TrySolitaire Half-Year Report: 215,359 Games, and What Changed

TrySolitaire Half-Year Report: 215,359 Games in the First Half of 2026

Published: July 2026 Data period: February 1 – June 30, 2026 Sample: 215,359 games started, across 23 game variants Reports in this series: February · March · April · May · June

We’ve published a report every month since February. This one looks at all five together.

In that time TrySolitaire went from a small site to one that sees more than 75,000 games a month. But the growth isn’t the most interesting part. The most interesting part is what people played. In February, this was a Klondike site. By June, it wasn’t — not because Klondike shrank, but because everything else grew up around it.

Here’s the first half of 2026 in numbers.

The Big Number

215,359 games started. 68,246 won.

That’s the first half of the year, February through June. A note on the window: our tracking only started on January 29, so January is two days of data and we’ve left it out. When we say “the first half,” we mean the five full months we can actually measure.

Of all the games started, players won a little under a third. We’ll come back to the win rate — there’s a wrinkle in comparing it across these five months, and it’s worth explaining properly rather than putting a single number on the whole period.

The Growth

The Games column counts only games that sent us an ending, and tracking fixes in March and April raised that share, so the Growth column counts every game dealt, played or not, outside Mahjong and Sudoku. On that count the site grew 3.5x in five months. It grew by more than half in March and again in April. Then the growth slowed.

That’s not a warning sign. Each month is still the biggest we’ve had. But the shape changed — from fast growth to steady growth. A site that grows by half two months running is catching a wave. A site that keeps growing 13-28% a month after that is building something people come back to. The second pattern lasts longer.

What Actually Changed: Klondike Stopped Being the Whole Site

This is the real story of the half-year, and it’s a share story, not a volume story.

Klondike’s share of all games played, month by month:

Klondike climbed to 81% of the site by April. Then it fell to 50% by June — a 31-point drop in two months.

Here’s the part that matters: Klondike didn’t lose any players. Its raw volume went 35,848 in April, 38,012 in May, 37,846 in June. Flat. It held every game it had. The share fell because everything else grew faster — a lot faster.

In February, if you played a game on this site, three times out of four it was Klondike. In June, it was a coin flip. The site didn’t move away from Klondike. It grew a second half that isn’t Klondike at all.

What Grew: FreeCell Went 18.5x

The game that did the most to change the mix was FreeCell.

Growth counts every game dealt, played or not. Mahjong’s is left out: until July its count also held games from our sister site.

FreeCell was a rounding error in February — 291 games, less than Golf. By June it was the second most-played game on the site, behind only Klondike and ahead of Spider. Counting every deal, it grew 18.5x.

Most of that happened in two months. FreeCell sat under 700 games a month through April. Then its deals jumped almost sevenfold in May and grew another 70% in June. Two things drove it: the game started ranking better in search, and a wave of players arrived from China. Both are real and organic. FreeCell earns its own full write-up soon.

Spider deserves a mention too — it grew every single month without a stumble, 734 to 10,159, and now sits at number three. It’s the quiet, steady climber where FreeCell is the breakout.

So the diversification wasn’t one game. It was FreeCell leading and nearly everything else rising behind it, all at once, while Klondike held its ground.

Win Rates: The Real Baselines

We fixed a tracking bug on April 14 that had been undercounting abandoned games. Before that fix, win rates read higher than they should have, because games people quietly walked away from weren’t all being counted.

That means we can’t put one win rate on the whole six months. February through April are measured differently than May and June. So instead of a blended number that mixes two tracking methods, here are the clean baselines from May and June — two full months after the fix, on large samples:

A few of these are now solid enough to treat as settled. Klondike Draw 1 wins 27.9% of the time across 58,310 games — that’s about as firm a number as we’ll ever have for a solitaire game. Draw 3 wins 11.5%. If you play Klondike here, those are your real odds.

FreeCell leads the site at 57.7%. It’s a game where almost every deal is solvable, so a high rate makes sense — winning usually means you found the path, not that you got a good deal.

