**Customer:** AI Companion App · Mobile Subscription **Published:** 2026-05-13 **Canonical:** https://cogny.com/case-studies/ai-companion-app-paid-acquisition ## Headline results - **New User Growth: 80×** — Direct result of building paid acquisition from zero - **CPA: ~$1.61** — vs. $30–$100 industry benchmark - **Video CTR: 15.5%** — vs. 1–3% industry average ## The challenge An AI companion app had no paid acquisition engine. No Meta Pixel, no campaigns, no conversion tracking. Growth depended entirely on word of mouth. ## The solution Cogny diagnosed the missing Meta Pixel as the #1 blocker, built an analytics warehouse in BigQuery, then launched and continuously optimized US-focused Meta Conversion campaigns, shifting budget to the highest-performing creatives in near real time. ## Background An AI companion app where users create personalized AI avatars through conversational onboarding. Before Cogny, the product had no paid acquisition engine of any kind. No Meta Pixel, no campaigns, no conversion tracking. Growth was 100% organic, driven by word of mouth. **Cogny took over paid acquisition on March 30, 2026.** This case study measures the acquisition results across two equal time windows: **89 days before Cogny** (January 1 – March 29) vs. **45 days with Cogny** (March 30 – May 13). ## What Cogny Did The first action was diagnostic, not tactical. Without a Meta Pixel firing conversion events, Meta's algorithm has nothing to optimize toward. Every dollar of ad spend would be flying blind. Fixing the tracking foundation came before any campaign launch. Cogny's work across the 45-day window: - **Diagnosed the missing Meta Pixel** as the #1 blocker to paid acquisition - **Built an analytics warehouse in BigQuery** for cross-channel attribution - **Structured and launched paid Meta campaigns** once the pixel went live (April 27) - **Ran continuous optimization**, shifting budget to the US market and highest-performing creatives - **Targeted men 35–44 in the US** with a Conversion campaign objective ## The Acquisition Explosion The results from the US Conversion campaign (April 14 – May 13): | Metric | Value | Context | |--------|-------|---------| | CPA | **~$1.61** (17.26 SEK) | vs. $30–$100 industry benchmark | | Video CTR | **15.5%** | vs. 1–3% industry average | | New user growth | **80×** | vs. pre-Cogny organic baseline | The **80× jump in new users** is a direct, attributable result of Cogny building and launching the entire paid acquisition engine from scratch. Not an optimization on top of existing campaigns. There were no campaigns to optimize. ## Why the CPA Is This Low A $1.61 CPA in a category where competitors pay $30–$100 doesn't happen by accident. Three things compounded: 1. **Tight targeting from day one.** US-focused, men 35–44, Conversion objective. No broad-and-pray phase, no wasted learning budget. 2. **A 15.5% CTR on video creative.** That's 5–15× the industry average. The algorithm rewards high engagement with cheaper inventory. 3. **A fresh pixel learning on a clean signal.** No legacy event noise, no bad historical optimization data. Meta's model started learning on the right outcome from the start. The compounding effect: high CTR drives lower CPM, lower CPM amplifies the conversion advantage, and the algorithm doubles down on what's working. ## Fast Targeting Optimization The headline metrics are the outcome, not the method. The method was speed of iteration. Cogny ran the account on a continuous loop rather than a weekly check-in cadence: - **Geo budget reallocation.** Once US performance pulled away from other markets, budget shifted to the US within the same week, not after a monthly review. - **Creative-level pruning.** Video variants were ranked by CTR and conversion rate continuously. Underperformers were paused; budget concentrated on the winners. - **Audience-level signal reading.** Men 35–44 emerged as the strongest cohort early. The bidding strategy doubled down on that signal rather than diluting spend across broader demographics. - **Real-time pixel-feedback loop.** Every conversion event fed straight back into Meta's optimization, accelerating the algorithm's learning curve instead of waiting on batched reporting. A human-only agency reviewing the account weekly would have made the same calls eventually, three to four weeks later. In a 45-day window, that lag is the entire campaign. ## Key Takeaways **For early-stage subscription apps considering paid acquisition:** - **Tracking is the foundation, not the polish.** A missing pixel isn't a minor oversight. It's the difference between paid acquisition working and burning cash. Fix it before anything else. - **You don't need a giant budget to validate.** A 45-day window with a single well-structured US campaign produced conversions at $1.61 each, in a category where the benchmark is $30–$100. Spend less, learn faster, then scale what works. - **Speed of optimization is the moat.** The CPA and CTR advantage came from acting on signal in days, not weeks. Most agencies can pick the right knobs. Fewer can turn them fast enough to matter. - **Tight targeting beats broad reach.** US-focused, single demographic, Conversion objective, and the algorithm did the rest. There is no learning budget to spend on audiences you don't actually want. ## What Comes Next The acquisition engine is built and running. The next 45 days are about scaling it: 1. **More creative variants** to feed the algorithm and fight ad fatigue at higher spend levels 2. **Adjacent audiences** beyond the validated US 35–44 cohort, using the existing winning creatives as the anchor 3. **Expanding geo coverage** to additional high-intent English-speaking markets once the US channel saturates A 45-day window took the business from no paid acquisition to a working engine running at sub-$2 CPA and a CTR 5–15× the industry average. That's the foundation. Now it scales. --- Source page: https://cogny.com/case-studies/ai-companion-app-paid-acquisition All case studies: https://cogny.com/case-studies Cogny for agents: https://cogny.com/llms.txt