The longest anyone has played each of my 9 games

In five of my nine games, one person has played more than everyone else together. Meet the record holders, and what each record taught me about its game.

Nine game icons standing on pillars like a podium, Plumb's tallest in the middle at 4:51:25, the others stepping down to each side, and Rinse and Seiche on empty dashed spots at the ends, on yellow

In five of my nine games, one person has played more than everyone else put together. In one of them, that person is the only one who has played at all. So I went looking for every record holder.

Here they are, one game at a time. Where they are, how long they played, how they played, and what each record told me about the game.

How I counted

Every number here comes from Google Analytics, for the four weeks from 17 October to 14 November. I left out Google’s test phones and my own phone, because together they make up most of what Analytics calls players. How I found that out is its own story: Google said 1,000 people played my games. It was 83.

The Google Analytics exploration called Games: real players, set to 17 October to 14 November 2026, with rows by country and game sorted by time played: Israel and Plumb on top with 1 user, 15 sessions and 4h 51m

A record holder here is one phone in one country. Analytics shows me the country, never a name, and the country is all I will ever say about anyone. The time is how long the game was open on their screen.

One more thing before the board. Five of my nine games are still in testing on Google Play: Hako, Rinse, Seiche, Sunwise and Stitchfall. Anyone who played those joined my testers group first, so those records belong to testers.

A records board on yellow listing each game's longest player: Plumb 4:51:25 in Israel over 5 days, Carillon 29:55 in the Philippines in 1 sitting, Stitchfall 20:35 in South Korea in 1 sitting, Hako 18:51 in Spain over 5 days, Pyxis 14:44 in South Korea over 12 days, Sunwise 10:39 in India over 4 sessions, Apsis 9:58 in India in 1 sitting, and Rinse and Seiche with nobody yet

Here is where those numbers come from.

A Google Analytics table of real players by country and game, sorted by time played: Israel, Plumb, 1 user, 15 sessions, 4h 51m, then the United States and India for Plumb, the Philippines for Carillon with 1 user, 1 session and 29m 55s, South Korea for Stitchfall with 20m 35s, Spain for Hako with 18m 51s, South Korea and Taiwan for Pyxis with 14m 44s and 14m 24s, India for Sunwise with 10m 39s, and Switzerland for Hako with 14 sessions and 7m 34s

A row with one user is one person. A row with more users is everyone in that country added up, so for Apsis, whose record holder shares India’s row with four others, I counted phone by phone.

Plumb: 4:51:25, Israel

The biggest record is not even close. One player in Israel played Plumb for 4 hours, 51 minutes and 25 seconds, across 15 sessions on five different days.

That is more time than my other eight games have had from everyone, all added up. It is 64% of all the time anyone has spent in Plumb. It is 41% of all the time anyone has spent in any game I have made.

And then they uninstalled it. That story needed a whole post of its own: A player spent 5 hours in my game, then deleted it.

Second place in Plumb is a player in the United States, with 1 hour and 14 minutes over four days. Plumb has had about seven and a half hours from real players, more than the other eight games together.

Plumb is a marble puzzle. You tilt the board, every marble rolls at once, and each one has to drop into the hole of its own colour. Its record tells me Plumb can hold someone for hours. What it cannot do yet is get found.

Carillon: 29:55, the Philippines

Carillon’s record is one sitting. One player in the Philippines opened it once and played for 29 minutes and 55 seconds without a break. Later that same day, they uninstalled it.

That one sitting is 74% of all the time Carillon has had from real players. Ten other people have played it, and together they have not reached 11 minutes.

Carillon is a chain-reaction puzzle. You tap one dot, its ring pops the dots near it, and every pop is a note. Half an hour in one go tells me the chains and the notes can pull someone in. Uninstalling the same day tells me something did not pull them back. Analytics cannot tell me what. I really wish it could.

Stitchfall: 20:35, South Korea

Stitchfall is my only game where you run and jump. You turn gravity a quarter turn at a time to gather buttons and reach each zipper exit.

Its record holder, a tester in South Korea, played it once, for 20 minutes and 35 seconds straight. That is 55% of everything Stitchfall has had.

Second place went the other way. A tester in the United States played for 10 minutes and 45 seconds, but spread across five different days.

A long single sitting is what I expected from a platform game. You find the rhythm and you keep going. Five short visits is what I hoped for and did not expect.

Hako: 18:51, Spain

Hako is about Japanese puzzle boxes. You slide wooden panels along their grooves until the lid comes free.

Its record holder, a tester in Spain, played on five different days, six sessions in all. They also pressed the hint button twice. After the Plumb player who never asked for help once, seeing someone ask made me oddly happy.

Hako’s second place is my favourite line on this whole board. A tester in Switzerland opened Hako 14 times, on ten different days, for 7 minutes and 34 seconds in total. Some days that was more than two minutes. Near the end it was a few seconds.

Hako’s boxes are small, and it looks like they suit a short visit. That is worth knowing before Hako comes out for everyone.

Pyxis: 14:44, South Korea, by 20 seconds

Pyxis has the closest race on the board. A tester in South Korea has 14 minutes and 44 seconds. A tester in Taiwan has 14 minutes and 24 seconds. Twenty seconds apart.

