More from Milk, Pillar One. The Thursday explainer, following Tuesday's hero investigation, What is a litre worth once it leaves the farm, and who decides?
A litre of milk used for butter and skimmed milk powder carried an estimated factory gate value of 35.62 pence in August 2026. Put the same litre through a mild Cheddar and whey route and the value was 35.48. AHDB publishes both calculations monthly, on declared assumptions about milk composition, yields, processing costs and co-product income. It also publishes the components, so butter's 16.20 pence and skimmed milk powder's 18.46 can be argued with line by line.
No transparent, separately published UK or Irish route series was found for milk processed into higher-value protein ingredients. That does not mean finance teams cannot value those products. It means they must build the comparison themselves.
Finance directors do not approve processing investments by choosing the product with the highest published price. They compare the cash an asset may generate with the capital required to build and operate it. The model will include the expected selling price, usable yield, energy consumption, labour, maintenance and working capital. It will also allow for the value, or the disposal cost, of every stream leaving the process.
Protein fractionation complicates that work because the route cannot be represented by a single product. Whey protein concentrate is not interchangeable with whey protein isolate. Milk protein concentrates have different economics from proteins recovered from cheese whey. The value of permeate depends on whether a processor can sell it, process it further, or must bear the cost of disposing of it. Plant design determines which of those outcomes is even available.
That does not make comparison impossible. It makes the assumptions more consequential.
AMPE and MCVE give finance teams a common external reference. Neither describes the precise economics of an individual dairy, and AHDB identifies both as estimates of market returns. Transport of the milk to the dairy is excluded. Processing costs are updated quarterly and yields annually. A processor still replaces the published assumptions with its own operating data before committing capital.
The value of the indicators lies elsewhere. They are an independent starting point. A lender, board or investment group can see how the reference was built and then compare it with the company's model. Where the two diverge, management then has to explain why.
AHDB publishes a third indicator. It is not a route value either. Milk Market Value stood at 35.51 pence in August and tracks farmgate price movements roughly three months ahead, valuing the average litre across typical UK utilisation. It tells a finance director what milk is worth in general but doesn’t help them choose between two configurations of the same processing plant.
Ireland's monthly indicator sits in the same category. Ornua's Purchase Price Index measures returns on a product portfolio that typically includes butter, cheese, whole milk powder and protein products. It is expressed against a 2010 base and converted into an indicative milk price after a deduction for estimated assembly and processing costs. So protein products already sit inside a published milk-value indicator. What is not available on either island is a figure that isolates a protein route, which is why the useful question is about separability rather than absence.
A protein project therefore begins without an external reference of its own. Ingredient prices are available from organisations including Expana, StoneX and DCA Market Intelligence, and Vesper provides tools that combine commodity prices with product yields. Those help establish the revenue side of a model, they do not produce a standard monthly route value directly comparable with AHDB's two.
So a project sponsor must decide which product configuration to model and what yield the processing plant can achieve. Finance and the exec must determine the appropriate processing cost and the treatment of co-products. The resulting number may be technically sound, and it remains specific to that asset and those assumptions.
That is an information problem rather than proof of an investment bias. To show that the absence of an external benchmark redirects capital, evidence would be needed from rejected projects, investment committee papers or lenders' credit decisions, and no such evidence is presently available. Nor is there evidence that financiers automatically increase a project's discount rate because a published milk-equivalent comparator is missing.
What can be said is that the burden of proof differs. A butter, powder or Cheddar proposal begins with a recognised market indicator and moves on to processing plant-specific economics. A fractionation proposal has to establish both the reference case and the performance of the proposed asset. That gives the finance team more assumptions to test and the board more points at which to challenge the case.
The demand signal is what makes the additional work necessary. In the 12 weeks to 8 August 2026, British shoppers bought 2.5% more cow’s milk cheese by volume and spent 0.6% more on it. Cottage cheese volumes rose 43% and the wider segment AHDB calls other cow’s milk cheese grew 13.0%, while Cheddar was among the lines in decline.
Those figures show purchasing moving towards parts of the cheese category associated with protein. They do not establish the return on a fractionation plant. Retail growth in cottage cheese cannot be converted into demand for whey protein isolate, because the products serve different customers and need different assets. What the figures do give a board is a reason to examine whether its existing route assumptions still describe where future value will be captured.
That question is more pressing as ownership of UK capacity changes. On 14 August 2026, Lactalis agreed to acquire Saputo's UK dairy division at an enterprise value of approximately £988 million. Five manufacturing sites are included, and brands including Cathedral City and Davidstow. Completion is expected by the end of the first quarter of 2027, subject to applicable regulatory approvals.
That price cannot be treated as a valuation of the plants alone. It covers an operating business with very familiar brands, customer relationships and manufacturing assets, and the announcement does not disclose how the enterprise value was divided between them. What it does show is substantial capital committed to established UK routes while the economics of alternative protein processing stay less visible.
The response is not to wait for somebody else to publish a third indicator. The immediate task is to impose comparable discipline on every route. Each model should begin with one milk composition, one financing period and one treatment of working capital, applied across all of them. Product prices should come from identifiable external benchmarks. Yield assumptions should be supported by plant trials, supplier guarantees or operating evidence. Co-product income should be shown separately rather than used to conceal weak core economics.
Sensitivity analysis is then more useful than any single route value. A board needs to know what happens when the ingredient price falls, when usable yield misses plan, or when energy costs rise. It also needs to understand whether the asset can change its product mix when relative prices move.
The absence of a published fractionation indicator is a genuine information gap. It does not determine the investment decision. It leaves each processor to build, alone, a standard the established routes already have. The board's responsibility is to see that the third route is tested with the same financial discipline as the first two, even when the first comparable number has to be built inside the room.
Route values: AHDB market indicators, published 25 August 2026. Retail data: NIQ Homescan POD, Total GB, 12 weeks ending 8 August 2026, published by AHDB on 25 August 2026. Ornua Purchase Price Index: Ornua's published PPI page and its monthly PPI reports.
Sources & data
- AHDB: Market indicators, AMPE, MCVE and Milk Market Value, published 25 August 2026, with methodology and calculation dataset
- NIQ Homescan POD, Total GB, 12 weeks ending 8 August 2026, published by AHDB on 25 August 2026
- Ornua: Purchase Price Index, published PPI page and monthly PPI reports
- Saputo Inc.: Saputo enters agreement to sell its United Kingdom operations, 14 August 2026
- Expana, StoneX, DCA Market Intelligence and Vesper: public product and methodology pages for dairy protein assessments and milk valorisation tools
