Chapter 28 — Key Takeaways

The core claims

  1. Off-premise is a second business line with a different margin structure that happens to share your kitchen. Price it as one. It is not a marketing channel and it is not "extra covers."

  2. Incremental revenue absorbs fixed cost only when the resource it consumes was going to sit idle. The moment the channel competes for a constrained resource, it stops adding and starts substituting — at roughly half the rate.

  3. The biggest gap between a dine-in dollar and a delivery dollar is not the commission. It is the beverage nobody ordered. Chapter 15's margin lives in the glass, and off-premise sells the low-margin half of your business.

  4. The commission rate is not the take rate. Promotions, sponsored placement, and refunds charged back can add ten points. Manage the channel from the payout statement, never from the contract.

  5. Cannibalization is measurable, and almost nobody measures it. A channel's entire value swings on it. Compute the break-even cannibalization rate before you launch, not after.

  6. The marketplace is a rational acquisition channel and an irrational fulfillment channel. If you list and do not put a conversion mechanism in every bag, you have chosen the irrational version by default.

  7. Menus for boxes are different from menus for plates, and the difference is physics, not opinion. Crisp things have four minutes. Braises have forty-five.

  8. The answer is rarely yes or no. It is a schedule. Open when the constraint is idle; closed when it is binding.

The formulas

  off_premise_contribution =
      menu_subtotal
      - commission (or processing + platform fee, on first-party)
      - plate cost
      - packaging
      - packing labor
      - error / refund allowance

  effective_take_rate =
      (commission + promotions + ads + refunds charged back + fees) / gross menu sales

  break_even_cannibalization =
      off_premise_contribution_per_order
      / contribution_of_the_dine_in_visit_it_replaces

  contribution_per_constrained_unit =
      contribution produced in the hour / units the constraint produced in the hour

  price_to_hold_contribution =
      (dine_in_price + per_item_packaging_and_labor) / (1 - commission_rate)

  packaging_cost_per_order = base (bag, cutlery, seal, label) + containers x count

The numbers from this chapter worth remembering
Marketplace commission range (illustrative) 15–30%, by service tier
Bellwether's modeled order $65.00 · plate cost $19.32 · packaging $2.24 · packing labor $1.30
Contribution — dine-in, two covers ($92.00) | **$61.76**
Contribution — first-party pickup $38.65 (63% of dine-in)
Contribution — marketplace @ 25% $25.89 (42% of dine-in)
Break-even cannibalization — first-party pickup 55%
Break-even cannibalization — marketplace @ 25% 34%
Hearth at the Saturday binding hour 29 items produced against a 28/hour sustainable rate = 104%
Contribution per hearth item at that hour $52.71
Net effect of one peak-hour delivery order **($26.82)** — while recording $65.00 of sales
Price needed to hold the Hearth Chicken's CM at 25% commission $40.27 (+38.9%)
Dedicated packer, 5 hrs × 5 nights @ $19.50 all-in | **$487.50/week** vs. $357.60 the channel produces
Volume at which a dedicated position starts to pay roughly 5% of sales
Error allowance — first-party / marketplace 1.5% / 3–4% of sales

Bellwether's Off-Premise decision, in one box

The $31,200 line stands. Everything under it changed.

First-party pickup only in year one — no marketplace, no delivery. Closed 6:00–8:45 p.m. Friday and Saturday to protect the hearth. Nine of the twenty-two items; the pasta is off, the Hearth Chicken is on with a documented compromise. Price parity. Labor absorbed, not added, with a 30-orders-a-week trigger to revisit. Contribution $28.82 an order, $17,984 a year — against $11,276** on a 25% marketplace, a **$6,708 difference on identical sales. Cannibalization assumed at 25%, flagged in the assumptions register as $7,800 of possible transfer, not growth. A marketplace listing is deferred to a written year-two test with a blackout control and thresholds set in advance.

What you should be able to do Monday morning

Pull one week of off-premise sales and one payout statement. Compute the effective take rate and the contribution per order after plate cost, packaging, packing labor, and refunds. Identify the constrained resource in your kitchen and its sustainable hourly rate. Chart when off-premise orders arrive against when dine-in demand arrives. Then close the channel during the hours where those two charts collide — and turn off any promotion or sponsored placement nobody explicitly authorized this quarter.