Portfolio Home
Case Study 01 · Furbo · 2017–2024
Furbo AI & Lifecycle
Systems
How I identified the app — not the hardware — as where Furbo's subscription value was earned or lost, and built the AI trust and lifecycle systems that sustained it across Spanish-speaking markets for 7 years.
Senior Product Strategist
·Furbo / Tomofun
·2017–2024
·Iberoamérica
AI trust systems · localization governance · lifecycle strategy · regional product operations
1M+
Global Users
Incl. 500K+ Google Play downloads, with Spanish-language traction across Iberoamérica
~30%
Revenue from Subscriptions
≈$6M GMV in 2024
#1
Amazon Category 2018–2024
00 · Scope & Authority
What I Owned
Owned the full regional system for Iberoamérica — subscription onboarding logic, AI alert quality, behavioral localization, and lifecycle communication across all product lines
Defined quality standards and rollout governance model for AI-generated alerts — including confidence thresholds, tier classification, human-in-the-loop validation, and release criteria
Led cross-functional execution across engineering, marketing, and CX — with authority to approve or block releases
Full ownership of the Spanish-language subscription experience across Iberoamérica — from onboarding through renewal, across every alert, launch, and market, for 7 years
01 · Context
Hardware + Subscription
Furbo Dog Camera sells a pet monitoring device with a free 1-month subscription trial. After that, renewal is optional — and entirely dependent on whether the app feels worth paying for. I joined in 2017 with ownership of brand strategy, app UX, and AI alert integration for Iberoamérica. Early on, I identified that the hardware drove acquisition, but the app drove revenue. Every cold or mistranslated alert was a direct threat to LTV. That insight anchored 7 years of product decisions.
02 · The Problem
A Revenue Risk Hidden Inside a Product Experience Gap
Furbo's recurring revenue depended entirely on one decision: whether a user who got a free month with the camera chose to renew their Furbo Dog Nanny subscription. Expansion into Spanish-speaking markets exposed a structural gap: the AI-generated alert system had no behavioral or emotional logic designed for the context of pet ownership in Iberoamérica. Alerts were machine-translated — producing cold, tonally flat messaging that eroded the perceived value of the subscription at every interaction. There was no classification system, no quality threshold, no regional adaptation layer. Every notification was a moment of truth for the subscription model. And most were failing.
"Every notification was a moment of truth for the business model. The alert had to feel worth paying for — in Spanish, in the right tone, every single day."
03 · What I Did
Four Decisions That Built the System
I designed and owned a trust architecture — a system governing how every alert was classified, validated, and delivered across Iberoamérica. Four decisions defined it:
Decision 01 · Problem Framing
The App Was the Product
The hardware drove acquisition. The subscription drove revenue. I reframed the entire Iberoamérica strategy around that insight — every lifecycle decision, every quality gate, every product operation existed to protect one thing: the moment a user decided whether the app was worth renewing.
Decision 02 · System Design
Classification System Over Case-by-Case
Instead of reviewing alerts reactively, I built a tiered classification system: four behavioral categories (Routine / Sensitive / Health / New Features), each with defined approval workflows, human-in-the-loop validation, regional ownership, and a 90%+ perceived clarity threshold. No alert shipped without clearing the gate.
Decision 03 · Rollout Governance
Authority to Block
I held and exercised veto power over regional launches. When a 2019 Dog Nanny alert tested at 82% clarity — below threshold — I blocked the global rollout over marketing and engineering pressure. Two weeks of iteration. 93% on retest. No rollback. No deterioration in regional ratings post-launch.
Decision 04 · Trade-off
LTV Over Activation
Every pressure in a hardware+subscription model points toward speed: ship fast, activate users, hit the campaign date. I consistently made the opposite call — protecting long-term subscription trust over short-term activation metrics. The 4.7★ rating sustained across 7 years is the compounded result.
04 · Real Alerts
Real Alerts, Real Decisions
Three alert tiers — each shown as the original English, the literal machine translation (the failure state), and the regionally adapted version. Each column reflects a behavioral and localization decision with direct implications for how the subscription felt to use.
