Product architectures should reflect the organizational structures that design them (the “mirroring hypothesis”). If they do not, modularity literature has documented a number of potential negative performance effects. We study how and why product and organizational architectures do not mirror analyzing the microdynamics and determinants of product cross-module dependencies for which there is no corresponding organizational tie. Building on complexity theory applied to technological and social systems, we first hypothesize that “non-mirroring” tends to increase over time as a function of the nature of the product development tasks, the proximity of developers, and time pressure. Then, we test these hypotheses on a unique micro-level dataset, documenting the development of an enterprise software over a course of 60 months and along 13 versions. Our estimates, robust to alternative model specifications and identification strategies, suggest that “non-mirroring” between product and organizational architectures increases when development tasks are routine, engineers are not co-located, and as the deadline of releasing new software version approaches. Moving from a subjective to an objective measure of non-mirroring (with software modules derived using the Louvain algorithm instead of relying on the engineers’ clustering of design elements), we find that the misalignment between product and organizational architectures increases when engineers perform innovative development tasks, as their imperfect perception of modules make them focus on the wrong technical dependencies. Our findings document how micro- organizational choices contribute to the misalignment between product and organizational architectures and offers engineers insights into how to prevent the negative performance outcomes that stem from non-mirroring.
- mirroring hypothesis