Always On Special: 16 Weeks of NZ Supermarket Mailers

Every Monday the Woolworths and New World mailers are transcribed, recording each lunchbox item on special along with the regular price where the mailer prints one. Any special that changes a price in the planner is first checked against that product's own online listing. That file now holds 16 unbroken weeks, 6 July to 19 October.

It answers a question a single week's mailer cannot: is the thing on special this week actually a bargain, or is it on special every week?

For three items, the answer is every week.

The 16-week record

"On special" counts a week where at least one product with that ingredient was discounted at either chain. "Discount measured" is the smaller number of those weeks where a regular price was printed next to the sale price, so I could work out how much came off. Where that number is under three I do not report a median at all — carrots were on special in nine weeks but priced in one, and a median of one reading is not a median.

Yoghurt, bread and cheese

Those three were discounted somewhere in all 16 weeks. Not most weeks. All of them.

But never the same packet

That sentence needs one qualification before it is worth anything, and it is the qualification that makes it useful.

"On special" here counts a week where at least one product carrying that ingredient was discounted at either chain. It does not mean the same cheese. So I went back and counted products rather than categories:

Three categories that never came off promotion, and the most weeks any single packet was discounted was six out of 16. Yoghurt ran through 74 different products to stay discounted for 16 weeks straight.

That is a different claim from the one a mailer invites you to make, and a more practical one. The promotional price is closer to what the category actually costs than the shelf tag is — if you are willing to take whichever brand and size is down this week. If you buy one particular yoghurt, you are not shopping a permanent discount. You are waiting for a rotation — and the most often any single yoghurt product came round was six weeks in 16.

There is a second-order effect worth naming. Because these items are always discounted, they are also the items where a mailer looks most generous. A front page full of yoghurt, bread and cheese is a front page full of things that were going to be discounted anyway.

The discount moves even when the discounting does not

Yoghurt's best discount, week by week:

14% · 30% · 39% · 51% · 37% · 32% · 25% · 34% · 33% · 23% · 34% · 20% · 24% · 22% · 30% · 28%

Discounted in every one of those weeks, and priced in every one of them too, so this row is as solid as anything here gets. But the size ran from 14% to 51%, median 30%.

That is the useful shape. Whether yoghurt is discounted is not a question. How much is, and the difference between a 14% week and a 51% week is real money across a term. It is also completely invisible if all you look at is the word "special" on the shelf edge.

Bread does the same thing with gaps in it:

35% · 40% · — · 52% · 18% · — · 27% · 20% · 27% · 27% · 23% · 19% · — · 21% · — · 36%

The dashes are weeks where bread was on special but the mailer did not print a regular price to compare against, or weeks it did not appear. Either way I will not put a number there.

The two chains do not discount the same things

Cheese and yoghurt are near-constant at both chains — 14 and 14 weeks for cheese, 15 and 16 for yoghurt. Below those two the chains diverge, and the split matters if you only shop one of them.

Bread is a Woolworths habit — 12 weeks of 16 against 11. Carrots run the other way, 9 against 3. Neither is a big enough sample to be a law, and both chains are perfectly capable of changing habit next month, but it is the sort of thing worth knowing if you have already decided which supermarket you use and are choosing what to buy there rather than where to go.

What I would actually do with this

Stop treating the always-discounted categories as opportunities — and stop treating your usual brand as one. Yoghurt, bread and cheese do not need timing, but they do need flexibility. There is a discounted yoghurt every week; there is not a discount on your yoghurt every week, and over 16 weeks the best any single packet managed was six. Buying whichever one is down pays every week. Waiting for a particular one does not.

Do not confuse a big discount with a cheap lunch. Across this week's mailers the dollars saved track the price of the thing being discounted almost exactly, while the discount rate does not track price at all. A big number off is mostly information about how expensive the item was.

Spend the attention on the rare ones instead. Of the 106 different items that turned up at all, 22 appeared in exactly one week. Those are the ones where a special genuinely is an event, and where buying two is worth it if the thing keeps.

I want to flag something about that second number, because it cuts the other way from everything else here. A "16 out of 16" claim gets stronger the patchier my collection is — if I missed items some weeks and yoghurt still showed up every week, that is more convincing, not less. The once-only count does not work like that. If my reading was thin in the early weeks, some of those 22 items were probably on special in weeks I did not catch. Treat it as a floor on how much variety there is, not a fact about the mailers.

Why there is no "specials are getting more common" section

I nearly wrote one. The count of promotions I recorded per week runs 29, 21, 36, 21, 33, 42, 54, 54, 56, 104, 132, 80, 121, 175, 184, 200 — from the twenties in July to past a hundred in September. That looks like a clear trend and would have been an easy paragraph.

It is not a trend. It is me. From September I started entering the mailer PDFs in full instead of picking items off the chains' websites, so I am recording roughly three times as much of the same reality. The supermarkets did not triple their specials; my notebook got bigger.

The same problem sank a second finding. Across those 16 weeks, 70% of New World's specials are member-only prices, which would be worth knowing. Split by week it goes 50%, 44%, 55%, 73%, 47%, 65%, 0%, 100%, 97%, 62%, 74%, 76%, 77%, 69%, 73%, 79%. A supermarket's membership policy does not flip between nothing and everything from one week to the next. I was labelling them inconsistently, and the total was hiding it.

Both are in this article as absences on purpose. If a number in a time series moves at exactly the point my method changed, the method is the more likely explanation, and splitting the total by interval is what shows it.

What I would not claim

This is two chains, not three. Pak'nSave is in the price index but not in this file — I read its mailer differently and do not have 16 clean weeks of it, so including it would mean padding.

"On special" here means on special in a mailer I read, which is a North Island mailer. Chains vary promotions by region and New World stores are individually owned, so your store may differ.

And the median discounts lean on weeks where a regular price was printed. New World frequently does not print one. Where that made a row too thin to trust, the table says so rather than guessing.

The planner's "Rank by cost after this week's specials" option reads the current week from this same file, so you can check this week's entries against what it offers you.

Plan a week of lunches →