Food cost for Indian restaurants: the formula, the GST nuance, and channel-level costing (2026)

The standard formula:

Food cost % = (cost of food consumed ÷ net food sales) × 100

where cost of food consumed = opening stock + purchases − closing stock for the period. Three places this goes wrong in practice:

• GST-inclusive sales in the denominator. A ₹300 menu item billed at 5% GST brings in ₹285.71 of revenue. Use net-of-GST sales, or your food cost reads ~1.4 points lower than reality at 5% — and much worse at 18%.
• Purchases instead of consumption in the numerator.A big Sunday vegetable buy lands in this week’s cost even though half of it is next week’s stock. Without opening/closing stock counts you are measuring purchasing cadence, not food cost.
• Untracked staff meals and wastage quietly inflating the number until someone blames the kitchen for a purchasing problem (or vice versa).

Getting consumption right requires stock data — which is why food cost is fundamentally an inventory problem, not a spreadsheet problem.

Generic (mostly Western, sometimes Indian) guides tell you to cost ingredients net of tax because “you claim the input credit”. In India that advice is wrong for the majority of restaurants:

• Standalone restaurants at the 5% concessional rate cannot claim ITC — the rate is conditioned on foregoing it. GST paid on ingredients, packaging, gas, cleaning supplies and aggregator commission is an unrecoverable cost, so it belongs in your food cost (and overhead) numbers.
• Restaurants at 18% with ITC (broadly, in specified hotel premises) recover input GST and should cost net of GST.

The two regimes give materially different recipe costs from identical purchase invoices, so your costing system must know which regime the entity is in. Note the asymmetry on the sales side: the denominator is always net of GST (that tax was never your revenue), even while the numerator may be GST-inclusive. For the full input-credit mechanics see the Complete Guide to GST in India.

Period-level food cost tells you that something is off; recipe costing tells you where. Per dish:

1. List every ingredient at recipe quantity — including oil, masala, garnish and packaging for delivery formats.
2. Apply yield percentages. A kilo of raw onion is not a kilo of usable onion — peeling, trimming and cooking losses mean the effective rate per usable gram is higher than the invoice rate. Yield-adjusted rates are where most recipe costings are 10-15% too optimistic.
3. Cost at current purchase rates, ideally pulled from your latest purchase bills rather than a rate card someone typed last Diwali.
4. Divide by portions producedto get plate cost, and plate cost ÷ net selling price = the dish’s theoretical food cost percentage.

When the POS tracks ingredient stock live and your purchase bills feed the same system, plate costs stay current as rates move instead of decaying in a rate card — the approach behind live stock in HelloBooks POS, with wastage tracking closing the loop.

Multiply each dish’s recipe consumption by the quantity actually sold (your POS item report) and you get theoretical consumption — what the kitchen should have used. Compare it to actual consumption from stock counts:

• Variance under ~2 points of food cost: normal operating noise.
• Persistent 3-6 point gaps: over-portioning, prep wastage, staff meals leaking into cost, unbilled items, or recipe drift (the cook’s ladle grew).
• Sudden spikes: usually a stock-count error, a missed purchase entry — or shrinkage.

This comparison is only possible if sales, recipes and inventory live in connected systems. When they also share ledgers with your accounting, the food-cost line in the P&L reconciles to the same data — no month-end argument between the chef’s spreadsheet and the accountant’s books. That is the core argument for an accounting-native POS over a standalone one — see how HelloBooks compares to Petpooja.

The same paneer tikka sold at the table and on Zomato has the same plate cost but very different order economics: the aggregator order adds commission (plus 18% GST on it, usually non-creditable), delivery packaging, and platform promotions. Many operators price aggregator menus 15-25% higher for exactly this reason.

Treat channel as a first-class dimension: tag every order dine-in / takeaway / aggregator, cost packaging into delivery recipes, and read food cost and contribution margin per channel. The settlement side of aggregator orders has its own traps (Section 9(5) GST, TDS under 194-O) — covered in the companion guide on Swiggy & Zomato reconciliation.

• Full-service restaurants: typically 28-35%.
• QSR and cloud kitchens: typically 25-30% — see the specifics for a delivery-first format in a cloud kitchen setup.
• Cafés and beverage-led formats: often lower on beverages, blended up by food.

Treat these as orientation, not law: menu mix, positioning and city rents change what “healthy” means. The discipline that actually protects margin is (a) net-of-GST math, (b) regime-correct ingredient costing, (c) a monthly theoretical-vs-actual variance review, and (d) channel-level reads. For the restaurant-specific books behind all four, see restaurant accounting in HelloBooks.