Pruning an iGaming push campaign in seven days
A working method for taking a broad casino push campaign from launch to a stable zone allowlist, without starving the optimiser of data.
Push is the highest-volume format in gambling and the one most often mismanaged. The failure mode is always the same: a buyer launches broad, panics at day two when the blended CPA looks terrible, and cuts so aggressively that the campaign never accumulates enough signal to find its good zones.
Here is the sequence that avoids that, assuming a casino offer, a Tier 2 GEO set, and server-to-server deposit tracking already in place.
Days one and two: buy data, not conversions
Launch broad with a frequency cap of one per subscriber per 24 hours and a daily budget you are comfortable writing off. You are not trying to hit CPA yet. You are trying to get at least 300–500 impressions into every zone that will eventually matter, because anything less will not distinguish a bad zone from an unlucky one.
Resist bid changes in this window. Every bid change resets the delivery pattern and mixes your samples together.
Day three: cut on click quality, not CPA
You still do not have enough deposits to judge CPA. You do have enough clicks to judge whether a zone sends humans. Cut on the upstream signals:
- Zones with click-through far above the format average. 12% on push is not a great creative, it is usually accidental or incentivised clicking.
- Zones with near-zero time-to-first-interaction on the landing page.
- Zones where the registration form is reached but never started.
- Anything already flagged by pre-bid filtering, which will show as credited invalid traffic rather than spend.
Older push subscriptions convert worse on average but are heavily over-represented in cheap inventory. Cutting them on day three removes most of your volume before you have measured whether the cheaper price compensates.
Days four to six: let deposits arrive
Deposits in casino lag registration by hours to days, and your day-four registration cohort is not finished depositing. Hold the surviving zones flat and let the postbacks land. This is the step most buyers skip, and it is why so many campaigns get killed one day before they would have turned.
Day seven: build the allowlist and bid per zone
Now you have deposit data against a stable zone set. Split the survivors into three groups and act differently on each:
- 01Zones clearing under target CPA: bid up 15–25% and let them take budget. These are your campaign.
- 02Zones within about 30% of target: hold and keep observing. Many of these move once creative rotates.
- 03Zones above that: pause, do not delete. Deleted zones lose their history, and supply changes; a zone that failed in July is worth retesting in October.
Convert the survivors into an explicit allowlist and run the next creative test against that list rather than against open traffic. From here the campaign is a creative-testing problem, which is a much easier problem than a supply-quality problem.