Both started while Pyxis was still in testing, and both kept coming back. In these four weeks, the player in South Korea opened it on twelve different days, and the player in Taiwan on eleven.

How they came back is completely different. The player in South Korea did most of their playing in the first three days, then came back only for short visits, and then uninstalled it. The player in Taiwan kept up a minute or two on most days and, as far as I can tell, still has it.

Pyxis is an untangling puzzle. You pull the knots of a web apart until no strand crosses another and the web turns green. It is also my most evenly played game. Eighteen people have played it, and its record holder has only 17% of its time. Pyxis does not get sittings. It gets visits.

Sunwise: 10:39, India

Sunwise is my one-thumb arcade game. You hold anywhere to swing your planets round a small star and smash what flies at it.

Its record holder is in India, with 10 minutes and 39 seconds over four sessions. Here I have to be honest. The same phone model, in India, shows up in five of my games. That is either one very loyal stranger or, much more likely, a friend or one of my testers.

That is Sunwise’s whole story so far. Three phones have opened it, and the other two have zero seconds. The record holder is 100% of Sunwise.

Apsis: 9:58, India

Apsis is an arrow puzzle. Arrows rest on orbits around a sun, and each one you tap flies straight out if its path is clear.

Its record holder, in India, played it once, for 9 minutes and 58 seconds, and has since uninstalled it. Apsis is the only one of my games Analytics has seen on that phone, so I cannot guess whether that was a stranger or someone I know.

Apsis has the flattest top of any board. Second place, also in India, has 9 minutes and 3 seconds over six days. Third, in the United States, has 8 minutes and 5 seconds, and it looks like the same player who came second in Plumb. The top three are within two minutes of each other.

Ten people have played Apsis, and nobody has gone deep. That is the record I most want to see broken.

Rinse and Seiche: nobody yet

Now the strange part. Rinse and Seiche have no record at all.

Two phones have installed each of them. Each phone opened the game once, and Analytics measured zero seconds of play. One of the two has already uninstalled each game.

I checked that it is not a bug. My own phone shows up in both games just fine, with all of my own play counted.

Both are in testing, and anyone in my testers group can install them, the same as Hako and Sunwise. Testers picked Hako. Nobody picked Rinse or Seiche.

Rinse is a cleaning puzzle. You pour water to wash every tile clean before the tank runs dry, and dirty water runs downhill. Seiche is a pond puzzle. Every ripple’s ring touches exactly one fish, and you work out where the rings meet.

I like both of them a lot, and that is exactly why the zero hurts. Zero is not even a bad review. A bad review would mean someone played long enough to have an opinion.

So my question for these two is not how to make them better. It is why nobody stayed past the first screen. Next time I write to my testers, I will ask the ones who skipped them. If the answer is the name, the icon or the first screen, those are things I can fix.

A Google Analytics table of real players by game: Plumb 21 users and 7h 34m, Pyxis 18 and 1h 28m, Carillon 11 and 40m 26s, Apsis 10 and 41m 21s, Hako 9 and 33m 56s, Stitchfall 7 and 37m 32s, Sunwise 3 and 10m 39s, and Rinse and Seiche 2 users each with 0s

Rinse's and Seiche's icons standing on two empty pedestals on yellow, each pedestal marked 0:00 and nobody yet

And if you want a record of your own, this is the easiest one I will ever offer. Join the testers, open Rinse or Seiche, and play for one minute. Right now, that would be a record.

What the board taught me

One person can be the whole game

In five of my nine games, one person has played more than everyone else together. At this size, a game’s audience is not a crowd. It is a few people, and sometimes one.

Bars for seven games showing how much of each game's play is its single record holder: Sunwise 100%, Carillon 74%, Plumb 64%, Hako 56%, Stitchfall 55%, Apsis 24% and Pyxis 17%, with a dashed line at half

That changes how I think about every change I make. When I make Carillon better, I am making it better for people who have not found it yet. The person who played it most has already gone.

A sitting or a visit

Carillon, Stitchfall and Apsis got sittings: one long go, and then nothing. Hako and Pyxis got visits: a little at a time, on many different days.

Seven players' time drawn as blocks, one block per day they opened the game: Carillon in the Philippines, Stitchfall in South Korea and Apsis in India as one long block each, and Hako in Spain and Switzerland and Pyxis in South Korea and Taiwan as rows of small blocks across five to twelve days, with no dates

I used to think a long sitting was the better sign. Now I am not so sure. A sitting tells me a game can hold someone. Visits tell me it fits into someone’s day.

The record holders leave

This is the line I keep coming back to. All four of my games that are out for everyone, Plumb, Carillon, Apsis and Pyxis, have lost their record holder. The person who played each of them the most has since uninstalled it. The three records still installed are all in games that are still in testing.

I do not read that as failure. People finish games, get bored, run out of space on their phones. But it does tell me where to look next. Not only at how people find my games, which I wrote about in How I earned $0 from 9 games, but at the second, third and fourth day. What is there for someone who has already played for an hour?

Beat a record

Every record on this board is small. Apart from Plumb’s, any of them could fall in one evening.

Plumb has the hardest record to beat. Carillon, Apsis and Pyxis are out for everyone too, and their records are all under half an hour. When the board changes, I will write it down.

On this page