Marlowe & Finch Kitchen tools · Margin
Menu engineering

Where the margin actually is.

Every dish ranked by what it earns and how often it sells — not by how it photographs. Five dishes are quietly working below the floor; here's what to do about each. Read by Eleanor Marlowe, Chef

Dishes on the menu
17
Across snacks, entrées, mains, sides & desserts
Below the margin floor
5
Plate cost above 40% of menu price
Gross profit at risk
$40,800 /yr
Recoverable by acting on the five flags below

The menu, on two axes

Popularity (sell-through) against plate gross profit. Thresholds: median 50% sell-through, median $17 GP.
Plate gross profit →
Stars Plough-horses Puzzles Dogs Pie floater Barramundi Duck Oysters Venison ↑
Sell-through →
Signature dishes (pie floater, barramundi) ringed in brass. Both pull volume — but the barramundi is also our worst margin breach.
How to read it

Stars — high volume, healthy margin. Protect them; never discount. The duck and the pie floater carry the room.

Puzzles — good margin, sells thin. The venison Wellington lives here: $55 a plate but only one table in three orders it. Reposition, don't reprice down.

Plough-horses — popular, slim margin. Mostly sides and puddings doing their job. Leave them be.

Dogs — quiet and slim. Rework or rest them, or keep them honest as a seasonal special.


Every dish, costed

Tap a column to sort. Brass-edged rows breach the margin floor.

Margin floor: plate food cost must stay at or under 40% of menu price (a 2.5× minimum mark-up). The rules engine blocked a draft suggestion to drop the venison to $74 to lift its volume — that breaches the floor, so it never reached this list. Recommendations are ranked for a human; nothing is applied automatically.


Recommendations by gross-profit impact

Ranked. Indicative annual figures, volume held flat.
1

Barramundi en papillote — reprice $44 → $49

It's a signature and a Star, and it's also the worst breach on the list at 54.5% food cost. People seek it out; it'll hold velocity at $49. This is the single biggest leak.

+$17,900per year
2

Smoked ocean trout — reprice $30 → $33

A Puzzle sitting below the floor. A $3 rise is well inside what the dish carries, and it nudges the margin back over the line without touching the kitchen.

+$6,450per year
3

Venison Wellington — reposition, don't discount

$55 a plate is real money, but it only lands at one table in three. Move it up the menu as the centrepiece "for the table" and let the floor sell it — three extra orders a week pays for itself. Holding the $96 price.

+$7,200per year
4

Crocodile boudin blanc — reprice +$3 or rotate off

Lowest-volume main and below the floor. Either take it to $45, or give the slot to something that earns it. A Puzzle that's been a Puzzle long enough.

+$4,300per year
5

Oysters Coffin Bay — leave the price, fix the cost

Below the floor, but it's a market-priced Star — guests know what oysters cost. The lever here is the Coffin Bay supplier price, not the menu. Hand to supplier tracking, don't reprice.

+$4,950per year
Under the hood How the pricing engine works — and where it should go next

Right now this is cost-based pricing with a margin floor: it costs each plate, ranks dishes by contribution and sell-through, and flags anything where food cost runs above 40% of the menu price. That's an honest, defensible floor — but it only knows what a dish costs to make, not what it's worth to a guest. The interesting work is closing that gap, from cost-based to value-based pricing.

On the roadmap

  1. What the street charges. Sample comparable dishes — a barramundi main, a share-to-table roast for two — at other Tanjong Pagar and CBD rooms, so a reprice is anchored to the local market and not just our own cost. Headroom we can see is headroom we can take.
  2. Listen to the reviews. If a dish is consistently called out as exceptional, that's pricing power sitting in plain sight. Feeding per-dish sentiment from the review-triage tool back into this model lets a quietly-loved plate carry a few dollars more without any drama — value-based, evidence-led, not a guess.
  3. Cost the waste, not just the recipe. Two dishes can look identical on paper while one quietly bins half a bunch of herbs or a fish frame every service. Folding yield, trim and spoilage into the true plate cost can flip the verdict — raise the price, re-portion, trim the inventory order, or drop the dish altogether.
  4. Learn the elasticity. Watch how velocity actually responds to past price moves, so the engine recommends the rise a dish can absorb rather than the rise we hope it can.
  5. The guardrails don't move. Every suggestion stays human-approved and rules-checked. The engine can propose a value-based price; it still can't apply one, and it still can't breach the floor.
Full use-case write-up — the problem, benefits and vendor landscape: UC07 — Menu engineering & pricing → Related concept — the review-feedback loop in point 2 lives here: UC04 review-triage mock-up →