How Long Does a Jigsaw Puzzle Design Stay Popular?

A novel gets a season. A toy gets a Christmas. A licensed film puzzle gets whatever is left of the cinema run. So when we sat down to reorder stock earlier this year and found ourselves buying back designs that first went on our shelves in 2020, it was worth asking whether that was sentiment or whether it was a real feature of this category. We went back through six years of our own sales to find out.

The question, and why it is harder than it looks

The obvious way to answer "how long does a puzzle stay popular" is to find the big sellers from your first year, then watch their sales fall away over time. Draw the curve, read off the half life, done.

That method is wrong, and it is wrong in three separate ways that all push the answer in the same direction. Understanding them is most of the work.

The customer base moves. Our list and our traffic are not the same today as they were in 2021. If a design sells the same number of units in two different years, that is not a flat line. Against a smaller audience it is growth, against a larger one it is decline. Raw unit counts cannot be compared across years at all, so every number in this article is a share of its own period, a rank inside its own period, or a rate per month inside its own period. The denominator cancels out.

Out of stock looks exactly like out of favour. This is the big one, and it is the trap that would have ruined the study. A puzzle that sold out in 2022 and was never reordered has a sales history that looks identical to a puzzle that people simply stopped wanting. Zero units either way. Any analysis that does not separate those two things is measuring the shop's buying decisions and calling it consumer demand.

A design that comes back gets a new identity. When a puzzle sells out here it comes off the storefront. When we buy it again, years later, it often gets loaded as a completely new product record with a new creation date. To the database that is a brand new puzzle. To the customer it is the same picture on the same box. We merged those back together by design, which pulled about 66 designs' histories back into one piece each. Every one of those merges made a design look older and longer lived than the raw data said.

What we did

We took every order from the store's first full month of trading in August 2020 through to 3 August 2026, joined each line to the product behind it, and merged product records that were re-listings of the same design by the same maker at the same piece count.

We then defined an opening era cohort: the 349 strongest selling designs from the store's first seventeen months, August 2020 to the end of December 2021. Those designs have had the entire life of the business available to them, so if a puzzle design has a natural shelf life we should be able to watch it run out.

Two exclusions. Funbox is out of the study, because that brand has stopped making puzzles, so its decline says something about a manufacturer rather than about design longevity. Accessories are out too, because glue and boards are not designs. What is left is overwhelmingly Ravensburger, with Cobble Hill, Gibsons, Jumbo, Falcon and HEYE behind it, which is a real limitation and we come back to it at the end.

Finding one: puzzles do not die, they get delisted

Of those 349 opening era designs, 228 sold nothing at all in the last twelve months. Taken at face value that is a 65% death rate over five years, and it would be a perfectly ordinary looking decay curve.

Then we asked the question that actually matters: could a customer have bought them?

Ninety eight percent of the apparent deaths were not deaths. They were empty shelves. Only five designs out of 349, fewer than one in seventy, were available to buy and ignored.

That single table reframes everything. The decay curve you would have drawn from the raw numbers is not a picture of customers losing interest in artwork. It is a picture of a shop that stopped reordering during a quiet period, which is exactly what happened here between 2022 and 2024. Demand did not expire. Supply did.

Finding two: age tells you nothing about rate of sale

If designs wear out, older ones should sell more slowly. This is testable, and it is the cleanest test in the study, because you can run it inside a single twelve month window where every design faced the same customers, the same emails and the same promotions.

The measure has to be units per month that the puzzle was actually on the shelf, not units per year. Get that wrong and you punish everything new: more than half our current range was listed within the last few weeks, so measuring their last twelve months makes them look dead when really they have barely arrived. We made that mistake in the first pass and it produced a lovely, wrong result showing old designs outselling new ones by a mile.

Corrected, and restricted to designs that have had at least six months on the shelf so the comparison is fair:

Flat. The rank correlation between a design's age and its rate of sale is -0.04, with a p value of 0.64, which in plain terms means there is no relationship there at all. A design first listed more than five years ago moves at the same speed as one listed eighteen months ago.

