F&B Cost Control: Recipe Cards, Portions & Menu Engineering
The menu is a price list for recipes nobody costed — recipe cards, portion discipline, waste as a transaction, and menu engineering: the four habits that hold food cost where the budget says.
Most Kenyan restaurants price by market: the burger costs what the competitor's burger costs, plus or minus ambition. What almost nobody knows is what the burger costs to make — this week, at this supplier's prices, at the portions the kitchen actually serves. Without that number, "food cost control" is a monthly autopsy. With it, every dish on the menu is a small business whose profit you can read.
Habit 1: recipe cards with live costs
- Every menu item gets a recipe card: ingredients, exact quantities, yield — the BOM discipline, plated.
- Ingredient costs update from purchasing automatically — when tomatoes double in the rains, every recipe using them re-costs itself the same day.
- The output is a live margin per dish: recipe cost vs menu price, ranked. The ranking is where every pricing and menu decision starts.
- Prep items (stocks, sauces, marinades) are sub-recipes — costed once, consumed by the dishes that use them.
Habit 2: portion discipline
The recipe card says 180g of beef; the ladle says whatever the cook's mood says. Portion drift of 10–15% is invisible on the plate and lethal on the month:
- Portioning tools where they matter — scoops, ladles, scales at the protein station — cost a few thousand shillings and repay weekly.
- Spot-weigh plated dishes against cards twice a week, publicly and blamelessly; measurement itself corrects most drift.
- Pre-portion the expensive proteins at prep time (butchered, weighed, bagged) so service-time speed never fights portion accuracy.
- Track yield on butchering and prep: the whole chicken → portions math is a yield reconciliation the kitchen should win, not wonder about.
Habit 3: waste is a transaction
| Waste type | Hides as | Record it as |
|---|---|---|
| Spoilage (produce, dairy) | General food cost | Waste entry with item, quantity, reason — trends expose ordering errors |
| Prep waste beyond yield | The kitchen's secret | Yield variance per prep batch |
| Returned plates & comps | Lost revenue nobody logs | Void/comp reasons on the POS, reviewed weekly |
| Staff meals | A perk that eats 2–3 points | A defined staff-meal allowance, issued and costed like any outlet |
The cost of "it's just a little"
A restaurant doing KES 3M monthly at 32% food cost spends ~KES 960,000 on food. Two points of drift — portions, waste, staff meals — is KES 60,000 a month, quietly. That is a salary, disappearing into generous ladles and unlogged spoilage.
Habit 4: menu engineering
With live recipe costs and POS sales data, every item lands in one of four boxes — high margin/high volume (stars: protect them), high margin/low volume (puzzles: promote or reposition), low margin/high volume (workhorses: re-cost, re-portion, or re-price), low margin/low volume (dogs: cut them). Reviewing the grid quarterly is how menus earn more without a single price increase the customer notices — portion architecture, plate composition, and placement do the work.
All four habits assume the plumbing underneath: store-to-kitchen issues, daily counts, a POS that consumes recipes, and purchasing that tracks ingredient prices. The bar runs the same play with tighter tolerances.
What AWRA OpsHub does today
- Waste and spoilage recorded as adjustments with a reason drawn from a configured catalogue and a named user.
- Live ingredient costs — weighted average cost per item, updated as goods are received, with landed cost included.
- Stock counted per location with blind entry, so kitchen and store variance is measurable at the item level.
- Purchase price history per supplier, so ingredient cost movement is visible.
What it does not do
- No recipe cards. There is no recipe, dish or bill-of-materials entity, so a plate cost cannot be assembled from ingredients.
- No theoretical usage and therefore no portion variance. Without a recipe there is nothing to compare actual consumption against.
- No menu-engineering grid. Popularity-versus-margin analysis is not produced, and could not be without plate costs.
- No yield or trim factors on raw ingredients.
Most of what this article teaches needs a recipe layer, and we do not have one. What we give you is the honest input side: accurate ingredient costs, counted stock per location, and waste captured with a reason. Plate costing itself is a spreadsheet exercise today — and worth doing, because the ingredient costs feeding it will at least be real.
Cost the burger, then the menu
Live ingredient costs with [landed cost](/glossary/landed-cost) included, waste captured with a reason, and stock counted per outlet — the accurate inputs a plate cost needs.
See F&B cost control in AWRAFrequently asked questions
How many recipe cards do we need before this works?
Start with the twenty items that drive 80% of sales — costing them takes a few kitchen afternoons and covers most of the money. The long tail follows at a card or two a week. What fails is trying to cost 200 items before turning anything on. Note that the cards themselves live in your spreadsheet, not in the product: nothing here has to be configured before you can start, and nothing you build is locked into us.
Ingredient prices change weekly at the market. How do cards stay current?
Half of that is automatic and half is not, so be clear which. Purchase prices do flow from receiving into ingredient costs, so the per-kilo and per-litre figures you cost from are always current without anyone retyping them. The re-costing of the dish is not automatic, because the system holds no recipe to re-cost — you rebuild the card from the current ingredient costs, which for twenty core dishes is a short monthly job rather than a research project. The volatility is still the argument for the system; it is just an argument about the input prices, not the cards.
Is cutting staff meals really worth the morale cost?
Don't cut them — define them. A costed staff-meal allowance (what, when, budgeted value) is a benefit everyone understands; an undefined "kitchen eats" policy is 2–3 food-cost points with no owner and endless suspicion. Definition is the win, not deprivation.
What margin should individual dishes make?
Think in shillings and mix, not just percentages: a 60%-margin soup earning KES 180 matters less than a 45%-margin platter earning KES 520. The menu-engineering grid weighs both — margin per plate and plates per week — which is why it beats a single food-cost target applied to everything.
