guest who wants Saturday cannot have it.
3. Product fence. The price attaches to a different product — three fixed courses, a smaller portion, a limited list. The guest who wants the full menu cannot have it.

Bellwether's early menu will use all three at once, which is standard and correct: three courses (product), Tuesday and Wednesday only (day), 5:00 to 6:15 seatings only (time). A guest who wants the Hearth Chicken at full size on a Saturday at 7:30 has no path to \$34, which is the entire point.

⚠️ Where the Money Leaks

The unfenced discount.

This is the most common and most expensive pricing mistake independent restaurants make, and it almost always arrives disguised as generosity.

An operator with a soft Tuesday runs a promotion: 20% off entrées. No time restriction, no day restriction beyond "this month," no product restriction. It is posted on social media on a Monday.

What happens is not that Tuesday fills. What happens is that the people who were already coming on Thursday, Friday, and Saturday — the ones who follow the restaurant on social media, because followers are your regulars — take 20% off. Bellwether's Friday and Saturday alone are 243 covers a week. Twenty percent off the food portion of a \$46 check is \$6.62 a cover:

$$243 \text{ covers} \times \$33.12 \text{ food} \times 20\% = \$1{,}610 \text{ a week}$$

About \$6,970 over a month, handed to guests who would have paid full price — and that is only the two nights that were already full. Add Thursday and it is over \$9,000. In exchange, the promotion produced whatever incremental Tuesday traffic it produced, which in my experience is close to nothing, because people who don't go out on Tuesday are not price-sensitive; they are busy.

An unfenced discount is not revenue management. It is a price cut you did not decide to make. If the same operator had run "three courses, \$34, Tuesday and Wednesday, 5:00–6:15 seatings," the Saturday regular has no access to it, and every dollar of discount is spent buying a cover that did not exist.

Where it backfires, and the fairness problem

I am not going to resolve this one cleanly, because it does not resolve cleanly.

The efficiency argument is genuinely strong. Capacity is scarce at 7:30 on Saturday and free at 5:15 on Tuesday. Charging the same for both is itself a pricing decision, and not an obviously fair one: it means the Tuesday guest subsidizes the Saturday guest, since the restaurant's prices must average out across a week in which one night carries the fixed costs of another. Variable pricing just makes an existing cross-subsidy visible. And restaurants already do it — nobody calls happy hour exploitative.

The counter-argument is also genuinely strong, and it is not sentimental. Dinner is not a seat on a plane. It is where people mark things: an anniversary, a birthday, a promotion, a last night before someone moves. Those occasions cluster on Friday and Saturday evening, and they cluster there because that is when working people are free. A peak-pricing scheme therefore charges the most to the guests with the least schedule flexibility, and schedule flexibility is not randomly distributed across incomes. The retired couple who can eat at 5:15 on a Tuesday gets the cheap price. The two people who both work retail get the expensive one. You can call that efficient. It is harder to call it fair, and the fact that the airline does it is not an argument, because nobody's daughter's graduation dinner is a flight to Denver.

Where I have landed, and I will tell you it is a compromise rather than a solution: the workable test is whether the guest can see the rule and choose the cheaper side of it.

• A posted early menu with a stated window: visible, choosable. Defensible.
• A holiday prix fixe at a higher price than a normal Tuesday: visible, choosable (they can come on a different night). Defensible.
• A surcharge that appears on the check without having appeared on the menu: not visible. Indefensible, and in a growing number of jurisdictions, not lawful.
• A price that varies by who you are — by loyalty-app history, by device, by inferred willingness to pay: visible to nobody, choosable by nobody. Do not do this, and not primarily for legal reasons. Do not do it because the day a guest finds out — and they will, because guests compare checks at the table — you have not lost a transaction, you have lost the relationship, and this book's third theme is that the relationship is the revenue model.

That test does not dissolve the problem. The guest who works Tuesday nights still pays more for being unavailable, and being unavailable is not a moral failing. I do not have a fix for that. What I have is a practical observation that should reassure the profit-maximizers reading this: guests punish pricing they find insulting far out of proportion to the dollars involved. A \$4 peak surcharge can cost you a review, a regular, and a week of arguing on the internet. A \$12 early-menu discount costs you \$12. Even on purely commercial grounds, the conservative position is the correct one.

⚖️ Code and Compliance

Pricing, fees, and disclosure — verify all of this locally.

