Back to AllotmentAllotment

What has changed

Allotment is in beta, at version 0.3. A new version is cut when the figures on the page could have changed for a reason you would want to know about: a data correction, a new measure, a changed method. Smaller changes to wording and layout are listed as they land. The reasoning behind each decision is on the methods page.

Since 0.3

  • Every dish and every ingredient now carries its score on the UK Nutrient Profiling Model, the arithmetic behind HFSS: points for energy, saturated fat, sugars and sodium per 100 g, minus points for fruit, veg and nuts, fibre and protein, with four or more marked less healthy. Dish cards and the recipe explorer show it, the ingredient index can filter by it, and the planner can leave less-healthy dishes out. The fruit, veg and nut share is calculated from the ingredient list, as the model's guidance directs. The word HFSS is never used as a label: that is a legal status needing a regulated product category too. No verdict is given where the record lacks an input the model needs.
  • The ingredient index can rank by any of the 25 nutrients it carries: choose most of a nutrient, pick iron, calcium, protein or another, and the ingredients with the most per 100 g come first, with where each sits among those measured for it. Names that land on one composition record are folded into one line.

Version 0.3, land measured against a boundary

24 September 2026

The land figure is rebuilt. It now measures the cropland a plan needs against the EAT-Lancet cropland boundary shared equally, about 31 m² a week each, with pasture shown beside it and not counted. Before, every plan came out at about a tenth of a line that a typical British week sat inside, for three reasons that are now fixed.

Figures that changed

  • The line is the EAT-Lancet boundary for cropland, 13 million km² (11 to 15), shared between everyone alive: 30.9 m² a week each. It replaces an equal share of all farmland in use, 119 m², two thirds of which is permanent pasture that could never grow a crop and which a typical UK week sat comfortably inside. A share of the status quo is not a limit.
  • Only cropland is measured against the line: arable, permanent crops, temporary grass and fallow. Permanent pasture is shown beside it, per week, with no verdict, because there is no boundary for pasture. Only beef, lamb, milk and sheep's milk carry any; for beef it is four fifths of the ground.
  • Every food is now priced on Poore and Nemecek's production-weighted figure, the basis of their published numbers and of the British diet references, instead of the median study. For grazing animals the median study sits far below reality, because most of the world's beef and lamb comes from extensive systems the literature under-samples: beef reads 232 m² a kilogram rather than 99, cheese 88 rather than 20. The spread across studies is still shown on every line.
  • The bar is grossed up for the weight that has no land figure, as the diet comparison already was, and says so. Where less than a quarter of a plan's weight has a figure no share is claimed.
  • Land now reaches about three quarters of a plan's weight rather than two fifths: semi-skimmed milk, chicken thigh, cheddar, parmesan, red lentils, tinned beans and eighty other common ingredients had no bridge to a land figure. Where the bridge is a stand-in (a lime costed as a lemon, a black bean as a field bean) the line says so. Wild-caught fish, salt and stock are counted as needing no ground.
  • A day is shown as a day: the land card had said m² a week whatever was on screen.

Known limits

  • The boundary is divided by today's population. The commission's own per-person figure divides by the ten billion expected in 2050 and is a fifth lower, 25 m² a week; both are stated in the drill-down.
  • Cream and yoghurt are costed as milk, which understates cream. Butter has no land figure at all and is outside the match.
  • The British diet comparison still counts all land, pasture included, because the published diets do; it is on the same production-weighted basis as the plan and is labelled as total land.

Version 0.2, second beta

21 September 2026

Two corrections that change figures, and the tool opening up to what people already cook. Every plan is now judged on eighteen nutrient checks rather than thirteen, with three more shown but not counted; dried fruit is no longer charged as free sugar; and a person can type in their own dishes, see them scored the same way, and start the week from them.

Figures that changed

  • Five nutrients join the checks: vitamin B12, folate, zinc, vitamin B6 and riboflavin, against UK reference intakes, where the composition tables carry them for about nine tenths of the food. A plan is judged on eighteen checks rather than thirteen, so a week that was adequate may now be short, most often on B12 in a fully plant-based week.
  • Vitamin D, iodine and selenium are shown against their references but not planned against or counted: vitamin D because food alone rarely reaches 10 µg and the advice is a supplement in autumn and winter; iodine and selenium because the tables carry them for too little of the food to judge a week by.
  • Dried fruit no longer counts as free sugar. The rule had depended on the word: dried apricots were counted and raisins were not. Ordinary dried fruit is outside the UK definition; purees, pastes, juices and candied fruit stay inside it.
  • Pea protein as an ingredient had been scored as raw peas, about a fifteenth of its protein; it now has its own record from product labels, as does whey protein isolate, which joins the protein shakes for plans that are not vegan or dairy-free.
  • Pitta, naan, flatbread and tortillas had been costed on a baguette factor; they now use the pita bread and tortilla wrap factors the library carries, and breadcrumbs use the sliced loaf.

