From networks to narratives: data-driven business model ideation

Purpose This study proposes a data-driven approach to business-model ideation using heterogeneous link prediction. It addresses limitations of traditional expert-driven ideation approaches, which may be constrained by cognitive bias and limited search breadth. Design/methodology/approach Business model data from 248 real-world companies were collected from businessmodelideas.com. Keywords representing three core components – value proposition, revenue stream and technology – were extracted and used to construct a heterogeneous network. Link prediction identified 725 retained connections, which were organized into six business-model ideation themes. Findings The method generated six categories: (a) Synergistic value proposition, (b) Revenue source diversification, (c) Technology complementarity, (d) Technology-enabled value creation, (e) Technology-powered new revenue source and (f) Revenue expansion through value aggregation. The generated themes were supported by illustrative cases and exploratory expert assessment, providing initial evidence of plausibility across both ex-post and novel opportunity cases. Practical implications The proposed approach offers a structured decision-support mechanism for translating predicted linkages into candidate business-model ideas. Managers can review robust high-probability links, summarize each as an opportunity concept with key assumptions and value-creation logic, and conduct low-cost exploratory tests for further evaluation. This process supports more systematic business-model experimentation while preserving managerial judgment. Originality/value This study contributes a scalable method for business model ideation using heterogeneous link prediction, moving beyond conventional manual ideation. It systematically discovers meaningful business model combinations from structured company-level data.