Forum

Browse every card by deck, every monster by biome, and every relic by rarity — with the real numbers the engine uses, not the printed ones.

The whole roster, grouped by the deck that leans toward it.

Every foe behind every door, its move kit, and how it scales with depth.

The passive items a run hands you, and exactly when each one fires.

Asking Reddit for players who win a run, not just clicks, cut the price of a winner from $2.58 to $1.07. Now the same goal and the same money go to TikTok, with ten short videos in three styles. In plain words and pictures: the videos, the budget, the calendar, and five guesses we wrote down before the first result.

56,019 players, 71,178 runs and 7.4 million logged moves, counted from the first line of code on June 18, and not a dollar charged. Month three held the level month two reached, paying Reddit for winners instead of clicks made 2.3 times more of them, a big games portal turned us down, and the number that matters has not moved: 2 in 100 players come back the next day. In plain words and pictures, with what comes next.

Where our advertising money goes, in pictures. For six rounds we paid Reddit for clicks and it never learned whether a click became a player, so it found the cheapest taps on the internet. Round 9 pays for a finished run instead: a player who wins one comes back seven times more often than one who only clicked. The ten adverts, the settings, the month, and the single number that decides whether it worked.

30,418 players and 3.78 million logged decisions. Season V doubled the world, we deleted our own home screen after three rounds of testing, and one switch of ad account halved the share of arrivals who play. The returning base tripled while first-timers got worse, and we can name the day it happened.

We have called one ad our champion for two rounds. Checking our own record, it was never significant, and the round meant to settle it was decided by the ad platform rather than by us. Round 7 is the mirror of Round 6: freeze the audience, move the creative, equal budgets, a month. Five predictions before a dollar is spent.

The readout, and the scoreboard we promised. One identical ad, three audiences, $367. The pool that has seen our ad for four rounds produced the best click-through in the round and the worst players in it, by a factor of two. All five predictions graded: two were wrong, and one of the wrong ones is the finding.

Five rounds taught us which ad wins. Round 6 freezes the champion and moves the only thing we have never moved: who sees it. The full targeting lists, the tracking line we refuse to cross, and five predictions written down before a single dollar was spent.

A million logged decisions, 10,171 players, two full seasons, $2,000 burned and $5 earned. Activation, the retention split that reframes everything, the two leaks that own the funnel, and six predictions written down so they can be checked.

Nine ads, three lanes, one week. The three that won, the copy pattern they shared, and the promised 7-day read on Round 3: retention came in at 1.14%, statistically identical to the campaign those hooks beat. Better hooks buy better first days and zero extra second weeks.

6,622 players, 6,638 runs, 4,894 finished. Activation, retention, the two leaks costing us most, and the 257 runs started on the hardest rung in the game.