The Shelf Price Is Not the Lunchbox Price

A supermarket shelf tells you what a pack costs. It does not tell you what a lunchbox costs, and the two disagree often enough to be worth writing down.

I track 180 lunchbox ingredients. For each one I hold two things the shelf does not put together: the pack the shop actually sells, and how much of that pack goes into one lunchbox. The second number is the one that took the longest to get honestly, and it is the one that changes the answer.

The trolley and the lunchbox

Across those 180 ingredients, the median pack is $4.95 and the median lunchbox serve is $0.47. The median pack holds 10 serves.

Buying one of everything would come to $1030.50. One lunchbox serve of everything comes to $125.22. That ratio is the whole point: most of what you carry out of the shop is not for today.

How many lunchboxes a pack is

Only 3 of the 180 are effectively single-serve. Those are the ones worth watching, and they are almost never the ones that look expensive.

The shelf ranking is about half right

Rank every ingredient by pack price, then rank it again by what one serve costs. The two orders agree, but loosely — the rank correlation is 0.58.

That number is doing real work in both directions. It is not near zero, so the shelf price is not noise: the 450g pack of chicken thigh at $17.90 really is expensive in a lunchbox too, at $3.58 a serve. But it is not near one either. 52 of the 180 ingredients sit on opposite sides of the median in the two rankings — cheap-looking and dear to use, or dear-looking and cheap to use.

Dear on the shelf, cheap in the box

These are the items people flinch at and shouldn't. Of the 87 ingredients whose pack costs more than $5.00, 26 come to under 50c a serve. A large bottle of oil is one of the most expensive single things in the shop and one of the cheapest things in the lunchbox.

Cheap on the shelf, dear in the box

The mechanism is the same every time: the pack is the serve, or close to it. The croissant row is the clearest case — $3.60, and one lunchbox uses the whole thing. It is the most expensive serve in the entire set while sitting in the bottom half of the shelf prices.

This is not an argument against buying any of them. It is an argument for knowing which line in the trolley is the one you are paying for repeatedly.

"Expensive" has three answers, and they are three different ingredients

Everything above ranks ingredients two ways: what the pack costs and what one serve costs. There is a third ranking this data can produce and a shelf cannot, because it needs the recipes as well as the prices — what an ingredient costs across the whole menu set. That is its cost per serve multiplied by the number of menus that use it.

Of the 94 lunch menus in the planner, 150 distinct ingredients appear that I can also price. Another 16 appear that I cannot — they have no pinned product yet, so they are missing from the totals below rather than guessed at.

Three questions, three answers, three different ingredients. And the first two are each in exactly one menu, which makes them trivia. Chicken is in 8, and at $3.00 a serve it carries more of the menu set's ingredient cost than anything else I track.

Those 10 lines are 40% of the $221.70 it would cost in ingredients to make every menu in the planner once. The top 3 alone are 22%.

This is the table the article should have opened with, and it is the one I did not have until the recipes and the prices were in the same place. Cheese appears in more menus than anything else — 20 of 94 — and comes 3rd on it.

Does the ingredient you use most often cost less?

It would be a tidy story, and the data does not support it. The rank correlation between how many menus an ingredient appears in and what one serve of it costs is −0.15 — near enough to zero to mean nothing.

The medians do split. The 21 ingredients used in 5 or more menus have a median serve of $0.28; the 129 used in fewer have a median of $0.45. That is the right direction for the tidy story. It is also 21 ingredients against 129, split at a threshold I chose, with a correlation of −0.15 sitting underneath it. I am writing the numbers down rather than the conclusion.

What the ranking does support is narrower and more useful: the recurring cost of a menu set is not carried by its commonest ingredient or by its dearest pack. It is carried by the few items that are both dear per serve and not rare.

What the serve sizes are, and are not

They are our portions. Recording a cheese serve as 30g out of a 500g block is a choice, not a standard, and a family that cuts a thicker slice will get different numbers out of the same shelf.

What they are not is reverse-engineered. When these were written down, the serve size for each ingredient was recorded first — how much actually goes in a lunchbox — and the portion factors were corrected afterwards where the two disagreed. Five ingredients had to be fixed that way. Had the serve sizes been derived from the factors instead, every table above would be a circular argument dressed as a finding.

Pack sizes are not estimates either. Each one was read off the product listing at the chain that sells it, which is why a 220g hummus tub is recorded as 220g and not as the 200g I had originally written down.

What I would not claim

The same pack-size effect shows up sharply in the free-from aisle, where the alternative often comes in a smaller bag: what gluten-free and dairy-free swaps actually cost puts numbers on that.

These are one household's portions against one chain's online prices — the per-serve figures use Woolworths. Fresh produce sold by weight — bananas, potatoes, courgettes — moves week to week, so its position in either ranking is the least stable part of this — and how much anything actually moves is measured separately.

And the 0.58 correlation is a snapshot. It is high enough that "expensive pack, expensive lunchbox" is a reasonable default; the useful part is knowing the 52 places where the default is wrong.

Every row behind this article is downloadable: lunchbox-ingredients.csv has each ingredient's pinned product, SKU, pack size, pack price, serve and cost per serve, so the whole table above can be rebuilt from it. The 20-item cross-chain basket is a separate file and a separate dataset — both are on the NZ Lunchbox Price Index page, published under CC BY. The chain-by-chain comparison of the same basket is in Where the supermarket gap actually is.

Plan a week of lunches →