Chris shares his experience improving his calorie tracking app Amy’s low week one user retention by fixing bugs, enhancing AI accuracy, and adding major features like barcode scanning and water tracking with innovative solutions for international users. Despite initial setbacks, these improvements aim to boost retention and user trust, contributing to steady revenue growth and ongoing development.
Five months after launching his calorie tracking app Amy, Chris shares his journey tackling the challenge of low week one retention, which measures how many users continue using the app one week after signing up. Initially, he believed retention had jumped from 12% to 20%, but later discovered a miscalculation and that retention had actually dropped back to 13%. Determined to improve this key metric, Chris focused on reducing friction in the app by fixing bugs, enhancing AI accuracy, and shipping two major new features: barcode scanning and water tracking.
Barcode scanning was a complex feature to implement because it involved two parts: scanning the barcode itself and then identifying the product from the barcode. Chris used AI-powered research tools to find suitable APIs and ended up combining two databases—Open Food Facts, which had better international coverage but inconsistent serving sizes, and FatSecret, which was more accurate for US products but limited internationally. He also built clever fallback solutions for edge cases, such as scanning nutrition labels when barcodes were missing or damaged, and handling multi-photo stitching for curved labels, innovations he had not seen in other calorie tracking apps.
After the barcode scanning feature, Chris developed a water tracking feature inspired by his own need to drink more water while traveling. He designed a simple yet engaging user interface that allowed users to log water intake by typing and included visual indicators like a water line and a floating duck animation. To handle international users, he implemented a language-agnostic detection system using lightweight AI models to distinguish water entries from food entries, even when both were logged simultaneously. The feature was made optional to keep the app simple and avoid overwhelming users.
Beyond these headline features, Chris emphasized the importance of improving accuracy and fixing small UX bugs to build user trust and reduce churn. He highlighted how even minor issues, like scrolling desynchronization between text and calorie columns, could frustrate users and cause them to abandon the app. These behind-the-scenes improvements, combined with the new input methods, are aimed at boosting week one retention, though Chris notes it will take a few weeks to see the impact.
Overall, Chris’s transparent update reveals the challenges of building and scaling a productivity app like Amy. Despite setbacks, he remains optimistic about growth, with monthly recurring revenue climbing to $2,800 and a strong profit margin. He invites viewers to follow his journey on social media and promises to share future insights, especially around improving AI accuracy, as he continues refining Amy to better serve its users.