A supplier quotes you a price. Is it fair? Should-cost analysis answers that by working out what the product should cost, from its ingredients up, before you look at the quote.
Procurement teams in automotive and electronics have done this for decades. In food it is harder, because most of the cost sits in raw materials that trade on commodity markets and move every week. This guide explains what should-cost analysis is, what goes into a food should-cost, how to find the cost drivers that matter, and how to use the result with suppliers.
What is should-cost analysis?
Should-cost analysis estimates what a product ought to cost, built from its inputs, instead of starting from the supplier’s quote or last year’s invoice. You list everything that goes into one unit, price each part at a fair market rate, add a normal margin, and compare the total with what you are asked to pay.
It goes by several names. Clean-sheet costing stresses that you start from zero and ignore the quote. Cost breakdown analysis describes the output, a line-by-line split of the cost. Teardown analysis is the engineering version, where a product is taken apart and each component is costed. In food, the recipe and the process sheet take the place of the teardown.
There are two ways to build one:
- Bottom-up: cost every input from the recipe. More work, more accurate, and the only way to see what drives the cost.
- Top-down: start from a known cost, such as last year’s price or an industry cost structure, and adjust it for what has moved since. Faster, good enough for a first check, but it inherits whatever was wrong with the starting point.
Most teams start top-down for a quick check and build bottom-up models for their largest or most volatile spend.
Why food should-cost is different
The method is the same as for any product. Three things make it harder in food.
The inputs are commodities. According to Lightsource, the raw materials that make up 60 to 80% of a food manufacturer’s direct material costs trade on global commodity markets. Their prices are set by weather, disease, exchange rates and trade policy, not by the supplier’s efficiency.
They move fast, and not together. Over the year to 21 October 2026, SMP prices on the Vesper Price Index rose 79.6% while cream prices fell 23.0%. Both come from the same milk. Cocoa prices more than doubled between February and July 2026. A model priced once a quarter misses moves of that size.
Recipes have yields. Water evaporates in baking, air is whipped into ice cream, a share of every batch is lost as trim and rework. The cost of one finished unit depends on how much raw material it takes to make it, not just on the written recipe.
The anatomy of a food should-cost
Every food should-cost has the same six layers. What changes from product to product is how the cost splits between them.
| Layer | What goes in it | Where the number comes from | How often it moves |
|---|---|---|---|
| Ingredients | Recipe share × market price, adjusted for yield and waste | Commodity price benchmarks matched to spec, origin and delivery terms | Weekly |
| Packaging | Film, trays, cartonboard, cases, labels | Packaging price indices and supplier quotes | Monthly |
| Energy | Gas and electricity for baking, cooling, drying | Gas hub and power prices | Daily to monthly |
| Conversion | Labor, overhead, maintenance, depreciation | Industry cost structures, your own plant data | Yearly |
| Logistics | Inbound freight, storage, cold chain | Freight indices, carrier quotes | Monthly |
| Margin | The supplier’s normal profit | Industry averages | Yearly |
The split varies a lot. Two products we have modelled in Vesper, using European industry cost structures for the non-ingredient layers:
| Product | Raw materials | Labor | Overhead | Energy | Margin | Source |
|---|---|---|---|---|---|---|
| Milk chocolate bar | 68.6% | 10.9% | 10.3% | 1.1% | 9.1% | Our chocolate bar breakdown |
| Bread | 53.2% | 22.4% | 12.7% | 1.9% | 9.8% | Our bread cost model guide |
Ice cream sits in between, with one extra layer that matters: packaging. According to Dairy Foods, ingredients make up 55 to 60% of the finished product cost and packaging 15 to 20%, more for small packs.
So a chocolate maker’s cost follows cocoa, sugar and dairy almost directly, while a baker carries nearly half the cost in labor, overhead and margin. The same 20% move in raw materials changes a chocolate bar’s cost by about 14% and a loaf’s by about 11% (our calculation). That is why a supplier’s “raw materials are up 20%” never translates one for one into the price.
For the product-level detail, see the worked breakdowns:
- What it costs to make a milk chocolate bar
- How much it costs to make ice cream
- What it costs to make butter biscuits
- What it costs to make a protein bar
- What it costs to make infant formula
- What it costs to make crisps
- What whey protein costs to make
Cost driver analysis: what actually moves your cost
Not every input deserves the same attention. Cost driver analysis ranks inputs by how much they can move the unit cost, which is their share of the cost times how much their price moves.
Impact on unit cost = input's share of cost × input's price change
Take an industrial bakery where energy is 8% of a product’s cost. Over the year to 13 October 2026, the German THE gas price on EEX rose 115.3%. That alone adds about 9.2 points to unit cost (our calculation), more than any single ingredient in the same product. A small share with a big move can matter more than a big share that barely moves.
