The big picture
A run is one playthrough, from the first door to the boss. It takes about five minutes.
If 100 people open the game…
All people who opened it, scaled down to 100.
What the game looks like
Pick a path, walk the map, fight with cards, take a reward.
Month 3 against months 1 and 2
Day by day
Dotted lines mark the seven seasons. Jun 18 – 26 is the build, before anyone was invited; the first ad ran on Jun 27. Hover or tap any day.
Where players come from
Free players who found us on their own, per week
Search, links, word of mouth. No ads, no portal.
…but they are far better players
Out of every 100 new players from each source, how many…
Do they stay?
Came back the next day, and a week later
Share of new players, by the month they first arrived.
Next-day return, week by week
Each dot is everyone who arrived that week.
Do paid visitors actually play?
Out of every 100 people an ad brought in, by the month they arrived.
Ads: we stopped paying for clicks
Same money, different order to Reddit
Before (Aug 4 – 27): "bring me clicks". After (Aug 28 →): "bring me people who win a run".
What a month of ads costs and buys
Spend is approximate (daily budgets).
What we built
Seven seasons in three months. 77 releases, close to one every day. There is now plenty of game, which is why the next stretch is about making it better, not bigger.
Four bets, and how they landed
Not everything worked. The ones that did not still told us who the game is for.
The people who stuck around
Registered accounts
Committed players active each day
Committed = played on 2+ days, 10+ minutes a day.
What's next
Better before bigger. One number to move: how many players come back tomorrow.
Make the game feel better
More animation and punch, smoother on slow phones, and sharper cards that combo. No new cards for a while.
Measured by: next-day return, time in the first session.
Earn the first money
Try different ways to sell Diamond Keys. Never pay-to-win: money cannot change the odds of a run.
Measured by: first sales, buyers per 1,000 players.
Try short video
AI video is now cheap enough to test TikTok-style clips as a second way to reach people.
Measured by: cost per player who wins a run, against Reddit.
Grow free traffic
More presence in more places, and a community that talks to each other. Free players are our best players.
Measured by: self-found players per week (now about 130).
Review the ad budget
Once the "pay for winners" campaign has a full month of data, decide whether to spend less.
Decision due: early October.
Season VIII, later
It is planned. It waits until the game we already have is worth coming back to.
Gate: the day-after number moves first.
Worth adding to the plan
How these numbers were made
- Source: the game's own telemetry mirror (every tap is an anonymous event). Test devices are hidden. Days are Europe/Warsaw calendar days.
- Counted from Jun 18, 2026, the day coding started, to Sep 20: 3 months and 2 days. Months run from the 18th to the 17th (30, 31 and 31 days). Month 1 includes nine days of building before the first ad on Jun 27, so it is smaller by construction. Sep 18 – 20 are on the daily chart and in the top tiles, but in no month bar.
- Totals, the 100-people funnel, the daily chart and the return rates use the dashboard's own definitions.
- Source split: a device is "paid" if it ever arrived with one of our ad-campaign tags, "CrazyGames" if it came through that portal, otherwise "found us on their own". Devices that loaded the page but never opened the game are in no group. The dashboard's own "organic" count also includes those devices and the CrazyGames trial, which is why it reads higher.
- Month 3's week-later return only counts players old enough to have had a week.
- Ad spend: month 1 is the ad account total on Jul 24. Months 2 and 3 are summed from daily budgets, so treat them as close, not exact.
- CrazyGames' playtime, return and conversion figures are theirs, from their decision email.