The Ravensburger 1000 Index

Puzzle prices are one of those things everyone has an opinion about and nobody has actually measured. So we measured them. We took twenty 1000 piece puzzles that have been listed continuously in Australia since 2017, priced each one year by year, and built an index out of it.

A puzzle that carried a $39.95 RRP in 2017 would need to be $52.95 today just to have kept pace with inflation. The same puzzle's RRP today is $49.82. Puzzle RRPs rose 24.7 per cent over those nine years, while prices in general rose 32.5 per cent. So in real terms a 1000 piece puzzle is about 5.9 per cent cheaper than it was.

That is a smaller claim than the cost-of-living headlines would suggest. It is also, we think, the honest one.

Why a 2017 dollar is not a 2026 dollar

A price from 2017 and a price from 2026 are not the same kind of number, because the dollars are not the same size. Comparing them directly tells you almost nothing.

A 1000 piece jigsaw is unusually well suited to fixing that. The piece count is fixed, the box is a standard size, the brands are the same brands, and a title from 2017 is physically identical to the same title today. Very few consumer goods hold that still. A television from 2017 is not a television from 2026 in any meaningful sense, which is why electronics prices are so hard to compare honestly.

So we hold the product completely still, then express its price in constant dollars using the ABS Consumer Price Index. What is left is the only question worth asking: did the puzzle get dearer faster or slower than everything else?

How the index is built

Three decisions do all the work here, and each one is a place where a sloppier version would go wrong.

1. Only titles that never left the shelf. We searched archived Australian retail listings for 1000 piece titles present both at the start of the period and at the end, which produced a basket of twenty. Every one of them is the same physical product across the whole series.
2. Each title is only ever compared with itself. For each pair of consecutive years we take the titles priced in both years, work out how much each one moved, and average those movements. Then we multiply the years together into a continuous series. This is a chain index, and it is how the Australian Bureau of Statistics builds the CPI. It matters because it means a title joining or leaving the range can never fake a jump in the level.
3. List price, not the price on the day. Sale prices depend on whatever promotion happened to be running when a page was archived, which measures the promotion and not the puzzle. List price is the stable series.

What it found

The column that tells the story is the second last one, the price of the same puzzle in today's money.

It goes up first. Through 2018 to 2022 puzzle prices ran ahead of both inflation and wages, and the real cost of a 1000 piece puzzle climbed from about $53 to nearly $56. If we had built this index in 2022, the honest headline would have been that puzzles were getting more expensive, and that would have been the correct answer at the time.

Then inflation arrived. From 2023 the CPI rose faster than puzzle prices did, and the real cost fell away to $49.82. The entire "cheaper" result is a story about the last three years, and it is driven far more by everything else getting dearer than by puzzles getting cheaper. Puzzle prices did not fall. They just stopped keeping up.

The denominator problem, and why we show our working

Here is the uncomfortable part, and the reason this section exists.

The answer to "did puzzles get cheaper" depends heavily on what you divide by, and the range is wide enough to drive a headline through.

The third row is the trap. The minimum wage was raised deliberately faster than general wages through the inflation of 2022 to 2024, precisely to protect the lowest paid. Divide by it and almost anything looks like it got cheaper. It would let this page claim puzzles are nearly 14 per cent cheaper, which is roughly two and a half times what the other measures support. Correct arithmetic, bad measure.

There is a more technical reason to avoid it, and it applies to median and average earnings too. The price side of this index is deliberately constant-quality: the same twenty puzzles, every year. Median and average earnings series are not constant-quality, because they move with who happens to be working and for how many hours. Pairing a constant-basket numerator with a shifting denominator is inconsistent. The CPI and the Wage Price Index are both built to hold quality constant, which is exactly why they belong here.

The second reading, and why it differs

We track the discounted price as well as the list price, and it tells a different story. On discounted prices the index rises about 45 per cent over the same nine years, well ahead of both inflation and wages.

Both numbers are real. The gap between them is the finding: discounting narrowed. In 2017 these titles were routinely sitting 20 to 25 per cent below list. By 2026 far more of them are simply at list price. So the RRP rose slower than inflation, while the price you actually paid at the till rose faster, because the reliable discount that used to sit underneath it thinned out.

We lead with list price because it is the stable series. We publish the other one because leaving it out would make the headline look better than the truth.

What this means if you buy puzzles

Common questions