Revenue-management instruments touch a surprising number of rules, and they vary by state, county, and city. Structure, not advice:

• Advertised-price and mandatory-fee disclosure. A number of U.S. jurisdictions have adopted or are considering rules governing how mandatory service charges, kitchen fees, and surcharges must be disclosed — some require inclusion in the advertised price, some require specific menu language. Assume your jurisdiction has a rule and find out what it is before you print.
• Service charges are not tips. Whether a mandatory charge belongs to the house or must be distributed, and how it interacts with the tip credit and overtime regular-rate calculations under the Fair Labor Standards Act, is a genuinely technical area. Chapter 20 covers the wage side; get it reviewed by counsel before implementing.
• Happy hour and drink pricing. Several states restrict or prohibit certain drink promotions — two-for-ones, unlimited-time pricing, volume discounts on alcohol. Your liquor authority's rules govern; a food discount that would be routine may be unlawful applied to drinks.
• Prepayment, deposits, and tickets. Prepaid dining can raise gift-card, unclaimed-property, and sales-tax-timing questions depending on the state — specifically, when the sale is recognized and when the tax is remitted. Ask your accountant before you sell your first ticket.
• Accessibility. Under the Americans with Disabilities Act, accessible seating must be available across comparable times and price points. A pricing or seating scheme that effectively restricts accessible seats to a narrower or more expensive band is a problem.

None of this is legal advice, all of it changes, and every item above should be verified with a licensed professional in your jurisdiction before money moves.

24.7 No-shows and deposits: policies, platforms, and the guest relationship

A no-show is the purest form of the perishable-inventory problem. You held capacity, refused other buyers for it, and it expired unsold.

Be careful with the numbers here. No-show rates vary enormously by market, by night, by season, by how far in advance the booking was made, and — consistently, in operators' reported experience — by channel, with free third-party booking generally worse than direct booking. Published figures range from low single digits to the low teens as a percentage of booked covers, and I would not attach a decimal to any of them. What matters is that you measure your own, weekly, from the reservation system, because it is one of the few operating numbers you can measure exactly and almost nobody does.

The arithmetic, and why it is smaller than people think

🧮 Run the Numbers

What no-shows actually cost Bellwether.

Plan assumptions, all constructed: 60% of dinner covers are booked in advance (the rest are walk-ins and the bar). A 4% no-show rate on booked covers.

$$475 \text{ dinner covers} \times 60\% = 285 \text{ booked covers a week}$$ $$285 \times 4\% = 11.4 \text{ covers a week that do not arrive}$$

Now the part that most treatments skip: a no-show only costs money if the seat cannot be refilled. Bookings skew to the weekend, so call it 7 of the 11.4 on Friday and Saturday and 4.4 on Tuesday through Thursday. On Friday and Saturday there is a wait at the door, and roughly 70% of those seats refill. Tuesday through Thursday there is no wait, and roughly 10% refill.

	No-show covers	Refill rate	Covers actually lost
Fri–Sat	7.0	70%	2.1
Tue–Thu	4.4	10%	4.0
Week	11.4		6.1

$$6 \text{ covers} \times \$46 \times 52 \text{ weeks} = \mathbf{\$14{,}352 \text{ a year}}$$

Fourteen thousand dollars is real money — about 5.5% of the plan's operating profit before debt service — and it is also half of the \$28,704 the same restaurant loses seating deuces at four-tops on Friday and Saturday. That comparison is the reason this section is short and §24.4 is long. No-shows generate more emotion than any other topic in restaurant operations and they are usually not the largest hole in the room.

Notice something else in that table. Two-thirds of the loss is on the slow nights, where the seat could not be refilled — and those are the nights where a deposit policy is hardest to justify, because you desperately want the booking and any friction costs you bookings.

A policy sized to the exposure

The instinct after reading the above is to require a card for everything. Do not. Every friction point in a booking flow costs bookings, and on a Tuesday you have no bookings to spare. Apply the policy where the exposure is, which is exactly two places: large parties and peak prime-time.

Bellwether's no-show and deposit policy (the plan's Year-1 position):

Four design points worth defending.

One: the peak-window policy charges on no-show only, not on cancellation. This is deliberate and it is the single most important line in the table. A cancellation at 4:00 p.m. is a gift — you can resell that table. Penalizing it teaches guests to no-show rather than call, which is precisely backwards. Make cancelling free and easy and one tap; charge only for silence.

Two: the free tools come first. A confirmation the guest has to actively respond to, plus a same-day reminder, is the cheapest no-show reduction available and most operators see a meaningful effect from it. Do that before you do anything punitive.