Official EN — Furbo's documented English
Literal ES — Machine translation (the problem)
Adapted ES — Trust architecture result
🟢 Routine — Barking Alert
✅ Official English
Real · Barking
🐕
Furbo Dog Camera
Now · Home
Max is barking. Furbo detects when your dog is barking — tap to check in on him now.
❌ Literal ES (Before)
Literal · Alerta
Max está ladrando. Furbo detecta cuando su perro está ladrando — pulse para verificar.
Ver cámara
Lanzar golosina
✨ Adapted ES (After)
Adaptada · Ladrido
¡Max está ladrando! Furbo te avisa para que puedas comprobarlo enseguida y calmarlo. 🐾
Routine tier. "Lanzar premio" vs "Lanzar golosina" — positive reinforcement framing in the user's emotional register. Uses tú throughout, signaling closeness from first alert.
🟡 Sensitive — Crying / Abnormal Behavior
✅ Official English
Real · Abnormal
😢
Furbo Dog Nanny
4:18 PM · Home
Luna is crying or whining. Furbo detected this as an abnormal behavior — she may need your attention.
❌ Literal ES (Before)
Literal · Comportamiento
😢
Furbo Dog Nanny
4:18 PM · Casa
Luna está llorando o gimiendo. Furbo detectó esto como un comportamiento anormal — puede necesitar su atención.
✨ Adapted ES (After)
Adaptada · Comportamiento
😢
Furbo Dog Nanny
4:18 PM · Casa
Luna está llorando o quejándose. Puede que te necesite — asómate para ver cómo está. 🧡
Asómate ahora
Lanzar golosina
Sensitive tier. "Luna está llorando o quejándose" — affective copy vs. clinical label. Emotional warmth matched to the user's attachment, not just a feature description. "Asómate" vs "Verificar" — intimate register vs. cold bureaucratic verb.
🔴 Health / Emergency — Home Emergency Alert
✅ Official English
Real · Emergency
🚨
Furbo Dog Nanny
11:03 AM · Urgent
Home Emergency Alert. Furbo detected the sound of an alarm at home. Please check on Rocky immediately.
❌ Literal ES (Before)
Literal · Emergencia
🚨
Furbo Dog Nanny
11:03 AM · Urgente
Alerta de emergencia en el hogar. Furbo detectó el sonido de una alarma en casa. Por favor verifique a Rocky inmediatamente.
✨ Adapted ES (After)
Adaptada · Emergencia
🚨
Furbo Dog Nanny
11:03 AM · Urgente
⚠️ Alerta de emergencia en casa. Furbo detectó el sonido de una alarma. Revisa a Rocky ahora — llama a emergencias si es necesario.
Health / Emergency tier. Urgent without alarming — calibrated tone for a high-stakes moment. The literal translation ("Por favor verifique a Rocky inmediatamente") adds friction when the user needs instant clarity. The adapted version removes formality and delivers urgency cleanly.
05 · Impact
Results at Scale
Over 7 years, the alert quality system and localization governance compounded into outcomes visible across engagement, perceived value, and subscription durability.
+35%
Notification open rate post-adaptation (vs. baseline Q1 2018)
4.7★
App store rating sustained 2018–2024 (20k+ reviews)
Incl. 500K+ Google Play downloads, with Spanish-language traction across Iberoamérica
32k+
Emergency interventions documented
~30%
Subscription revenue share of total GMV — the business model the trust system was built to protect
Est. based on Grips Intelligence (furbo.com online GMV, 2024–2025). ≈$1.8M of $6M GMV.
90%+
Perceived clarity threshold maintained
7yr
Unbroken ownership through 5 hardware generations
Seven years. Five hardware generations. One subscription system that had to earn trust — in the right language, at the right emotional register, across every alert, every tier, every market in Iberoamérica.
The full case study goes deeper into the system —
governance logic, classification design, and the decisions behind the work.
Includes the alert tier framework, rollout governance model, regional adaptation rationale, and the 2019 blocked launch in detail.
Read Full Case Study →