To appreciate customers and partners who have journeyed with Coolmate, since 2023, every December the brand launches its annual “Lookback” experience — a personalized recap of the year spent with Coolmate.
I have been involved in the Lookback project since its early versions. However, in the 2025–2026 edition, I participated more deeply in shaping the product strategy and experience direction, beyond just UI execution.
This year, we expanded the experience with a stronger AI-driven approach. Users could chat with CoolGPT, generate personalized chibi avatars, join lucky draw activities, and explore customized titles generated based on their real shopping behavior.
Working closely with the Marketing and Development teams, we simplified visual and textual elements, focusing instead on training AI models to analyze behavioral data and deliver more meaningful, personalized outputs.
Lookback 2025 evolved from a gratitude microsite into an interactive, data-powered brand experience — where AI, personalization, and emotional engagement come together.
A 27.6% image generation rate is strong for a newly introduced AI feature, indicating high curiosity and engagement.
Approximately 65% of users who generated images set them as avatars (553/840), showing strong personalization appeal.
Share button clicks reached ~94%, suggesting effective CTA placement and visibility.
AI integration did not significantly impact overall funnel retention, maintaining strong Lucky Draw participation.
The actual share rate (13.9%) is significantly lower than share button clicks, indicating potential friction in the final sharing step.
Enhancing “achievement showcase” elements in avatars may encourage more organic sharing behavior.
With further Vietnamese language training, the AI could generate copy that is more concise, culturally nuanced, and trend-aware while maintaining brand tone consistency. Improving linguistic quality may enhance emotional resonance and increase organic engagement and sharing behavior.
The working file might look a little messy at first glance 😅
But behind it was a tightly coordinated effort between Marketing, Development, and Product.
To bring AI into the Lookback experience, we trained models, prepared APIs, refined personalization logic, designed interfaces, optimized flows, and built sharing features — all within a tight timeline.
.svg)