Three: budget the waivers, and mean it. Every forfeiture you charge is a hospitality event with a person on the other end, some of whom have a genuinely good reason and all of whom will tell somebody. The plan therefore books half of what it could theoretically charge:

$$\approx 124 \text{ chargeable seats a year} \times \$20 = \$2{,}480 \text{ gross opportunity}$$ $$\text{waive roughly half} \rightarrow 62 \text{ seats charged} = \mathbf{\$1{,}240}$$

That \$1,240 is a line in the revenue bridge in this chapter's Business Plan checkpoint, and it is the only line in the entire plan that we are rooting against. If it comes in at zero, the policy worked perfectly and nobody no-showed. Give the manager on duty explicit authority to waive without asking, and require only that they log the reason — because the log is how you find out whether the policy is working or whether your confirmation flow is broken.

Four: it is a policy, which means it is written down, disclosed at the moment of booking, and applied consistently. Disclosed at booking means on the booking screen, in the confirmation, and in the reminder — not in a footer. A charge a guest did not know about is a chargeback, a review, and possibly a compliance problem (see the callout in §24.6).

Platforms, briefly

Reservation platforms offer holds, prepayment, ticketing, waitlist management, and no-show tracking, and their fee structures differ substantially — per-cover fees, subscription tiers, marketplace placement, or combinations. Chapter 26 owns the selection decision. The revenue-management points are three:

• The no-show tracking is worth more than the marketing. A platform that lets you see no-show rate by channel, by night, and by lead time gives you the data to size a policy. Most operators never look at it.
• Price a cover fee against the covers it actually adds, not against the covers it processes. A platform that charges per cover on a booking your own website would have taken is a cost, not an acquisition.
• Own your direct channel. Direct bookings have better no-show behavior in most operators' experience, cost less, and produce guest data you keep. The marketplace is a supplement.

🔍 Check Your Understanding

1. Why does Bellwether's policy charge for a no-show but not for a cancellation, even a late one?
2. A restaurant computes that no-shows cost it \$40,000 a year and imposes a \$25-per-seat deposit on all reservations. Name two things that are likely to go wrong.
3. Using the decomposition from §24.2, which of the three RevPASH levers does a no-show policy move, and which does a prix fixe move?

(1: Because a cancellation returns the inventory in time to resell it, and a no-show does not. Charging for both teaches the guest that calling is punished, which converts cancellations into no-shows and makes the problem worse. 2: Booking volume falls, most sharply on the slow nights where the restaurant can least afford it and where no-shows cost the least; and the restaurant now has to actually charge the cards, which produces disputes, chargebacks, reviews, and a manager spending Saturday night on the phone. A policy sized to the exposure would cover peak prime-time and large parties only. 3: A no-show policy moves capacity utilization. A prix fixe moves the other two — it lowers average check and shortens dine time — with a net gain in revenue per occupied seat-hour and a net loss in dollars per cover.)

24.8 Shoulder hours, prix fixe, and filling the room you already pay rent on

Shoulder hours are the low-demand hours inside your open window — the ones adjacent to the peak, where the doors are unlocked, the crew is paid, the lights are on, and almost nobody is there.

Bellwether's shoulders are the 5:00–6:00 hour every night, the 9:00–10:00 hour on the slow nights, and, functionally, all of Tuesday and Wednesday.

Sizing the prize, honestly

Take the first hour of dinner across the whole week. Occupied seat-hours in the 5:00–6:00 hour run roughly 7 on Tuesday, 8 Wednesday, 9 Thursday, 10 Friday, 10 Saturday — 44 occupied out of 340 available, a utilization of 12.9%, producing about \$1,278 of revenue and a RevPASH of **\$3.76** against the dinner week's \$12.85.

Annualized, that hour is 17,680 seat-hours a year. If it ran at the dinner average instead of at \$3.76:

$$17{,}680 \times (\$12.85 - \$3.76) = \$160{,}711$$

Now let me immediately take most of that away from you, because a chapter that leaves that number standing has lied to you. You will never fill the five-o'clock hour to the dinner average. Nobody does. People do not want to eat at five, the guests who do are a small and largely fixed population, and any cover you do attract at 5:15 pays a lower check by design. A realistic program captures a small single-digit fraction of that \$160,000, and \$160,000 is useful only as a way of saying: this hole is bigger than your annual profit, so a program that recovers ten percent of it is worth building.

What actually fills a shoulder

Four things, in descending order of how well they work for a restaurant like Bellwether.

