Purpose: This study develops a web-based matic motorcycle damage prediction system using the K-Nearest Neighbors (KNN) algorithm at Bengkel Sahabat Motor to support early damage detection, preventive maintenance, and cost reduction.

Methodology: A quantitative approach with waterfall System Development Life Cycle (SDLC) was used. Data were collected through observation, interviews, and workshop records. The system was built using Personal Home Page (PHP), html, Cascading Style Sheets (CSS), JavaScript, and MySQL. KNN with Euclidean distance and K=3 was applied, using a three-level symptom scale. System design used Unified Modeling Language (UML) and validation was conducted through black box testing.

Results: The system accurately classifies motorcycle damage, with test outputs correctly identifying "Engine Overheating" based on nearest neighbor distances. Black box testing achieved 100% acceptance across 143 test items, categorized as “Very Good.” Diagnosis time decreased from 30 to 10 minutes per case.

Conclusions: The KNN-based system effectively automates motorcycle damage classification and improves diagnostic efficiency.

Limitations: The study is limited to a single workshop, small dataset, no IoT integration, and lacks formal accuracy metrics.

Contributions: This study provides a practical machine learningbased predictive maintenance system for motorcycle workshops, offering a replicable framework for digital diagnostics in the automotive service sector.