Choosing a Default AI Persona: How a Scheduling Startup Balanced Trust, Tone, and Conversion

AI · 6 min read

Schedly built an AI assistant to manage meeting scheduling and initially shipped with a neutral, efficiency‑focused persona. Early feedback showed the tone felt robotic to enterprise users and too casual to SMB clients. The team designed an experiment to test three personas — Formal, Friendly, and Expert — across cohorts and measured open rates, reply tone, and meeting acceptances.

The results were surprising: while open rates were similar across personas, the Expert persona drove a 14% higher meeting acceptance rate and fewer reschedules, suggesting users trusted the assistant’s recommendations more when it sounded authoritative. However, enterprise legal teams pushed back on certain phrasings in the Expert set, so the product team adopted a hybrid approach: Expert by default with configurable tone toggles for organizations.

Beyond conversion, the experiment surfaced important UX decisions about defaults and customization. Schedly prioritized a safe default that maximized acceptance while introducing admin controls and a per‑user personalization slider in the settings. The lesson for startup designers: test persona as a functional product lever, not just as copywriting flavor.