One: a fenced early menu. Three courses at \$34, Tuesday and Wednesday, 5:00–6:15 seatings only. The economics were worked in §24.5: \$29.87 of contribution per cover against a seat-hour that is otherwise worth zero, at a better contribution-per-hour than the à la carte cover. Build it around the Hearth Chicken at a quarter portion so it advertises the thing you are actually good at, and never, ever let it appear on a Saturday.

Two: a bar program with its own reason to exist at 5:15. Twelve bar seats at 5:00 on a Tuesday are the cheapest inventory in the building — no table to reset, no server section to staff, one bartender already on the clock. A short, sharply-priced early bar menu is the highest-margin shoulder program available and it is the one most chef-driven restaurants under-build because the kitchen finds it uninteresting. Chapters 15 and 16 own the program; the capacity case is here.

Three: reshaping the peak-night arrival curve. Discussed in §24.3. On Friday and Saturday, moving covers out of the 7:00–9:00 window into the shoulders is worth real money because it converts hearth capacity that is currently idle at 5:30 into covers, without asking the fire to exceed twenty-eight items an hour. The tool is the host stand and the reservation grid, not price: when the 7:30 slot is gone, the offer is "I have 5:45 or 9:00" — and 5:45 said warmly, first, with a reason, is accepted far more often than 5:45 said apologetically, second, as a consolation.

Four: events on nights you are dark. Bellwether is closed Sunday evening and all day Monday. Those are 100% unsold seat-hours with the rent already paid. This is Chapter 29's territory and I am not going to price events here — but I am going to give Chapter 29 the pricing floor, because it is a revenue-management calculation and it is the one operators get wrong.

🧮 Run the Numbers

The same \$3,000 buyout, on two different days.

A company calls and offers \$3,000 for a private buyout of the dining room, four hours.

On a Monday, when Bellwether is dark: those 68 seats × 4 hours = 272 seat-hours would otherwise produce nothing at all.

$$\$3{,}000 \div 272 = \$11.03 \text{ RevPASH}$$

That is better than a Tuesday dinner (\$8.39) and better than a brunch (\$9.71), on a day the building is currently earning zero. Take it, subject to the incremental labor and the fact that somebody has to be there.

On a Friday, the same \$3,000 buys out a room that was going to do:

$$120 \text{ covers} \times \$46 = \$5{,}520$$

The event is \$2,520 of revenue worse than an ordinary Friday. In contribution terms — the à la carte night contributes $\$5{,}520 \times 72.2\% = \$3{,}986$; a \$3,000 set-menu event at a 28% food cost contributes \$2,160 — it is **\$1,826 worse**, and that is before you count the regulars you turned away who booked somewhere else on Friday and may stay there.

The floor price for a Friday buyout is therefore not "whatever covers our food cost." It is the RevPASH of the hours it consumes. Bellwether's Friday runs \$16.24 per available seat-hour, so a four-hour Friday buyout of the full room must clear $68 \times 4 \times \$16.24 = \$4{,}417$ before it is even neutral — and in practice you price above that, because you are also selling exclusivity and taking execution risk.

Same event. Same food. Same labor. One is a good week and one is a \$2,520 mistake, and the only thing that distinguishes them is which seat-hours it eats.

👨‍🍳 On the Line

What a five-fifteen seating actually feels like, and why nobody wants to sell it.

The honest reason most restaurants do not work their early hour is not that guests refuse it. It is that the building is not ready and everyone knows it.

At 5:15 the sauté cook is still finishing mise. The dish pit is stacked with prep containers. The bread came out of the oven eleven minutes ago. The floor was mopped at 4:45 and there is one damp patch by the service station. The bartender's ice well was filled at 5:05. The playlist has not been switched from the prep playlist, which is somebody's aggressive hardcore record. And the room is empty, which means the two people you seat at 5:15 are sitting in a large silent space being watched by nine staff.

That last one is the real problem, and it is a hospitality problem rather than an operations problem. An empty room feels like a mistake to the person sitting in it. Guests read emptiness as a verdict.

If you are going to sell the early hour, you have to actually build it:

• Seat the early tables together, in the warmest part of the room, near the bar or the window — never scattered across a dark forty-seat expanse. Three occupied tables clustered read as early; three occupied tables scattered read as failing.
• Set the pass at 4:45, not 5:10. If the early menu is fenced to three courses, the kitchen can pre-stage it in a way it cannot pre-stage the à la carte board.
• Change the music at 4:30 and light the room for evening before the first guest, not after.
• Brief it at pre-shift like a real service, because the early guests are frequently the ones who become regulars — they are local, they are repeat, and they are the only guests who will ever come on a Tuesday in February.

