AppLovin, AI Bowls, and the Measurement Stack Running Ollie’s Growth

Ollie’s algorithm doesn’t just calculate portion sizes — it incorporates live health data from the app (body condition, stool condition, teeth health) to continuously optimize what and how much a dog should eat. This creates a flywheel where the more a customer engages with the app, the more personalized and accurate the product becomes. That depth of personalization is what makes it genuinely hard to churn off: you’re not just canceling a food delivery, you’re walking away from a health system built around your specific dog.

Ollie’s most recent performance surprise has been AppLovin — placing ads in native mobile gaming environments. Kalina’s read on why it works: Ollie’s audience is already primed for health-app engagement (they use Whoop, Apple Watch, track their own metrics), and that audience over-indexes in mobile gaming. The alignment between the product experience (app-native, data-driven, wellness-oriented) and the AppLovin context is producing strong performance signals that Meta alone couldn’t surface.

Ollie ran an AI-generated ad that depicted their dog bowl reshaped into a heart. It went viral on Meta — high engagement, strong comments, genuine consumer excitement. The team’s response was not to ship more AI ads; it was to manufacture the heart-shaped bowl as a real product and offer it as a gift in the member experience. Kalina’s framing: AI is a powerful signal detector for what resonates — but the best outcome is when the human team takes that signal and makes something real with it. Keeping the human in the loop is the philosophy.

The Van Leeuwen partnership — a dog ice cream co-created with a premium human ice cream brand and sold in Van Leeuwen retail locations — worked on every dimension: awareness, acquisition, and retention gifting. Kalina’s view is that the future of marketing increasingly looks like partnerships with brands that share the same consumer and the same values. The dog wellness ecosystem (Embark for DNA, Fi for GPS wearables) gives Ollie both distribution surface area and credibility through association.

Ollie runs a weekly measurement loop that combines media mix modeling (macro channel attribution), multi-touch attribution (journey-level), and first-party surveys asking customers directly how they heard about the brand. No single layer tells the full story. The MMM tells you which channels are driving incrementality; the MTA shows you the path; the survey catches what both models miss. Kalina’s team uses this triangulation to set expectations before every major channel bet and then assess against those expectations once data comes in.

Ollie’s 2026 strategy requires holding two things at once: investing significantly in brand awareness and TV for long-term reach, while maintaining the payback metrics (CAC, PNL health) that keep the business sustainable. Kalina’s approach is to pre-name the expected outcomes for every awareness bet — what good looks like, what off-track looks like — so the team can evaluate whether the bet paid off rather than defaulting to either optimism or anxiety when the data arrives.