What Britain actually keeps in the cupboard

Most lists of “store cupboard essentials” are somebody’s opinion, and the ones that cite anything cite a survey asking people what they think is in their kitchen. There is better data than that.

DEFRA’s Family Food survey measures what roughly 5,000 UK households actually buy, in grams per person per week, as part of the Living Costs and Food Survey. It is National Statistics, it runs to 1974, and nobody writing about pantries seems to use it.

Here is what it says about the long-life half of the shop, for the financial year ending 2025.

What Britain buys

What this measures matters. These are purchases, not contents. A jar of cumin that lasts a year ranks nowhere against pasta bought weekly, so this is not a list of what sits in cupboards. It is a list of what moves through them, which is the more useful thing for planning: it tells you what people will genuinely eat rather than what they will buy once and resent.

Four things worth noticing

Biscuits come first, by a distance. More than rice and pasta together. That is not a moral failing, it is a fact about what a cupboard is for: things that need no cooking, keep for months and require no decision. Any honest pantry plan has to account for that rather than pretend people will only eat lentils.

Beans outrank pasta. 94 g against 89 g, and that is before the separate 65 g of baked beans. Tinned pulses are the single most-bought long-life protein in the country, which is a reasonable argument for building a corpus on them.

Canned tomatoes are 63 g a week per person, roughly a 400 g tin per person per six days. The tomato base is not a niche technique, it is what a large part of the country is already doing.

Canned soup is 53 g a week, which is more than flour. This site stocks no tinned soup at all, uses it in nothing, and has never mentioned it. That is a straightforward gap between what we assume a pantry contains and what one actually does.

What we got wrong

Writing this found the soup thing, which is the point of looking at data rather than at other people’s lists.

There is a second, subtler gap. Breakfast cereals are third at 100 g a week and this site carried oats and nothing else. Ready-to-eat cereal is genuinely outside what the site is about, since it needs no cooking and no skill. But cornflakes crushed into a coating, or used to bulk out a crumble topping, are real techniques that a cupboard already holding a box can use, and dismissing the category entirely was a decision made without looking. There is a box in the ingredient list now.

Then we looked again, and the table above was hiding something

The first pass at this ranked the categories, which is the table above, and stopped. Going back to the same spreadsheet and ranking every individual line in it turned up one thing that matters more than anything else here.

Row 14 is not what it looks like. “Vegetable and salad oils” at 54 ml a week is not one thing. Underneath it, olive oil is 16 ml and everything else, sunflower and rapeseed and blended, is 38 ml. The country buys well over twice as much plain cooking oil as olive oil, and every recipe on this site had been written assuming the olive.

That is the clearest example of the thing this whole page is for. Olive oil is what a recipe writer reaches for. It is not what is in most kitchens, and the difference was invisible while we were reading the category total.

It buys no shelf life, which is worth saying plainly: FoodKeeper gives olive and vegetable oil the same six to twelve months. Switching does not make a cupboard last longer. It means the recipe matches the bottle already on the shelf.

It does buy something else, and it was not the reason for doing it. British vegetable oil is largely rapeseed, and CoFID puts rapeseed-based blends at 9.6 g of n-3 fats per 100 g against olive oil’s 0.7 g. Forty-five recipes here now fry in it, and the corpus carries noticeably more n-3 than it did yesterday for no other reason.

That is not a licence to skip the fish. CoFID’s column counts all n-3 fats together, and the kind in rapeseed is not the kind in mackerel: the NHS advice about a weekly portion of oily fish is about the marine forms specifically. A bottle of vegetable oil is a good thing to have found. It is not a tin of sardines.

What else came out of the wider look, and is now in the ingredient list: tomato ketchup and mayonnaise, from the “pickles and sauces” category that ranks second; and yeast extract, which sells in quantities too small to rank anywhere and sits in a very large number of British cupboards doing a job nothing else of its weight can do.

Mayonnaise is the surprise of the four. Most people assume it is impossible without a fridge. USDA FoodKeeper says commercial shelf-stable mayonnaise is safe at room temperature after opening, and that the label telling you to refrigerate it is about quality rather than safety.

What this does not tell you

Purchases are not stocks. Somebody buying 96 g of rice a week is not keeping 96 g of rice; they are keeping a kilo and working through it. For planning a 30-day pantry, the useful figure is consumption rate, which this gives, applied to a horizon, which it does not.

Averages hide households. The same dataset breaks down by income decile, region, age and household composition, and the differences are large. A single national number is a starting point rather than an answer.

It says nothing about what is good. Chocolate biscuits rank thirteenth. The data describes the country as it is, not as a cupboard ought to be stocked, and the store cupboard build is the opinionated version for a reason.

Reproduce any of this with scripts/defra-pantry.py in the repository, which reads the published spreadsheet directly.