Do those four things and a 5:15 seating stops being a consolation prize. Skip them and no discount in the world will fill it twice.

Prix fixe economics, stated plainly

Prix fixe economics — the trade a fixed-price, fixed-course menu makes — comes down to four effects, and you should be able to name all four before you write one:

1. The check is fixed, which removes variance and makes forecasting exact. You know what 40 prix fixe covers produce to the dollar.
2. The mix is fixed, which means you can buy to a count. Chapter 13's purchasing gets dramatically easier, waste falls, and the ideal-versus-actual variance narrows.
3. The duration compresses — fewer decisions, coursed pacing, no dessert menu negotiation — typically fifteen to twenty minutes on a three-course format.
4. The check falls relative to à la carte, and the amount it falls is the price you pay for the first three.

A prix fixe is worth it when effects one through three are worth more than effect four. At 5:15 on a Tuesday, that is not close: effect four is measured against a seat-hour worth zero. At 7:30 on a Saturday it is a mistake, because effect three buys a turn the hearth cannot cook and effect four is measured against a \$46 cover you already had.

Which is to say: prix fixe is not a menu format. It is a fence with food on it.

🍽️ The Business Plan

Checkpoint 24 of 40 — the Revenue Model, and the bridge closes.

This chapter owes the plan four things: covers by daypart, a RevPASH baseline, a no-show policy, and the closed revenue bridge that has been open since Chapter 1.

Covers by daypart

Note what that adds to the plan: the annual cover count with the patio is 37,740, not 36,140. Chapter 1's figure was the indoor base and every chapter since has quoted it. From here forward the plan carries both.

The RevPASH baseline

Two things in that table are worth saying out loud, because both are counterintuitive and both are correct.

First: adding the patio lowers RevPASH, from \$12.09 to \$11.60, and raises profit. The patio's seat-hours are cheap — no incremental rent, since the plan's occupancy already covers the space — but they sell at \$6.14 blended against the room's \$12.09, because a patio fills only in good weather and never fills completely. If you managed this restaurant to maximize RevPASH you would close the patio, and you would be \$64,800 of revenue poorer — and, after the incremental server and the furniture, meaningfully less profitable. RevPASH is a comparison metric within a fixed capacity. It is not an objective function. Use it to compare a Tuesday to a Saturday, an hour to an hour, this March to last March. Do not use it to decide whether to add capacity — that is a contribution question.

Second: do not put off-premise revenue in the numerator. The plan's headline is \$1,550,000 and its on-premise seat-hours are 127,248. Divide one by the other and you get \$12.18, and that number is a lie, because \$74,440 of the numerator — events, takeout, forfeited deposits — never touched a dining-room seat. A restaurant that grows its takeout business and computes RevPASH on total revenue will watch the metric rise while the dining room does nothing. RevPASH is a dining-room metric. Keep the numerator and the denominator in the same building.

The revenue bridge — and which one the plan picks

Chapter 1 built the bottom-up estimate from seats, turns, check, and days and got \$1,410,760. The plan's headline is \$1,550,000**. The **\$139,240 difference has been open ever since, and Chapter 4 laid out two ways to close it and said the plan must pick one. This chapter picks.

Bridge A — itemized streams (Chapter 4's first version):

Bridge B — turns and covers (Chapter 4's alternative):

  Dinner at 105 covers a night (1.55 turns) instead of 95:
     (105 − 95) = 10 more covers × 5 services × 52 weeks = 2,600 covers
     2,600 × $46                                                 = $119,600
  Brunch at 118 covers a service instead of 110:
     (118 − 110) = 8 more covers × 2 services × 52 weeks =   832 covers
     832 × $24                                                   =  $19,968
  ────────────────────────────────────────────────────────────────────────
  Total                                                          = $139,568
  Against a $139,240 gap, an overshoot of                        =     $328

The plan takes Bridge A, with one change. Here is the reasoning, and I want to give Bridge B its due first, because it is a better-looking argument than it gets credit for.

Bridge B's dollars are better dollars. Every one of them comes from the business the plan already describes — no patio furniture, no heaters, no event sales effort, no takeout packaging, no third channel to manage. And they cost almost nothing to serve: ten more covers on a night the crew is already working carries maybe an incremental server hour, so it contributes something like \$272 a service against the patio's roughly \$219, because the patio needs a body outside. On pure contribution per dollar of revenue, Bridge B wins and it is not especially close.

Three reasons the plan takes Bridge A anyway.

