PixelForge Games: Install Cost Down 60% and Retention Up Across All Geos
How a hybrid-casual studio rebuilt mobile UA around D7 ROAS instead of D1 install CPA, and why that single shift changed the entire game.
The Challenge
PixelForge had cycled through five agencies in two years. Every one of them treated mobile UA the same way: install CPA was the headline KPI, the buying optimization event was the install, and the creative team chased D1 install volume above all else.
What the studio actually cared about was D7 ROAS, installs that retained, played, and monetized through ad-monetization plus IAP. They had a great game and bad players coming in the door. The fix was not a creative fix. It was a buying-optimization fix that required rebuilding the entire UA stack from the SDK level upward.
Our Approach
SDK + measurement rebuild around D7 events
We rewired the AppsFlyer setup to send richer post-install events, session-2, level-completion, ad-impression-served, IAP, back to Meta, TikTok and Google App as bid signals. Buying moved off the install event onto a post-install retention proxy event.
Creative segmentation by retention archetype
We segmented the install base by what made them retain. Players who returned for the meta-game got one creative lane; players who returned for the core loop got another. Creative production split into two parallel tracks against those archetypes, and the ratio of retaining-installs climbed every month.
Soft-launch geos as the testing surface
We used PH, ID and TH as the soft-launch testing geos for new creative concepts before promoting them to NA and EU. Cost-per-install in soft-launch is a fraction of NA, and the retention-curve signal is strong enough at low spend to call winners early.
Hybrid-casual ad-monetization stack
Worked with the LiveOps team to tune the rewarded-video and interstitial cadence so retention did not break under the weight of ad load. The buying side cannot fix monetization decisions made by the studio, but it can tell you when monetization is killing your retention curve.
Execution Timeline
Results
- ✓Install cost fell 60% on a blended basis, driven less by CPI compression and more by the mix shift toward D7-retaining installs.
- ✓D7 retention lifted 38% versus baseline. The game looked the same. The players coming in were different.
- ✓Active geos expanded from 3 to 7 over the engagement window. Each new geo went through the same soft-launch validation cycle before scaled spend.
- ✓Studio reported that ad-monetized ARPU in the retained cohorts was approximately 2.1× the previous cohort average, meaning the install was worth roughly twice as much, on top of being cheaper to acquire.
Results
We've worked with 5 agencies before Unled. None understood mobile gaming UA the way this team does. Install costs dropped 60% and retention improved across all geos.
What This Engagement Taught Us
- ›Optimizing on installs in 2026 is malpractice for any game with a non-trivial ad-monetization or IAP revenue line. Optimize on a post-install retention proxy.
- ›Soft-launch geos are the cheapest creative testing surface in mobile UA. Use them.
- ›UA cannot fix bad LiveOps decisions, but it can be the early-warning system that tells the studio when ad-load is killing retention.
More Client Stories
Bring Us a Real Account
Show us your numbers. We will tell you what we would do differently, and whether your account is one we can move. No deck, no pitch.