Innovation Streams and Ambidextrous Organizational Designs: On Building Dynamic Capabilities

Abstract

This paper explores the roots of dynamic organizational capabilities. We suggest that streams of innovation are central to a firm’s dynamic capabilities. We argue that organizational design is an important determinant of innovation streams and that these streams can be effectively executed within organizations through ambidextrous organizational designs. Based on in-depth data on 15 business units and 36 innovation episodes we explore the consequences of organization design choices on innovation outcomes. We explore the phenomena of ambidextrous organizational designs what they are, how they operate, and their association with innovation outcomes. We also explore their boundary conditions as well as how these complex organizational forms evolve. Functional designs, cross-functional teams, and unsupported teams are significantly less effective in hosting innovation streams than ambidextrous designs. Ambidextrous organizational designs are effective for all types of non-incremental innovations except for product substitution events. Further, the use of ambidextrous designs was associated with the sustained performance of existing products. While not all ambidextrous designs were successful, firms that did not employ ambidextrous designs were unable to implement streams of innovations. Two business units initiated spinouts but only after the non-incremental innovations were initiated within an ambidextrous design. These exploratory results suggest that organizational design and senior leadership behaviors are powerful levers in driving streams of innovation and that ambidextrous organizational designs permit a business unit to simultaneously explore and exploit. We discuss the implications of these results for the ongoing debate on the nature of dynamic capabilities and organizational evolution.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 39 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.