One: Bridge B is not a bridge. It is the contingency, and a plan that spends its contingency in the base case has no base case. This is the decisive argument. Look at what the turns lever is worth as a recovery instrument. One additional dinner cover per night, every night, all year, is $5 \times 52 \times \$46 = \$11{,}960$. So:

Read the bottom row: recovering the entire bridge from covers alone requires 1.57 turns — which is Bridge B. Bridge B is the total-failure recovery plan. If you spend it up front to close the plan, then when the patio has a wet June you have nothing left to say.

Two: Bridge A can be audited line by line; Bridge B cannot be audited at all. Four named streams are four things that can be forecast separately, tracked separately in the POS, and killed separately if they do not perform. "We will do 1.55 turns" is a single undifferentiated assumption, and when it comes in at 1.43 there is nothing to diagnose and nothing to cut — only a hole.

Three: Bridge B's covers are not distributed where the capacity is. Ten more covers a night applied across the week takes Tuesday from 62 to 72 and Saturday from 123 to 133. The Tuesday increment is a pure demand problem — those guests do not currently exist and no seating decision conjures them. The Saturday increment lands on a kitchen that Figure 24.5 already shows firing thirty-two items an hour against a rated twenty-eight, in the exact hours the additional covers would want. And Bridge B overshoots the gap by \$328, which is not a large number but is the wrong sign: a plan whose bridge exceeds its own headline invites the question of why the headline is not \$1,550,328.

The one change to Bridge A: the \$1,240 gets a name. A revenue bridge with a line called "disclosed rounding" is more honest than one that hides a plug inside a tuned assumption — and I want to be plain that tuning an assumption to make a total land exactly does not remove a plug, it conceals it. But we can do better than disclosure. Section 24.7 built a no-show and deposit policy whose expected forfeitures, after waiving roughly half of what could be charged, come to $62 \times \$20 = \$1{,}240$. That is a real mechanism with stated assumptions, which means next March somebody can check it. A plug you can audit is not a plug.