What it does now

  • The page opens on a simple view: one control line, the week's dishes coloured by climate impact, one chart of the four budgets, and two buttons. Everything else is behind Drill down, which opens on the same plan. Every dish opens its card with the ingredient drill-downs behind it; each can be remixed on its own, or the whole plan with one button; the chart rows open their layers; the nutrition goals can be edited in place.
  • My dishes: type in what you already cook, one ingredient a line, picking each from the library as you type, and see it scored the same way as every dish here with a review that says what was matched loosely, what drives its climate impact, and where it is low or high. Every matched line opens the full record; wrong? lets you say what a line really is. Dishes are kept on your device, rated, and can start your week. Choose which meals you cook, batch cook, and save a week to reopen later. The cupboard uses the same rows.
  • Every ingredient you pick or correct is remembered, with who decided and why, so the next person who types the same words is offered the right thing.
  • An ingredient index: every ingredient the library can score, what a kilogram costs the climate and the pocket, what 100 g carries, where it sits among the rest, and up to four side by side.
  • A menu on every page and a recipe explorer that searches every dish by name or ingredient. Cuisine filters above the plan. Every meal card colours its climate impact and cost by where the dish sits among the dishes for that meal. The nutrition bars fill as the table comes into view. The targets control offers rather not say instead of either.

Known limits

  • Your dishes, cupboard and saved weeks live in this browser only. There are no accounts yet.
  • A dried or tinned amount is restated to the record it lands on only where the line says so; once an ingredient is picked from the library, put dried or tinned in the amount if it matters.
  • Whey protein's climate figure is the dried whey powder factor, the nearest the library carries, and understates an isolate. Naan is costed on the pita factor.
  • The comparison ladder's rungs are drawn from the whole collection at default settings; the week of vegan dishes still cannot be built to the reference intakes.
  • Climate factors are Agribalyse, modelled on French supply and used as a proxy for UK food. Prices are DEFRA Family Food averages to March 2024, valuing the ingredients used, not the packets bought. It is not dietary advice.

Version 0.1, first beta

17 September 2026

The first version anyone outside the team can use. It plans a day or a week of real recipes against your nutrition goals at the lowest climate impact and cost it can find, and it says where every figure comes from and where it stops. Beta means the arithmetic has been audited three times and the data is still being corrected as people find things.

What it does

  • Plans a day or a week from 6,175 BBC Good Food recipes, inside your energy, nutrition and spending goals, at the lowest climate impact and cost.
  • Measures every plan on four things: climate impact against a 14 kgCO2e a week goal and a typical UK week, land against an equal share of farmland, nutrition against UK reference intakes, and cost against your weekly budget.
  • Stands the plan beside a typical week of vegetarian, fish and meat dishes from the same collection, costed the same way, so the comparison is like for like.
  • Offers a cleaner dish, or one thing changed inside the dish, for any meal, and holds the rest of the plan still when you take one.
  • Starts from a shopping list if you would rather: type what you bought and it says what that costs the planet and what you can cook from it.
  • Ranks every dish on all three budgets at once, and draws the best of them into a month, a quarter, or a year of dinners.
  • Shows what any food costs the climate, what makes that figure, and how much it varies between the farms it was measured on.
  • A methods page that says which data is used, how it is joined, and what the numbers do not cover.

Corrected on the way here

  • Fruit, vegetables and pulses were being charged as added sugar. Free sugars are now decided per ingredient, which took a third of the counted sugar off every plan.
  • 139 ingredients had been costed on a limoncello record because a matcher fell back to the wrong median. Red Leicester was the one that gave it away.
  • Every plant milk had been matched to sheep's milk, and coconut milk to cow's milk, across 226 recipes.
  • Nut butters were costed as dairy butter. An almond butter now has 21 g of protein rather than 0.4 g.
  • Twenty-one sauces were costed on a duck a l'orange record. Dishes with sausagemeat in them were being offered as vegetarian.
  • Nitrous oxide from fertiliser had been filed under fossil fuel in the Agribalyse split. It now sits with land management, which is where it comes from.
  • Salt a recipe adds but never weighs is now counted, from the recipe's own published salt figure.
  • A nutrient the composition tables do not carry for a food is treated as not known, not as zero, and the plan says how much of its weight that covers.
  • Comparisons to published UK diet studies were dropped: they are costed on a different database that runs about twice as high on the same basket, and were turning a real 55% saving into a claimed 88%.

Known limits

  • Climate factors are Agribalyse, modelled on French supply and used as a proxy for UK food. The library carries the 2025 refresh; foods added in later releases are missing.
  • Prices are DEFRA Family Food averages to March 2024, not today's shelf prices, and they value the ingredients used, not the packets bought.
  • A week of vegan dishes from this collection cannot yet be built to the reference intakes, so that rung is missing from the comparison. That is a fact about the recipe book, not about eating that way.
  • Spices, chorizo, olives and passata are costed on stand-in records because the library has no closer row.
  • It is not dietary advice, and it does not cover water, biodiversity or animal welfare.