T-Life app · Home Internet — Self-serve Performance Diagnostic
Showcase

Sr. Product Designer · 2024–2025
Consumer · Mobile 
I designed the Performance Diagnostic experience for T-Mobile's T-Life app — a self-serve system that helps Home Internet customers diagnose and resolve performance issues on their own.
CONTEXT
Performance issues were the leading driver of churn and support calls for T-Mobile Home Internet customers, but users had no meaningful way to diagnose problems themselves. The system was complex; the customer experience was opaque. Performance Diagnostic translates network telemetry into clear, guided resolution — directly inside the app users already had.
6 remediation flows  ·  40+ screens  ·  4 months  ·  Sole designer
Customers didn't need more data. They needed clarity and guidance.
​​​​​​​
The system tests five layers of connectivity and walks users through resolution:
— Cellular connectivity, gateway speed, Wi-Fi health, mesh placement, broadband address 
— Technical signals translated into plain-language states 
— Step-by-step remediation, never just diagnostics
SOLUTION
Decision 1 — Action over information.
A diagnostic that only reports problems makes users more frustrated, not less. Every detected issue routes into a guided remediation flow — Gateway Placement Assistant for cellular, in-app address update for broadband mismatches, interactive step-by-step guidance for Mesh relocation. The system never leaves users at a dead end.

Decision 2 — Plain-language categories, not technical tests.
The user-facing diagnostic categories don't map 1:1 to the backend tests. Multiple network checks collapse into single human-readable states ("Cellular connection," "Wi-Fi health") so customers see what they need to act on — not the underlying telemetry.
OUTCOME
Shipped at global scale 
Mini Profile launched on News Tab, a surface available to Facebook's billion-plus users.

Scope expanded mid-project 
The work was elevated to a universal bottom sheet for use across other Meta surfaces — and the team shipped a parallel solution for the original use case.

You may also like

Back to Top