We want to be careful about what that does and does not prove. It is not proof that puzzle designs never fade. The five year old designs still on our shelf are there because we chose to buy them again, so they have passed a filter. The honest statement is this: a puzzle design that is still being stocked after five years sells exactly as well as a new one. Survivorship is not a flaw in that sentence, it is the whole point. The filter is real and it works.

Look at what that does to the shape of the range:

A tenth of the range is doing nearly a quarter of the work, and every design in it is older than five years.

Finding three: the tail is very long, and we cannot see the end of it

Treating this properly as a survival problem, with designs that are still selling correctly marked as unfinished rather than dead, the median survival of an opening era design is 3.7 years. Sixty one percent of them stayed commercially alive for three years or more, a third for four years or more.

But the honest headline is the one we cannot put a number on. Our data is 5.9 years long, and the longest lived designs in it are 5.9 years old. They are still selling. The oldest thing we can observe is exactly as old as our ability to observe it, which means we have not measured a lifespan at all. We have measured the length of our own dataset.

The dormancy numbers make the same point from another angle. Among designs with more than one selling day, 63% had a silence of six months or more and then sold again. Nearly a third went quiet for over a year and came back. The longest gap we found was 1,654 days, four and a half years of complete silence, followed by another sale.

And on the buying side, where the decision costs real money, 66 designs in our catalogue have been listed, sold out, and then listed again. The median gap between first listing and newest listing is 1.9 years, a third of them are three years or more apart, and the longest are 6.3 years, which again is the limit of the data rather than the limit of the design.

What the rest of the industry does with this

We are one Australian shop with six years of history, so the sensible check is whether the manufacturers behave as though this is true. They do, and openly.

The pattern holds in our own top sellers too. The Bizarre Bookshop 2, Colin Thompson's illustrated wall of literary puns, was one of the strongest designs of our first year and is still selling for us today. It is also still listed in Ravensburger's own catalogue. It has outlasted several complete cycles of everything else on a gift shop shelf.

What it means if you are choosing a puzzle

Three things follow, and they run against the instinct that new is better.

1. An older design is a stronger bet than a new one, not a weaker one. A puzzle still in print after five years has survived a filter that a new release has not faced yet. A manufacturer kept printing it, a distributor kept carrying it, and shops kept reordering it. A new arrival has no track record at all. That is not a reason to avoid new puzzles, it is a reason to stop treating "released this year" as a quality signal.
2. If a design you want has vanished, ask before you give up. Almost every puzzle that disappears from a shop does so because it sold out and was not reordered. That is a decision, and decisions can be changed. We have brought designs back after years off the shelf, and they sell as well as they ever did.
3. Be more careful with anything tied to a date. The designs that hold up are the ones that were never chasing the moment: landscapes, cityscapes, illustrated interiors, animals, fine art, detailed comic scenes. Film tie-ins and anything with a year printed on it are the genuine exception, and they behave the way you would expect.

Where this study is weak

We would rather say this ourselves than have it said for us.

It is mostly a Ravensburger study. Once Funbox is excluded, Ravensburger dominates our 2020 and 2021 range, with Cobble Hill, Gibsons, Jumbo, Falcon and HEYE well behind. The finding may be a finding about how Ravensburger manages its catalogue as much as about puzzles in general.

Six years is not long enough. Every longevity number here is bounded by the age of the store. The correct reading of "the longest lived design is 5.9 years old" is not that puzzle designs last 5.9 years. It is that we cannot yet see how long they last.

Design age is a floor, not the truth. We date a design from when we first listed it, which for our founding range means 2020 regardless of when the artwork was actually published. Many of these designs were years old before they ever reached us. Every age in this article is therefore an understatement.

We cannot reconstruct historical stock levels. We know what is in stock now, which is what makes the finding in section one possible, but we cannot say precisely how many days each design was buyable in past years. The exposure adjustment in section two handles the current window properly and is the reason we trust that result more than the others.

We will rerun this every year. The interesting question is not whether the numbers hold, it is how long it takes before we finally watch a strong design actually die.

Common questions