The Story

Seven Apps, a leading portfolio of top-ranking wellness and creative apps, wanted to scale Face AI, its rapidly growing AI-powered photo app. The goal: acquire users who don’t just install — but actually engage deeply and convert into long-term subscribers. To achieve this, Seven Apps partnered with Churney and became one of the early adopters of Meta’s Non-Purchase Value Optimization (NPVO) product. Churney supplied a high-quality predictive signal that allowed Meta’s new optimization model to target users who were most likely to deliver long-term value, even before they made a purchase. Within the first month, the impact was clear: +26% improvement in key campaign performance metrics by month 1 compared to Seven Apps’ standard Meta optimization.

The Goal

Goal

Face AI aimed to scale profitably on Meta by acquiring high-value users who would convert into paying subscribers — not just trial starters or early engagers.

‍

Challenge

Meta’s traditional optimization models are built around clear purchase signals. But subscription apps — especially trial-to-paid funnels — often don’t produce immediate purchase events.

This meant Meta’s standard algorithms would frequently optimize for shallow actions (e.g., installs or trials) rather than users with true revenue potential.

Face AI needed a way to give Meta a value signal earlier in the funnel.

The breakthrough

Solution

Churney built a predictive lifetime value (pLTV) model using Face AI’s historical behavioral and subscription data. The model identified early patterns — within the first sessions and first day — that indicated whether a user would likely become a paying subscriber.

Churney transformed this into a custom predictive value event, which was then integrated into Meta’s new Non-Purchase Value Optimization (NPVO) product.

This allowed Meta to:

Receive a high-quality, future-value signal
Optimize toward predicted subscription value — not just short-term conversions
Scale campaigns efficiently without waiting for delayed subscription events

Over several weeks, the predictive NPVO setup was compared directly against Face AI’s business-as-usual optimization.

‍

The Results

By focusing on long-term value instead of early funnel actions, Face AI achieved:

+26% improvement in campaign performance in month 1

The predictive signal dramatically strengthened Meta’s ability to identify and target users most likely to convert into subscribers.

‍

Why It Worked

Churney’s predictive modeling + Meta’s new Non-Purchase Value Optimization formed a powerful combination:

Smarter Value Targeting — Meta focused on users forecasted to deliver higher subscription revenue.
Earlier Optimization Signals — Churney provided reliable value predictions before a purchase occurred.
Efficient Scaling — Campaigns grew without sacrificing user quality or long-term monetization.
Predictive Revenue Focus — Seven Apps gained visibility into early behaviors that drive Face AI’s profitability.

Signal engineering for ad platform optimization is a discipline that sits at the intersection of data science, infrastructure, and measurement methodology. Most vendors underestimate at least one of these components. When we evaluated Churney on pLTV integration with Meta's NPVO product, we focused on three criteria: Could they construct a predictive signal with appropriate calibration? Could they integrate it reliably into our data pipeline? And could they measure the outcome without bias? They delivered on all three. The team demonstrated fluency in the technical tradeoffs—signal timing, value magnitude, event incrementality and freshness that determine whether an optimization model receives actionable information or noise. Their infrastructure work was clean, and critically, they maintained proper holdout methodology so we could attribute results with confidence. We observed a 26% lift in campaign performance during the first month. That's a strong initial result, though the real test is sustained performance over multiple cohorts and varying market conditions. What differentiated Churney was execution quality. The pLTV optimization space has no shortage of promises; what's scarce is teams who understand the engineering and statistical rigor required to deliver on them.

İlker

CTO & Co-Founder, SevenApps

x

+26%

+0%

Higher