From KISS to Feature Creep: How a Mini-Program Made Me the PM I Dislike

After years of coding, the author increasingly embraces the KISS principle and dislikes complexity in projects. While code complexity can be refactored, product complexity often has no way back: every added feature incurs maintenance, cognitive, and user-understanding costs. The author argues a product should ideally do one thing, or at most serve as a unified portal; tightly related features belong together, not a stream of ad-hoc ideas justified by "I think customers want this."

The author's own mini-program "Mood Diary Calendar" began with a clear, restrained scope: daily text and emoji mood entries, plus monthly and yearly statistics. This initial version solved a specific personal need — tracking monthly mood distribution to better control emotions.

Feature creep started during travel. To avoid monotony, the author added image upload (referencing an earlier article on acme.sh + DNSPod for Let's Encrypt certificate automation). Then came geolocation (referencing an article on fuzzy positioning, reverse geocoding, and manual point selection). The result resembled a stripped-down WeChat Moments — lacking social interaction and with lower implementation quality.

Next, weather data was added using the date and location (referencing an article on QWeather integration and optimization). At this point the author felt discouraged, considering deletion, but discovered a small stable user base despite zero promotion. The low hosting and certification costs justified keeping the service running.

Further travel revealed a need to record multiple locations per day ("special forces tourism"), so a travel-log feature was built: each day could contain multiple spots with photos and coordinates. The author envisioned later adding AI to polish travel notes into PDFs or images for sharing on Xiaohongshu or Moments, because mini-programs cannot include social features and the author lacks operational capacity.

An anniversary reminder was then added — the author often forgot dates, and the feature would also boost daily active usage. This became another entry point justified by personal convenience.

In retrospect, the product evolved around the author's immediate needs with no long-term roadmap. It remained organized around the date dimension, and AI assisted with UI layout improvements. The author apologizes to early users for the initially poor visual design.

The experience underscores that product management is genuinely difficult: maintaining restraint while responding to real usage is a constant tension.