🧾 Read the Numbers

```text FIGURE 24.9 — "The bridge, closed" [the Bellwether plan] THE ARTIFACT The Year-1 revenue reconciliation for Bellwether's business plan: the bottom-up estimate from Chapter 1, the named components that close the gap, and the plan headline. THE CONTEXT Chapter 1 estimated $1,410,760 from 68 seats, 1.40 turns, a $46 dinner check and a $24 brunch check. The plan headline is $1,550,000. Chapter 4 itemized a bridge and left the choice open. This is the choice.

       Chapter 1 bottom-up estimate                        $1,410,760
         + patio dinner    100 svc × 12 cov × $46             +55,200
         + patio brunch     40 svc × 10 cov × $24              +9,600
         + private events   14 × $3,000          [Ch. 29]     +42,000
         + takeout          52 wks × $600        [Ch. 28]     +31,200
         + forfeited deposits  62 seats × $20    [§24.7]       +1,240
         ──────────────────────────────────────────────────────────
         bridge total                                        $139,240
       PLAN HEADLINE, YEAR 1                               $1,550,000

       Owned by this chapter:  patio $64,800 + deposits $1,240 = $66,040 (47%)
       Owned elsewhere:        events $42,000 + takeout $31,200 = $73,200 (53%)

WHAT IT SHOWS A closed bridge in which every line has a mechanism, a unit build, and an owner. It foots to $139,240 exactly, and $1,410,760 + $139,240 = $1,550,000. It also shows the plan's largest single dependency in one line: 53% of the bridge belongs to two business lines Bellwether has never operated. WHAT IT DOESN'T It says nothing about cost or profit. The patio needs furniture, heaters, and an incremental server on patio nights; events need a salesperson's time and set-up labor; takeout needs packaging and, if it runs through a third party, a commission that can reach a quarter of the order. Chapters 28, 29, and 31 carry those costs. Nothing here should be read as $139,240 of profit — at the plan's blended margin it is closer to $23,000 of operating profit before those incremental costs are subtracted. It also does not model weather. "100 patio services" is not 20 calendar weeks × 5 nights; it is roughly 24 calendar weeks at an 83% weather- availability factor, and that factor is the assumption most likely to break. THE DECISION Adopt Bridge A. Track all five lines separately in the POS from week one: patio covers tagged at the table, event revenue in its own department, takeout in its own department, forfeitures logged with a reason. Review monthly against the build, not annually against the total. THE LESSON A revenue bridge is not a total; it is a list of promises. Write it so that each promise can be checked by somebody who was not in the room when you made it — and never close it with the lever you are counting on to save you. ```

FIGURE 24.10 — The bridge as a picture                            [the Bellwether plan]

  Ch.1 bottom-up            $1,410,760  ████████████████████████████████████████████
    + patio dinner             +55,200  ██
    + patio brunch              +9,600  ▎
    + private events  [Ch.29]  +42,000  █▍
    + takeout         [Ch.28]  +31,200  █
    + forfeited deposits        +1,240  ▏
  ──────────────────────────────────────────────────────────────────────────────────
  PLAN HEADLINE             $1,550,000  ████████████████████████████████████████████████

  One block ≈ $32,000. The bridge is 9.0% of the headline; the base is 91.0%.
  That ratio is the reassuring part: the plan does not depend on the new stuff.

The no-show policy

As specified in §24.7: card on file for parties of six or more and for Friday/Saturday 6:00–8:00 seatings; \$20 per seat charged on no-show only for the peak window, and on no-show or inside-48-hour cancellation for large parties; prepaid ticketing for the three major holidays; nothing at all on every other booking; active confirmation at 24 hours and a same-day reminder on everything. Manager may waive without asking and must log the reason.

The unbanked levers — the plan's actual cushion

These are quantified and deliberately not in the bridge, because a plan should be held only to what it banked:

Do not add that column. The levers overlap heavily — the arrival reshaping, the bar covers, and the table-mix discipline are all partly the same covers, and all of them are subsets of the turns lever. The honest summary is the one number at the bottom of the recovery table above: there are several independent routes to about 5.4 more covers a night, and 5.4 more covers a night replaces the patio entirely.

What this checkpoint does not settle

• Whether the patio's 100 weather-available services happen. This is the plan's most fragile revenue assumption and it is a climate question, not a management one.
• Whether events and takeout deliver \$73,200. Fifty-three percent of the bridge belongs to two business lines the partners have never run. Chapters 28 and 29 have to defend those numbers, and if they revise them, this bridge reopens.
• Whether any of this is profitable. This chapter has produced revenue lines only. The patio's incremental labor and equipment, the events' set-up cost, and takeout's packaging and commission all land in Chapters 28, 29, and 31.
• Whether brunch earns its labor. Figure 24.2 shows brunch producing less revenue per service than a Tuesday dinner on nearly twice the covers, which means nearly twice the servers. Chapter 32 owns that question and it is a real one.
• Chapter 1's open question — can this concept produce 95 covers a night on 68 seats? The answer from here is: yes, at 44.2% capacity utilization, which is unremarkable and well within what the room and the hearth can do. The harder question was never 95. It is whether Tuesday can produce 62.

Open questions carried forward:

1. Does the patio deliver 100 usable services in a Midwestern spring and summer? (Chapter 31's seasonal forecast; Chapter 33's cash timing)
2. Can Chapters 28 and 29 defend \$73,200 of bridge, and what does it cost to earn? (Chapters 28, 29)
3. Is brunch a profitable daypart or a habit? (Chapter 32)
4. What does the early prix fixe do to food cost and to the kitchen's prep day? (Chapters 12, 13, 14)
5. Does the reservation platform selected in Chapter 26 actually report no-show rate by channel — and if it doesn't, how will anybody know whether the deposit policy worked?

Conclusion

A restaurant sells two things simultaneously and only counts one of them. It counts plates. What it is actually renting is time in a chair, and the chair keeps producing that inventory whether anybody buys it or not.