Spider’s difficulty spread holds firm. 1-suit 55.8%, 2-suit 31.3%, 4-suit 21.8% — a 2.5x gap between the easiest and hardest setting, steady across the whole clean sample. Same pattern we found in the spring, now on more games.

Mahjong Turtle wins 51.3%, with the same caveat we always attach: Mahjong lets you reshuffle the board up to five times by default, which raises its rate compared to card games that have no such option. It’s not directly comparable to Klondike or Spider.

What We Learned: Most Games End in Someone Leaving

The biggest thing we learned this half wasn’t about any one game. It was about how games end.

For most of the year we reported win rates and treated everything else as “not a win.” When we fixed our tracking in April, May and June let us split that open — and the split was surprising.

Across the open-ended games, here’s what happens after someone starts:

Klondike gets abandoned three times out of four. FreeCell only two times in five. The pattern is a mirror: the games people finish are the games people win.

And losses barely exist. In Klondike, across 75,858 games in May and June, players recorded a completed loss three times. FreeCell: zero. Spider: three. The reason is mechanical — in an open-ended game, you don’t lose, you get stuck. The board just sits there. You leave. So a game you didn’t win shows up as an abandon, not a loss.

That’s why win rate remains the number we trust most. It’s the one measure that stays fair whether a game ends by declaring a loss or by someone closing the tab.

What’s Next

Two things change the second half of the year.

First, we’ve consolidated. Our sister sites for Mahjong and Sudoku were retired earlier this month and now point to TrySolitaire directly. For six months our data quietly mixed traffic from those sites into two games’ numbers, Mahjong and Sudoku, and never the card games. It was about 6% of all games in June, but a third of Mahjong’s and two thirds of Sudoku’s. That’s over. From here on, every number is TrySolitaire and only TrySolitaire.

Second, that cleanup unblocks the posts we’ve been holding. A full look at what winners do differently, whether undo actually helps, and deep data portraits of Klondike and FreeCell — all of it needed clean, single-site data we didn’t have until now. Those are coming.

Six months in, the picture is clear. The site grew three-and-a-half-fold, stopped being a one-game site, and taught us that most solitaire games don’t end in defeat — they end in someone walking away, usually from a Klondike deal that wouldn’t break.

The most-played game of the first half was Klondike, holding half the table by June. Play it here →

Or try the game that grew 18.5x. Play FreeCell →

All data is anonymous. TrySolitaire does not track individual users. Win rates are calculated as wins divided by total starts (treating abandons as non-wins). Data covers February 1 – June 30, 2026; January is excluded because tracking began January 29. Win rate and abandonment figures are drawn from May and June only, the two full months after our April 14 abandon-tracking fix; earlier months undercounted abandoned games and are not directly comparable. Volume, growth, and popularity figures cover the full February–June period. Growth from February, March or April to a later month counts every game dealt, played or not, except Mahjong and Sudoku, whose counts until July also held games from our sister sites. The other game counts here include only games that sent us an ending (a win, a loss or an abandon), and tracking fixes on March 14 and April 14 raised that share, so those counts grew faster than play did. Until March 5 the first game of some page loads went unrecorded, so February’s counts are about a tenth low. Mahjong allows up to 5 reshuffles by default, which raises its win rate relative to games without that option; only the Turtle layout is reported. TriPeaks, Golf, and Pyramid each include a smaller “standard” sub-variant we are still characterizing; their rates are the game-level blend. May and June figures include games from now-retired sister sites (Mahjong and Sudoku only, no card games): at least 4% of starts in May and about 6% in June. That is about a third of Mahjong’s games and two thirds of Sudoku’s, so the Mahjong (Turtle) and Sudoku baselines above are partly, and for Sudoku mostly, sister-site players. For game rules and strategy, see our Klondike Solitaire Guide, FreeCell Guide, and Spider Solitaire Guide.

Published July 2, 2026 | TrySolitaire Blog · Play Free Solitaire