15 launches, dissected line by line

Half the budget went into cutting sources in the first two weeks. The remaining four weeks ran at half the starting price.

The start moved three weeks because of certification. Dates were counted from the certificate, so the weekly plan held.

The peak fell on a league round: placements were planned around the fixture list, not around convenient dates.

The first-deposit event was passed to the store from day one — the algorithm reached target cost by the end of week two.

The first two months produced almost nothing. By month six the channel delivered deposits without daily spend — which is the point of it.

Stream terms were agreed in writing before launch. Not a single broadcast had to be pulled afterwards.

Expensive per action and justified by average volume: the audience traded rather than watched.

Of 140 channels offered, 38 were selected — by views-to-subscribers, not by the seller’s rate card.

The cheapest line on the list: these people had already registered and verified, and none of that was paid for twice.

Of 4,100 domains, 3,700 went to the blacklist. The rest produced 89% of deposits.

KYC pass rate was 61%. Without that number the cost per sign-up would look twice as good and mean nothing.

22% remained on day seven. That is the real price — measured by entries it would look five times lower.

Push gave volume and almost no repeat deposits. In month two the budget shifted toward in-app.

Display produced nothing on its own, yet search converted a quarter better alongside it — so they were measured together.

A third of the trades arrived in months two and three after publication — the YouTube search tail only shows on a long window.