Table A1.

Empirical data: path origin of detailed learnings in relation to their impact on innovation paths

	Impact on Path 1—FCs	Impact on Path 2—SWTs	Impact on Path 3—FS	Impact on Path 4—WHR
Detailed learnings 1–14 from Path 1 (FCs)	T1: On‐path learning 1. TechLtd learns about FC technology → L1 2. Size of controller needs to be small to fit vehicle applications → L1 3. Strict compliance with many safety standards is needed in the car industry → L1 4. Learns about feed‐in technology (how electricity generated by FCs needs to be transformed to be grid compatible) → L1 5. Discovers characteristics of green markets (numerous small niche markets with different regulations, some are at very early stage or still in formations and thus difficult to predict) → L1 6. Discovers automotive supplier markets → L1 7. Meet actors knowledgeable about feed‐in technology and green markets → L1 8. Norms in car industry (product quality and compliance, need for dedicated department) → L1 9. Automotive markets have high entry barriers due to long supplier selection processed. High risk of important upfront investment → L1 10. High‐volume automotive markets exceed TechLtd's current manufacturing capacity → L1	T2: Path‐initiation learning 4 → L2 5 → L2 7 → L2 11. Mature technologies offer better commercial prospects, unlike those still in emerging as in path P1 → L3 12. Large well‐established markets difficult to penetrate due to important, formalized entry barriers (such as strict norms and supplier selection standards) → L3 13. Market size should not exceed TechLtd's production capacity of <1000 pieces/year → L3   T3: Cross‐path learning 14. R&D collaboration is very helpful for new product development (to share costs and risks) → L5	T2: Path‐initiation learning 2 → L2 3 → L2 4 → L2 5 → L2 6 → L2 7 → L2 8 → L2 11 → L3 13 → L3   T3: Cross‐path learning 14 → L5	T2: Path‐initiation learning 4 → L2 5 → L2 7 → L2 11 → L3 12 → L3 13 → L3   T3: Cross‐path learning 14 → L5
Detailed learnings 15–31 from P2 (SWTs)	n/a	T1: On‐path learning 15. Learns about SWTs and associated technologies → L1 16. Wind turbine management technology (manage highly fluctuating turbine velocity) → L1 17. High efficiency inverter and grid feed‐in technology to minimize conversion losses → L1 18. Meet many SWT market actors → L1 19. Discover global SWT niche market dynamics, and the difference among countries → L1 20. SET markets often come with high product development costs due to national regulation; high costs for market monitoring → L1 21. SET markets are highly dependent on government regulation of renewable energies; require intensive market exploration and adapted exploration methods → L1 22. Early‐stage markets typically involve → L1 – bankruptcy risks – long and difficult trust‐building – lack of professionalism – lack of industry networks – lack of industry lobbies 23. In the SWT market, many poorly functioning turbines create legitimacy problems (that can even impact reputation for component suppliers) → L1	T2: Path‐initiation learning 17 → L2 24. Most inverter and feed‐in components developed for SWT can be reused for FS and WHR → L2 25. Do not go into B2C markets as TechLtd lacks competence to handle large consumer sales and after sales requirements → L3   T3: Cross‐path learning 20 → L4 21 → L4, L5 22 → L4 26. Industry workshops are good to rapidly meet many market actors and gain market information → L5 27. In SET markets, upfront research is needed about the numerous regulatory frameworks (for instance related with the different feed‐in tariffs in Europe) → L5 28. Approach industry leaders (to understand their business model) allows to rapid have an idea of the market dynamic at play → L5 29. Involve end users in product development (to better understand their needs) → L4, L6 30. Assess viability of the business model of potential clients. Sometimes it even allows to estimate if the overall market is viable; the business model of a waste truck manufacturer showed that flywheels are not ready for automotive applications → L5, L6 31. Rapid prototyping instead of (upfront) full product development → L5, L6	T2: Path‐initiation learning 17 → L2 24 → L2 25 → L3   T3: Cross‐path learning 20 → L4 21 → L4 26 → L5 27 → L5 29 → L4, L6
Detailed learnings 32–39 from P3 (FS)	n/a	n/a	T1: On‐path learning 32. Learns about FS technology → L1 33. Two main application areas: vehicles and electricity grid → L1 34. Learns to developed and emergency shutdown module in case of electricity cut → L1 35. High complexity as FS can be used for multiple applications in several still emerging markets → L1 36. Large FS fit electricity‐grid, micro FS rather automotive applications → L1 37. FS diffusion dependent on government feed‐in legislation → L1 38. Automotive market: strong dependence on very few market gatekeepers (e.g., automakers, train manufacturers) → L1	T2: Path‐initiation learning n/a   T3: Cross‐path learning 39. A good method is to adopt a “wait‐and‐see” approach in promising markets with high uncertainty about future development → L5, L6
Detailed learnings 40–43from P4 (WHR)	n/a	n/a	n/a	T1: On‐path learning 40. Learn about WHR technology → L1 41. Heat recovery is still a niche market with many different customer applications → L1 42. High search costs for other/new customers in markets with very diverse customer applications → L1 ○– case by case approach (no standard solutions) ○– high‐speed solutions only needed in some applications ○– need to demonstrate advantage of high‐speed solutions to customers 43. WHR market highly dependent on government regulations → L1

Note: FC: fuel cell; FS: flywheel storage; R&D: research and development; SWT: small wind turbine; WHR: waste heat recovery. The detailed Learnings 1–43 are listed and explained in relationship between the Paths P1–P4 they originated on and the Paths P1–P4 they had an impact on. Some detailed learnings have an impact on more than one path. Each detailed learning is related with an “→” to one or more of the six learning outcomes (L1–L6). Some detailed learnings related with two (or more) different learning outcomes. In that case, the table only features their number when they appear the second time (e.g., 2 → L2). Single‐loop learning outcomes that were directly useful for the same path (on‐path) are described under T1. Single and double‐loop learning outcomes useful to initiate new paths (path‐initiation learning) are described under T2, and double‐loop learning outcomes that allowed an improvement of exploration (across path) under T3.

Coding scheme: first order concepts: detailed learnings (1–43); second order themes: learning outcomes (L1–L6); aggregated dimensions learning types (T1–T3).