Growing Wearable Engagement Through Experimentation

Context

Over two years on Meta’s Wearables Onboarding, I partnered with Product Manager, Engineering, Data Science, UX Research, and Content Design to continuously improve the first-time user experience. Rather than treating onboarding as a fixed flow, we treated it as a product that could be measured, tested, and optimized.

Every experiment began with behavioral data, a clear hypothesis, and measurable success metrics. Together, these experiments increased activation, feature adoption, and long-term engagement across millions of AI glasses users.

Hypothesis

Users are more likely to connect optional features when they can quickly understand their value and only see features relevant to them.

The original experience presented optional setup as a text-heavy list, making it difficult for users to understand the value of each feature or quickly identify what was relevant to them.

I redesigned the experience into a visual, card-based flow that surfaced each feature’s value upfront while dynamically prioritizing recommendations based on the user’s context. By making the experience more glanceable and personalized, users could make faster decisions with less cognitive load and fewer unnecessary steps.

Impact

• +23% provider linking
• +2–4% Communications WAU
• ~10% Partner Music WAU
• Statistically significant improvements across every shipped provider

Hypothesis

Users acquiring a previously owned device shouldn’t discover ownership issues halfway through onboarding. Detecting registered devices early and guiding users through ownership transfer upfront would reduce friction and improve setup completion.

Previously, users pairing a device that was still registered to another account didn’t encounter the ownership issue until much later in onboarding. This unexpected interruption forced them to leave the flow, unpair the device, perform a factory reset, and restart the entire setup process, resulting in significant drop-off.

I redesigned the experience to detect previously registered devices immediately after discovery. Users were guided through unpairing and factory reset before onboarding began, and once the device returned to discovery mode, pairing resumed automatically. By resolving ownership transfer upfront, the rest of the onboarding experience became uninterrupted.

Impact

• +3.3 percentage point increase in 24-hour setup completion

Hypothesis

Users are more likely to adopt Meta AI when onboarding demonstrates practical, everyday use cases instead of generic voice commands.

Working closely with Data Science, I redesigned the Meta AI Tour by introducing an additional education screen featuring data-informed sample prompts that highlighted high-value, real-world scenarios. Rather than simply teaching voice commands, the experience helped users discover meaningful ways to control their device and interact with Meta AI from day one.

Impact

• +1.0% AI Daily Active Users
• +1.7% AI Weekly Active Users

Hypothesis

Users are more likely to adopt music features immediately after successfully linking a music provider than during optional feature setup.

Instead of ending the experience after users linked their music provider, I introduced a contextual education moment that immediately demonstrated how music could be controlled hands-free using voice. Showing users what they could do at the moment they completed setup reinforced the value of the feature and encouraged immediate adoption.