Once you can see the seat-hour, the metrics fall into place. Seat turns normalizes covers for the size of the room. RevPASH normalizes for the size of the room and the length of the service, and it is the only common restaurant number denominated in the thing you have a fixed supply of. And RevPASH decomposes into exactly three levers — capacity utilization, average check, and dine time — which means any revenue proposal anyone brings you can be tested by asking which of the three it moves and what it costs in the other two.

Applied to Bellwether, the metric found things a cover count could not. That Saturday brunch does 77% more covers than Tuesday dinner and \$212 less revenue. That the room is 43% empty on its busiest night, while the hearth is over its rate for two hours of it — which means the constraint on a Saturday is the fire, not the seat, and every duration technique aimed at the seat is aimed at the wrong thing. That Tuesday's guests spend the same and sit the same length as Saturday's, so Tuesday is a demand problem and discounting it attacks a variable that is not broken. That three deuces a night at four-tops on the weekend is a \$28,704 line item nobody will ever see on a P&L.

And it closed the bridge. The plan takes the itemized version — patio, events, takeout, and a named deposit line summing to \$139,240 — not because those dollars are better than the turns dollars, but because they are different dollars. The turns lever is what you have left when something goes wrong, and something goes wrong. Five and a half more covers a night replaces the entire patio. Spend that in the base case and there is no answer to a wet June.

There is one more thing to say about the fairness question in §24.6, because it does not go away and you will meet it. Every instrument in this chapter is a way of charging different people different amounts for the same room, and the ones that survive contact with guests are the ones a guest can see coming and decline. That is a commercial test as much as an ethical one, which is convenient, but it is worth noticing that it is not a complete answer. It never was.

Chapter 25 changes register entirely. Everything in Part V so far has been about producing revenue from a room; the next chapter is about the single category of failure that can end all of it before lunch. Food safety is the one area of this business where being pretty good and being negligent are indistinguishable right up until the morning they are not.

Key Terms

RevPASH (revenue per available seat-hour) — total revenue for a period divided by available seats multiplied by hours open. The restaurant metric denominated in perishable capacity; it decomposes into capacity utilization × (average check ÷ average dine time). Bellwether's planned dinner RevPASH is \$12.85. (Ch. 24)

Seat turns — covers divided by seats for a service; the number of times each chair was rented. Normalizes for room size but carries no information about the length of the service window. (Ch. 24)

Capacity utilization — occupied seat-hours divided by available seat-hours. Occupied seat-hours are covers × average dine time in hours. Bellwether's dinner week runs 44.2%. (Ch. 24)

Table mix — the composition of a dining room's seating inventory by table size. Decided at build-out and constraining forever after; the near-universal failure is too few two-tops. (Ch. 24)

Yield management — the discipline of maximizing revenue from fixed, perishable capacity by varying price and availability across customer segments and times. Imported into restaurants from commercial aviation and hotels. (Ch. 24)

Demand-based pricing — charging different prices for the same or similar product by time, day, or channel in order to shift demand toward unsold capacity. Works as a discount on the trough; reliably backfires as a surcharge on the peak. (Ch. 24)

Fencing — a restriction (time, day, or product) that prevents a guest who would have paid full price from accessing a discounted price. An unfenced discount is not revenue management; it is a price cut. (Ch. 24)

No-show and deposit policy — the written, disclosed rules governing card holds, deposits, cancellation windows, and charges for reservations that do not arrive. Should be sized to the actual exposure — large parties and peak prime-time — and should charge for no-shows, never for cancellations. (Ch. 24)

Shoulder hours — the low-demand hours inside an open service window, adjacent to the peak, where fixed costs are being paid against little or no revenue. Bellwether's are 5:00–6:00 nightly and 9:00–10:00 on slow nights. (Ch. 24)

Prix fixe economics — the trade a fixed-price, fixed-course menu makes: it fixes the check, fixes the purchasing mix, and compresses duration, in exchange for a lower check than à la carte. Worth taking where the seat-hour would otherwise be unsold; damaging where it is not fenced. (Ch. 24)

The shape of the week — the stable, repeating distribution of demand across a restaurant's operating days and hours. Highly forecastable after about eight weeks of trading, and the foundation of every revenue, purchasing, and scheduling decision. (Ch. 24)

Spaced Review

1. Compute RevPASH for a 120-seat restaurant open 11:00 a.m. to 10:00 p.m., six days a week, doing \$61,000 of weekly sales. Then compute it for a 40-seat restaurant open 5:30 to 10:00, five nights, doing \$26,000. Which is using its capacity better, and what does the cover count tell you about that?
2. Bellwether's Tuesday and Saturday produce identical revenue per occupied seat-hour (\$29.05 and \$29.02) and RevPASH figures that differ by a factor of two. Explain in one sentence why, and say what that rules out as a fix for Tuesday.
3. From Chapter 14: why does wave-splitting create zero capacity, and why is "reshaping the arrival curve" nevertheless a real revenue technique rather than the same idea renamed?
4. From Chapter 11 and this chapter: the Hearth Chicken costs \$8.52 and sells at \$29.00 for a \$20.48 contribution margin. In the \$34 early prix fixe it appears at a quarter portion, costing \$5.66 inside an \$8.81 three-course plate cost. Why is contribution margin per cover the wrong comparison between the two, and what is the right one?
5. From Chapter 19 and this chapter: the plan budgets Tuesday's labor at 32.3% of Tuesday's sales, or \$921. Explain why that is both arithmetically correct and operationally wrong, and then explain why the same fact makes an incremental Tuesday cover more valuable than an incremental Friday cover.
6. The recurring question: an operator tells you their covers are up 8% year over year and they are delighted. What are the first two numbers you ask for, and what would make you conclude the restaurant is actually worse off?