The outcome will reflect communication patterns!

In 1967, Melvin Conway observed that organizations design systems that mimic their communication structure.  The social interactions between programmers and groups of programmers cause this outcome. There have been studies at Harvard, MIT and elsewhere that support Conway’s Law (know in academic circles as the mirroring hypothesis). Conway’s Law is important because it links organizational structure to outcomes. Conway’s Law effects not only the organization of work but what work a team does. Two competing organization structures illustrate the impact of Conway’s Law. The first occurs where the teams organized by functional area and the second is the organization of people by value stream.

Tiny Case Study Context:  Tom Cagley & Associates (TCA) is an agile consultancy. The most basic functions of performed at TCA include: work acquisition (sales and contracts), work delivery, billing, and collections. TCA has a set of processes supported by SAAS and package solutions to address each of these functions.

Organizing the systems personnel (that sounds grander than reality) by function would have teams focused on work acquisition, work delivery systems to the collection system. In the most extreme case, solutions delivered using the functional model are an assembly of individual functional changes cobbled together. Each change reflects the same handoffs observable in the org chart. Every large organization I have dealt with leverages this model. Even when organizations organize to support products they often subdivide to address specialties for many reasons. In TCA’s example, most of the packages address a specific function rather than the entirety of the product stream. We could take the example to absurd level (for TCA) and premise that the work delivery team at TCA might subdivide between consulting systems and presentation systems (PowerPoint and Prezi support). In this model, as work moves from acquisition to delivery there a large number of handoffs across team boundaries. The systems and processes, built or acquired, reflect the organization structure or worse, dictate the organizational structure. The theory for this approach is that the gains from specialization overwhelm the inefficiency of handoffs. Specialization works best where there are economies of scale, in software development and maintenance, this is rarely true but rarely measured.

Organizing people to support the value stream, moving from acquisition of work to final disposition, crosscuts the functional view. In this case, the work follows the steps required to deliver value without crossing boundaries. This approach to solution delivery takes a systems view.  The focus is on the flow and transformations required to deliver value which minimizes handoffs between functional areas. This approach is relatively rare in large organizations. The approach is more common in small organizations and start-ups due to constraints in the number of people, the lack of development of specialized systems, and little historical baggage generated by role specialties. This approach has fewer boundaries and thus less overhead but requires building a full-stack team which more planning to recruit and construct. Adherents of this approach believe that the lack of handoffs reduces overhead and cycle time which outweighs recruiting costs and lower utilization rates for some specialties (which is why you want T shaped people).

Conway’s Law is generally expressed from the point of view building systems. In the classic case, the patterns of communication between departments and teams define how the system will be constructed.  As importantly, in today’s build, buy and borrow environments, the systems impact organization structures. Conway’s Law for process and system changes can be stated as changes to the systems including business processes require changing communication patterns.

Next:  Reflections On Conway’s Law and Agile