Three NZ Supermarkets: Where the Lunchbox Basket Differs

Until recently I could only compare two chains honestly. New World is now in the index, and this is the first write-up with all three columns in it.

What it is not is one morning's shopping, and an earlier version of this article said it was. Behind the table are 51 readings — 17 items across three chains. 50 of them were taken in the week ending 18 October 2026. The other 1 is older than that, the oldest by 14 days.

That is the rule the table runs on rather than a slip: a row joins the three-way comparison when each chain's reading is within 30 days of the Woolworths round. All three columns are re-read automatically every week from each chain's online shop, but a Pak'nSave or New World row keeps its previous regular price in any week the item is on special or the read fails, because a special is not a regular price. Dropping every such row until all three land in one week would leave less to compare.

But it does mean part of what follows is measuring the calendar as well as the shops, so here is exactly which part:

Every other reading in the table was taken within a week of 18 October 2026.

The totals

On the 17 items I can price at all three on those days:

Pak'nSave is cheapest, which surprises nobody. Two things about that are worth more than the ranking.

The first is the size. Woolworths is $7.24 dearer than Pak'nSave across 17 items — real money, but not the chasm the "Pak'nSave is way cheaper" shorthand implies.

The second is that New World is dearest — $3.53 above Woolworths — and New World and Pak'nSave are the same co-operative. Foodstuffs runs both. Whatever the gap between the two banners is, it is not a difference in buying power.

Most of the gap is five items

Those five lines are 90% of the entire Woolworths–Pak'nSave gap. The other 12 items, added together, account for the rest.

This is the practical finding and it is not the one the shorthand gives you. If you do one shop a week at one chain, you are not losing money across the board by picking the "wrong" one. You are losing it on a short list — and it is a list you could buy somewhere else without driving anywhere twice.

Which aisles the gap comes out of

Five items is a list, not a pattern. Grouping the same 17 rows by aisle gives one, and it says something the list cannot: the gap is not thinly spread everywhere and then piled up in five places. It leans the same way in every aisle, by different amounts.

Pak'nSave has the cheapest sub-total in five of the five aisles, Woolworths in none, and New World in none. New World being dearest overall is one thing; not having a single aisle where it comes out lowest is a different and harder fact about it.

Every aisle leans the same way this round — Pak'nSave is cheaper in all five of them.

I would not turn that into shopping advice yet — one round of aisle totals is one round. But it is a sharper thing to re-test next round than "is Pak'nSave cheaper" — that one is answered, and it stopped being interesting.

And it runs the other way on fresh produce

Woolworths is cheaper than Pak'nSave on three of the 17 items:

Two of the three are fresh produce. The other line — sliced ham — sits inside $0.50, which is a rounding difference rather than a finding. That is not a coincidence I can prove from one round, but it is the second time the fresh produce rows have leaned this way, and it is the sort of thing worth watching rather than announcing.

Counting cheapest-per-item rather than totals: Pak'nSave wins 13 of the 17, Woolworths 3, New World 1.

What is not in the table

Wholegrain bread, cheese block and eggs are missing from the three-way comparison — I do not have a New World price for them in this round, so they sit out rather than being estimated. An estimated number in a price index is worse than a missing one.

The same rule applies to specials. Every figure here is a regular shelf price. If a chain has something on promotion this week, that is real money in your pocket, but it is not what an index measures — let specials in and the index reports deflation every time something goes on sale. The weekly deals are tracked separately, and what that record shows is its own article.

What I would not claim

These are online Click & Collect prices, not shelf tags, and the two can differ.

Pak'nSave and New World price by store, and mine are both Ormiston in East Auckland. This is one corner of one city's answer, not the country's.

And it is one round, not a trend. Whether these prices move at all is a separate question I track separately: what actually moved across every reading so far. The five items carrying the gap here are not guaranteed to be the five carrying it next month. On the two-chain basket I can check that against May. On the 14 rows priced at both chains in both rounds, the four widest gaps made up 44% of the Woolworths–Pak'nSave difference in 16 May 2026 and 90% in 18 October 2026, and two of those four rows are the same in both rounds — and which chain wins which row has already changed on three items. Which items carry the gap is the part to re-check every round.

The full table, with every row's reading date and the reason for each exclusion, is on the NZ Lunchbox Price Index page, and the same rows are downloadable as price-index.csv — one row per chain per item, each carrying its own reading date.

One more thing that file will tell you about itself. A pinned product and SKU — so you can see exactly which brand and size was priced — is recorded for 17 of the 17 Pak'nSave rows, 17 of the 17 New World rows and 15 of the 17 Woolworths rows. New World and Pak'nSave share product codes, so unless a row needed a different product at New World, both banners are read against the same SKU. Where a SKU is missing, the price is still a real reading, but you cannot check exactly which product was priced. The CSV marks every row either way. Three-chain totals are the kind of number that is easy to assemble wrongly, so the rows are there to be checked. Published under CC BY.

Plan a week of lunches →