Rapid advancements in consumer technologies enable peo-ple to collect a wide range of personal data. With a proper means for people to ask questions and explore their data, longitudinal data feeds from multiple self-tracking tools pose great opportunities to foster deep self-reflection. How-ever, most self-tracking tools lack support for self-reflection beyond providing simple feedback. Our overarching goal is to support self-trackers in reflecting on their data and gain-ing rich insights through visual data exploration. As a first step toward the goal, we built a web-based application called Visualized Self, and conducted an in-lab think-aloud study (N >= 11) to examine how people reflect on their personal data and what types of insights they gain throughout the reflection. We discuss lessons learned from studying with Visualized Self, and suggest directions for designing visual data exploration tools for fostering self-reflection.