Posted by Ray Poynter, 10 June 2019
Recently, my son, William, introduced me to Conway’s Law, and it immediately struck me that this ‘Law’ is an important consideration for market research and insights.
Conway’s Law is: “Any organization that designs a system will inevitably produce a design whose structure is a copy of the organization’s communication structure.” For example, if we divide a problem so that part of it is answered by a Qualitative team and part by a Quantitative team, we will tend to produce an output that comprises a qualitative and quantitative part.
Conway’s Law is often quoted in the world of software development. For example, in “Exploring the Duality between Product and Organizational Architectures: A Test of the Mirroring Hypothesis” a report published by MacCormack, Rusnak and Baldwin and the Harvard Business School in 2008 looked at how Conway’s Law impacted software products. The team looked at a range of software projects in fields as diverse as word processing and financial management and compared the outcomes from teams that worked in silos and from loosely-coupled, open teams. The study found that the silos tended to produce monolithic systems whereas the open teams tended to produce modular, inter-operable elements.
Conway’s Law and the World of Customer Insights
Turning this thinking to the world of insights, we can see a similar consequence of different communication structures – and one of the reasons that there has been such interest amongst clients to break down silos. Silos being the consequence of creating pockets of restricted communication. The consequences of Conway’s Law are visible at the research agency level, the end-client level, and between the two.
At the Research Agency Level
If an agency breaks a project into sales, field, analysis and presenting, it runs the risk of producing an output that details: 1) Why we won the project, 2) Who we interviewed, 3) How we analysed the data, 4) Our conclusions and recommendations. Exactly the sort of reporting that clients have been criticising. For an alternative output see the recent NewMR presentation by Mike Sherman and Neil Gains.
Similarly, if a research project is broken into country teams, or into qual and quant, or into teams looking at survey data and big data, we will tend to produce outputs that reflect those structures. The alternative is to focus on a project team approach, perhaps ensuring that people whose job is at the front-end of the process stay engaged throughout, and that people whose tasks are nearer the end of the project are present at the design and kick-off stages.
The implications of Conway’s Law are even more profound for the end-users of insights, i.e. clients. If an organisation divides it sources of insight into, say, an insights team, an ops team, and a data analytics team it will tend to get three strands that it will then have to try to marry together after they have been created. If a client has research being developed by several suppliers, and the suppliers are each working in a silo, talking only to the client, the product will be insight that is riven into components, that somebody will have to try to integrate. In the recent client interviews with people like Unilever’s BV Pradeep, Danone’s Elaine Rodrigo, and Nissan’s Mr Takahashi we can see that they are all seeking to tackle this issue, moving away from being order takers (which automatically silos them) and towards integrated solutions.
To some extent, thinking about Conway’s Law simply provides additional backing for the widespread concern about silos and the move to break out of the silos. However, as my son William pointed out to me, you can’t break Conway’s Law, but you can leverage it. If you think about the sort of outputs you want, you can then design your teams and communication networks in ways that are likely to produce it. For example, if you are an end client, you should be connecting your suppliers in ways that create integrated insights. If you are an agency, think about integrated project teams, and seek opportunities to communicate with other suppliers and end users.