Run that sum for every input and you get a ranked list of drivers. Three things usually come out of it:
- One or two inputs dominate. In a chocolate bar it is cocoa butter and cocoa mass, in ice cream the dairy solids, in many baked goods butter or vegetable fat.
- Small inputs can surprise. Energy, packaging and minor ingredients rarely matter, until they double.
- Inputs can offset each other. When one dairy input rises and another falls, a product that uses both may see little net change. A supplier quoting only the input that rose is telling half the story.
This is also where should-cost connects with price risk. The inputs at the top of your driver list are the ones worth covering forward or indexing; see our guide to commodity price risk management.
How to build a food should-cost model in five steps
- Define the unit. Per tonne, per case or per consumer pack. Pick the unit you buy in, so the result compares directly with the quote.
- Write down the recipe and the yield. Recipe shares by weight, then the loss from raw material to finished unit: moisture loss, trim, rework, overrun. Ask the supplier, or use public formulation guides as a starting point.
- Price each ingredient at a matching benchmark. Same specification, same region, same delivery terms. A French wheat flour price and an Italian one moved very differently over the last year; the benchmark must match what the supplier actually buys.
- Add packaging, energy, conversion, logistics and margin. Use industry cost structures if the supplier will not share theirs, and your own plant data if you make similar products.
- Compare with the quote and keep it live. The gap is your agenda for the next conversation. Then refresh the price inputs as often as your largest inputs move.
For a step-by-step walkthrough with screenshots, see how to build a should-cost model for a food product.
Using should-cost with suppliers
A should-cost model is a tool for better questions, not proof that a supplier is overcharging. Gaps have innocent explanations: a better specification than you assumed, forward contracts bought at a different time, a smaller plant with higher conversion costs. Use the model to find out which it is.
Three common uses:
- Testing a price increase. Weight each input’s price change by its share of the supplier’s cost and compare the total with the increase asked. Our guide on how to respond to a supplier price increase works through a full example.
- Open-book negotiations. Share your model’s structure and ask the supplier to correct it. Suppliers often prefer that to defending a single number.
- Index-linked contracts. Once both sides agree on the cost structure, the largest inputs can be linked to published benchmarks, so the price moves with the market instead of being renegotiated. See how a price adjustment clause works.
Common should-cost mistakes in food
- The wrong benchmark. A spot price for a product bought on annual contracts, or the wrong origin. The model ends up precise and wrong.
- Ignoring yield. Costing the written recipe instead of the raw material it takes to make one finished unit.
- Static prices. A spreadsheet priced once and never refreshed. In food, a model older than a month is usually out of date somewhere.
- A margin of zero. Suppliers need to make money. A should-cost without a normal margin will always say the quote is too high.
- Mixed dates. One input priced this week, another from last quarter. Every input in the model should share a pricing date.
Should-cost analysis with Vesper
Vesper’s should-cost model tool, Cost Models, links every ingredient in your recipe to a live commodity price series, adds packaging and non-ingredient costs from industry data for your country and sector, and recalculates whenever a price moves. You see the total, each layer’s share and which input is driving the change, in one view your procurement, finance and sales teams can share.
Frequently asked questions
What is should-cost analysis?
Should-cost analysis estimates what a product ought to cost by adding up its inputs: each ingredient at its market price and share of the recipe, plus packaging, energy, labor, overhead and a normal margin. You then compare that figure with the price a supplier quotes or the cost you actually pay. The gap tells you where to ask questions.
What is the difference between should-cost and clean-sheet costing?
They are close to the same thing. Clean-sheet costing usually means building the cost from zero, ignoring the supplier's quote and history, while should-cost is the wider practice of estimating a fair cost and comparing it with the quote. Teardown analysis is the hardware version, where engineers take a product apart to cost each part. In food, the recipe and the process take the place of the teardown.
Why is should-cost analysis harder for food than for manufactured parts?
Because most of the cost moves every week. Commodity raw materials make up a large part of direct material cost in food, and their prices can double or halve within a year. A should-cost model for a machined part can be updated quarterly. A model for a dairy dessert priced in January can be badly wrong by spring.
Which costs go into a food should-cost model?
Six layers: ingredients (recipe share times market price, adjusted for yield and waste), packaging, energy, conversion costs (labor, overhead, maintenance, depreciation), logistics, and margin. Ingredients are usually the largest layer, but in products like bread, labor and overhead can be close to half of the total.
How often should a should-cost model be updated?
As often as its largest inputs move. For products dominated by dairy, cocoa, sugar or vegetable oils, that means weekly price inputs. Conversion costs and margins move slowly and can be reviewed once or twice a year.
