95% CPA reduction in 6 weeks
A FinTech startup was struggling to convert users into paying subscribers. Cost per acquisition sat stubbornly at $595—unsustainable for a startup trying to scale.
This wasn't a "sign up and go" product. The platform had been designed thoughtfully—a detailed onboarding flow that qualified users, ensured compliance, and set customers up for long-term success. The product experience was working. Paid acquisition wasn't.
The team had tried everything to fix it: refreshing ad creative, tweaking audience targeting, adjusting bids, testing new copy angles. Nothing moved the needle.
I was brought in to diagnose what was actually going wrong.
The Goal
Reduce customer acquisition cost to a sustainable level while scaling weekly conversions from single digits to triple digits—without shortcutting the funnel or sacrificing user quality.
The Strategy
Tactics
Funnel mapping and conversion path analysis, Mid-funnel event identification and correlation analysis, Meta Ads event optimization restructuring, Weekly performance dashboards with CPA trends
Funnel Diagnosis. I mapped the complete user journey and found the root cause: the purchase event was 32 steps deep. That thoughtful onboarding flow that qualified users and ensured compliance? It meant the final conversion event was happening only a handful of times per month. Meta's algorithm had almost nothing to learn from.
Every "optimization" they'd tried—new creative, adjusted targeting, bid changes—was rearranging deck chairs. The real issue wasn't the ads. It was the optimization architecture.
During the initial strategy session, I asked a question: What event are you optimizing toward, and how often does it fire?
The answer pointed me in the right direction. They were optimizing toward Purchase—which made intuitive sense. But how often was that event actually firing?
Event Restructuring. Working with the product team, I identified a milestone at step 19 that strongly correlated with eventual conversion. Shifting the optimization target to this event gave the algorithm a faster, stronger signal while maintaining correlation with downstream value.
Scaling Discipline. As CPA dropped, the client scaled daily budget 4x while maintaining efficiency—proving the system could grow without breaking.
The Results
| Metric | Before | After |
|---|---|---|
| Cost per Acquisition | $595 | $30 |
| Weekly Conversions | <10 | 100+ |
In this case the businesses was optimizing for the "right" event (purchase), but that's actually the wrong signal for the algorithm.
Meta's machine learning needs volume and speed to optimize effectively. When your conversion event is too many steps deep and fires only a few times a month, the algorithm is flying blind.
The fix isn't always "optimize for something cheaper." It's finding the event that:
Correlates with eventual value
Happens frequently enough for the algorithm to learn
Occurs early enough in the journey to influence targeting
The 32-step funnel existed for good reasons—compliance, qualification, user quality. The solution wasn't to gut the product experience. It was to find the right signal within that journey and let the algorithm do what it does best.