More from Milk, Pillar One. The Friday briefing, closing a week that opened with What is a litre worth once it leaves the farm, and who decides? and continued with Two routes start with a number. The third starts with an argument.
Five days, one question, two pieces of reporting. On Tuesday this publication asked what a litre of milk is worth once it leaves the farm, and who decides. On Thursday it set out how the two published answers are built, and what a protein project has to establish before it can be compared with either.
What follows is what the week returned: five things a reader in a particular job might do differently on Monday, and the one thing the reporting still cannot show.
For boards. The two published route values are useful for a reason that has little to do with the figures themselves. AHDB publishes the components underneath each one. The butter and skimmed milk powder route is built from butter at 16.20 pence a litre, buttermilk powder at 0.96 and skimmed milk powder at 18.46, less stated processing costs. The mild Cheddar and whey route is built from Cheddar at 29.67, whey powder at 4.41 and estimated whey butter at 1.39. A processor who disagrees with either total can name the line they disagree with.
That is the standard to hold a fractionation paper to. Ask the sponsor for the components rather than the conclusion, and ask what each one would have to be wrong by to change the decision. A model a board can argue with line by line is worth more to that board than a model that arrives as a single figure, however carefully the figure was produced.
For brands. The fastest-growing line in the cheese fixture over the 12 weeks to 8 August was cottage cheese, up 43% by volume. It has been on British shelves for decades. We looked for new British capacity behind that growth, and for a campaign explaining it, and found neither.
We are not able to say why shoppers are buying it, because the data measures purchases, not motives. What we can say is that we looked for a British producer or a brand claiming the growth and did not find one. Occasions get defined by whoever explains them first, and on the evidence of this week nobody has started.
For processors. Thursday's explainer split the work of a conversion business case in two. The project sponsor decides which product configuration to model and what yield the plant can achieve. Finance sets the processing cost and decides how each stream leaving the process is treated, as a revenue or as a cost of disposal. The value of permeate, to take the clearest case, depends on whether a plant can sell it, process it further, or must pay to remove it, and plant design decides which of those is even available.
Neither half stands up without the other. In many businesses those two decisions sit with people who have never taken them in the same room. Getting them into one room is cheaper than any of the capital the decision would commit.
For retailers. Cow's milk cheese volume grew 2.5% over the period while spend on it grew 0.6%, average prices having fallen 1.8%. More cheese left the fixture and barely more money came in for it. The growth went to the tub rather than the block: Cheddar was among the lines in decline, alongside extra mature slices, down 16.1%, reduced fat and functional lines, down 12.7%, and grated mild, down 11.6%. On the previous read, to 11 July, Cheddar still accounted for 43.2% of all cow's milk cheese volume.
That is volume growth without value growth, and it puts two options in front of a range decision. List more of what is growing, or ask a supplier what the protein inside it is actually worth. The second is the more useful question, and nobody in the chain can currently answer it with a published figure.
For suppliers. A conversion project begins with no external reference of its own. Every business case is therefore built on the sponsor's own assumptions about configuration, yield and cost, and defended alone in front of whoever is being asked to fund it.
The firms that sell membrane, filtration, separation and drying plant hold better data on all three than most of the firms that would finance the asset. That is a commercial position, not a technical one, and it has an expiry date. It lasts exactly as long as the absence of a published comparator does.
What the week did not establish. Whether the absence of a published comparator changes any capital decision. Showing that would take evidence from rejected projects, from investment committee papers or from lenders' credit decisions, and none of it is available to us. Nor is there evidence that a bank raises a project's discount rate because a published milk-equivalent comparator is missing.
That makes this an information problem rather than proof of a bias. Two of the three routes out of a British dairy can be compared against a declared standard every month. The third has to be argued from first principles each time. What that costs, and who it costs, is a question this week has opened, not answered.
Tuesday brings the retail and markets briefing, and with it the quarter map: six questions across 13 weeks, and what each one asks. The packaging arithmetic follows on 21 September, worked in public against fees that are already published and invoices that land in November. And the question this week hands on, whether the UK and Ireland have the plant to convert the whey they already produce, returns in October.
Retail data: NIQ Homescan POD, Total GB, 12 weeks ending 8 August 2026, published by AHDB on 25 August 2026, with the 11 July read published on 31 July 2026. Route values and their components: AHDB market indicators, published 25 August 2026.
Sources & data
- AHDB: Market indicators, AMPE and MCVE with their published components, published 25 August 2026
- NIQ Homescan POD, Total GB, 12 weeks ending 8 August 2026, published by AHDB on 25 August 2026; 12 weeks ending 11 July 2026, published 31 July 2026
