PulseApp Reached 50K Downloads at €1.80 CPI Through Creative Testing
A fitness app broke through high CPIs by systematically testing 60+ creative variations, finding 3 winning formulas that cut acquisition costs by 64%.
Context
PulseApp is a fitness app offering personalized workout plans and nutrition tracking. Available on iOS and Android, with a freemium model — free basic features, €9.99/month premium subscription. The app had 12K organic downloads and was looking to accelerate growth through paid acquisition.
Previous attempts at paid user acquisition had been expensive. Running basic Meta and Google app campaigns with a handful of creatives yielded CPIs of €5.10 — too high given the LTV of roughly €18 per averaged install (including free users who never convert).
Funnel type: App install → trial → paid subscription Target CPI: Under €3 for sustainably profitable growth LTV per install (blended): €18
Constraints
- Budget: €8,000/month total across all platforms
- Tracking: Meta Events Manager and Google Firebase configured. SKAN for iOS with limited post-install visibility
- Creative: In-house designer could produce static and motion graphics but no UGC or live-action video
- App Store rating: 4.4 stars on iOS, 4.2 on Android — decent but not top-tier
Hypotheses & Experiments
Hypothesis 1: CPI is high because creative quality is low, not because the market is wrong
With only 4 active creatives, the algorithm had limited options. The winning creative formula had not been found yet.
Experiment: We built a creative testing pipeline producing 8-12 new variations per week. We tested across 4 dimensions: hooks (first 3 seconds), messaging themes (transformation, discipline, community, simplicity), formats (static, carousel, short video, UGC-style), and CTAs.
Hypothesis 2: TikTok can deliver cheaper installs for a fitness app
TikTok’s fitness community is massive and engaged. The native video format rewards authentic, fast-paced content.
Experiment: We allocated 20% of budget to TikTok with native-style creative — quick before/after transformations, workout clips, and “day in my life” formatted ads.
Hypothesis 3: Optimizing for trial start (not install) will improve LTV
Standard app campaigns optimize for installs. But an install without a trial start is worth almost nothing.
Experiment: We configured campaigns to optimize for the “Start Free Trial” in-app event instead of the install event, accepting higher CPI in exchange for higher trial rates.
Week-by-Week Progress
Weeks 1-2: Audited existing setup. Fixed Firebase event tracking gaps. Built first batch of 15 creative variations across hooks and messaging themes. Weeks 3-4: Launched creative test structure on Meta (5 ad sets, 3 creatives each). Started TikTok campaigns. First insights: “transformation” theme hooks outperformed “discipline” by 3x. Weeks 5-8: Identified 3 winning creative formulas. Scaled spend behind winners. Killed 40+ underperforming variations. CPI dropped from €5.10 to €2.80. Added Apple Search Ads for brand and competitor keywords. Weeks 9-12: Switched Meta campaigns from install optimization to trial-start optimization. Initial CPI spike to €3.50, then settled at €2.10 as the algorithm calibrated. Trial-to-paid conversion improved because the algorithm was finding higher-intent users. Months 3-4: Mature campaign structure. TikTok achieving €1.40 CPI (best channel). Meta at €2.10. Apple Search Ads at €1.90. Google App Campaigns at €2.30. Blended CPI: €1.80. Day-7 retention improved to 38% — the algorithm was finding better users.
Results
| Metric | Before | After (Month 4) | Change |
|---|---|---|---|
| Blended CPI | €5.10 | €1.80 | -64% |
| Monthly Installs | 1,570 | 12,500+ | +696% |
| Day-7 Retention | 22% | 38% | +73% |
| Trial Start Rate | 31% | 48% | +55% |
| Trial-to-Paid | 6.8% | 11.2% | +65% |
| Monthly Paid Subscribers (from ads) | 33 | 672 | +1,936% |
| Blended LTV:CAC | 3.5:1 | 10:1 | +186% |
What Did Not Work
- UGC-style content without real UGC: Our designer created “fake UGC” that looked polished. Authentic UGC (from actual users) performed 2-3x better. We shifted to sourcing real user content.
- Long-form explainer videos (30+ seconds): Anything over 15 seconds had terrible completion rates. The audience wanted quick value — show the app, show the result, done.
- Google App Campaigns alone: Google UAC was the most expensive channel. It works as part of a multi-platform mix but should not be the sole channel.
- Targeting gym-goers specifically on Meta: Counter-intuitively, broad targeting outperformed interest-based targeting. Meta’s algorithm found our best users when we gave it freedom.
Takeaways
- Creative volume is the unlock for app install campaigns. Testing 60+ variations in 4 months is what found the winning formulas. 4 creatives is nowhere near enough data.
- Optimize for deeper funnel events, not installs. When we switched from install optimization to trial-start optimization, CPI went up slightly but the users were dramatically more valuable.
- TikTok deserves real budget for consumer apps. It was our best-performing channel by CPI, but only when creative felt native to the platform.
- Day-7 retention is the north star metric for app growth. If people do not come back, no amount of ad optimization will build a sustainable business. Our campaigns improved retention because the algorithm was finding better-fit users.
Want Results Like These?
Book a free strategy call and we will show you exactly how we would approach your campaigns.
Book a Free Strategy Call