Abstract

Agile management has emerged as a critical paradigm for sustaining innovation and adaptability in dynamic entrepreneurial environments. This study employs an unsupervised machine learning framework to identify data-driven archetypes of agile management among 5,000 successful startups, utilizing the publicly available Startup Failure Prediction Dataset. Unlike traditional predictive models that classify startups as successful or failed, this research focuses exclusively on successful ventures to explore internal variations in agility and operational strategy. The methodological pipeline comprised four key stages: data preprocessing using Min–Max scaling, dimensionality reduction through Principal Component Analysis (PCA), clustering via the K-Means algorithm, and post-hoc interpretation of cluster profiles. Optimal clustering was achieved with k = 4, validated using the Silhouette Coefficient. The resulting clusters—Innovative-Stable Performers, Lean-Traditional Operators, Agile-High-Innovation Leaders, and Resource-Intensive Scalers—exhibited distinct financial, structural, and innovation-related characteristics. Cluster 2 (Agile-High-Innovation Leaders) demonstrated the strongest innovation capability and customer retention, while Cluster 3 (Resource-Intensive Scalers) emphasized growth through capital intensity and larger teams. Findings reveal that agile management practices among startups are not homogeneous but multidimensional, reflecting diverse strategic equilibria between innovation, resource efficiency, and experiential learning. The study provides empirical evidence that agility manifests through multiple viable pathways—ranging from lean innovation to scale-oriented execution—each corresponding to different configurations of financial and human capital. The research contributes to both management theory and data science by demonstrating how unsupervised learning can empirically derive typologies of agile behavior, thereby bridging the gap between computational modeling and organizational studies.

Keywords: Agile Management, Startup Archetypes, K-Means Clustering, Unsupervised Learning, Innovation Analytics

How to Cite:

Subiyakto, A., Huda, M. Q. & Hakiem, N., (2026) “Identifying Data-Driven Archetypes of Agile Management in Successful Startups Using Unsupervised Machine Learning”, Agile Management 1(3), 155-168. doi: https://doi.org/10.63913/am.v1i3.133