Classic Solitaire Game

The situation

An India-based, VC-backed mobile publisher — 10 partner studios, 20+ titles — came to us with one of their games, a classic solitaire title. As is typical of the category, the game was almost entirely ad-monetized (~97%), with no real IAP economy.

Incent UA was working, but spend was capped — any incremental scale was unprofitable. They were trying to scale non-incent UA in parallel, but no non-incent channels looked profitable, even at modest scale.

The problem

ROAS was being measured incorrectly (in gross, not net!), but there were also other data-pipeline issues that prevented a clear understanding of ROAS per campaign and network.

On top of that, with ~97% of revenue coming from ads and no IAP economy, LTV was structurally constrained, leaving little room to bid up and win more installs.

What we did

We opened with our typical deep teardown — product, data, UA, ad monetization, and the competitive set — to build a list of hypotheses for high-ROI initiatives.

As always, we consulted our vault of past initiatives, including over 900 A/B test results, for perspective on what types of interventions were most likely to pay off.

The first job was the data: we cleaned up the UA data pipes, fixed how ROAS was measured, and redirected spend away from what was quietly losing money. That alone changed which campaigns deserved budget.

Then we ran 15 high-ROI experiments across product, UA, and ad monetization, segmenting incent and non-incent users so we could treat two very different populations correctly.

We also identified and attacked some technical issues constraining performance — ANRs and cold-boot times. Throughout, we taught the team to run the rapid-experimentation process themselves, so the methodology would outlast our engagement.

The result

• D14 ARPU rose 60% blended — +67% for incent users and +52% for non-incent.
• The winning experiments, each with strong statistical confidence:
  • Early-levels tuning +8.3% (incent, p .054)
  • Rewarded interstitials +8.2% (incent, p .031)
  • Onboarding stack +8.0% (incent, p .059)
  • Mid-game interstitials +6.3% (blended, p .014)
• Fixing ROAS measurement ended the unprofitable overspend and unlocked more profitable scale.
• The team left with a repeatable experimentation methodology to use on every title they build next.

Testimonial

"Turbine drove 15 experiments across product, UA & admon and helped drive D14 ARPU gains of 60%. But more valuable long-term, they taught our team to run that rapid experimentation process ourselves — a growth methodology we'll use on whatever we build next."

— Founder & CEO, VC-backed mobile publisher

Matt's take

"This smart, fast-moving team had data-infra issues, inflating ROAS signal and driving unprofitable spend. We deployed our test playbook, drove 15 high-ROI initiatives, and cleaned up UA to unlock more scale."