The whole loop, one team
Media buying and creative production fail separately: buyers blame assets, studios blame bids. Campaign mode closes the loop by making one team own both, bid against your retention cohorts, and answer for the blended number.
What we bid against, and why it isn't installs
An install is a purchase of a maybe. Across verticals, day-1 retention averages about 26% and day-30 about 7%, which means the median install stops existing within a month. So the loop optimizes cohorts: cost per retained day-7 user, early ROAS signals, payback windows. On iOS that means living honestly inside SKAdNetwork 4's three postbacks and the AdAttributionKit transition rather than pretending the IDFA era is coming back.
The channel spine
Google App Campaigns carry Android volume and now offer tROAS bidding on iOS globally; Apple Ads (renamed from Apple Search Ads in April 2025) owns high-intent App Store real estate and adds multiple search-results slots from March 2026, which will reshuffle everyone's CPTs; TikTok and Meta supply the creative-driven discovery volume where our bench earns its keep. SDK networks join when the economics say so, with fraud filtering on from day one, since roughly 22% of non-organic installs carry fraud signals industry-wide.
Soft launch to global
For games, the loop starts narrow: a soft-launch ladder of two or three geos chosen for signal quality per euro rather than cheapness alone, creative testing running from day one, and a written set of gates (retention, early ROAS slope, CPI ceiling) that decide the global button. The puzzle-studio story walks the whole eleven weeks, gates included, and the soft-launch guide publishes the geo logic.
Campaign questions
Google App Campaigns and Apple Ads as the spine, TikTok and Meta as the creative-hungry flanks, and the SDK networks (AppLovin, Unity, Mintegral, Moloco) when scale demands them. Channel mix follows your economics: Apple Ads' cost per tap averages around $2.50 now, so it earns its slot on high-value apps and loses it on thin-margin ones.
8% of managed spend with a €3,500 monthly minimum, creative bench included. Mobile UA agencies typically land between 5% and 15% of spend; we sit low in that band because the creative work, priced separately elsewhere, is what makes the media side efficient enough to be cheap. No setup fee after a working session.
SKAdNetwork 4 postbacks where networks support them, AdAttributionKit as it rolls out, and conversion schemas designed around your day-1 signals, since with ATT opt-in averaging 35% you bid mostly on modeled data. We set the schema before scaling a euro; bidding on a broken signal is the fastest way to buy 30,000 wrong users.
No. We want games and consumer apps where retention curves exist to bid against: if your D30 rounds to zero and the roadmap won't change that, UA spend is a subscription to a leaky bucket and we'll say so in the first session. The bar isn't genre; it's whether the product keeps what we catch.
Hand over the loop
The working session covers your retention curves, your channel mix, and the first three things we'd change. Free, and occasionally uncomfortable.
Start a run