Radiology for managers: network analysis

Network analysis has moved from anthropology and sociology into management science. In a case study of the engineering firm MWH, Professor Rob Cross shows how organizational network analysis can help managers to improve connectivity and collaboration.

Take a large global engineering services firm, MWH, which decided to reorganize its IT department along functional lines while trying to improve its collaboration across continents. Vic Gulas, the manager in charge of the reorganization, realized that information sharing within and across functions and continents was going to be key to the success of the project. Rob Cross, professor of Management at the McIntire School of Commerce (University of Virginia) has written a case (see reference below) on Gulas’s work at MWH: “The work of Vic Gulas at MWH shows us how the technique of organizational network analysis can help visualize the strengths and weaknesses in an organization and how it can help managers focus on specific individuals or biases in the network and thereby improve performance”
 
MWH
Colorado-based MWH is a large engineering services firm ($975 million in 2003 revenues), considered to be a leader in water, hydropower and environmental remediation. A certain amount of fragmentation goes with its turf. Each year the firm manages hundreds of different projects for different types of institutions ranging from municipalities to multinational companies. The company is spread over the world, from the Americas (e.g. the 5 de Noviembre Hydroelectric project in El Salvador) to New Zealand (regional IT office) via Iran (Karaji Dam).

To add a further degree of fragmentation, the company has grown through mergers, meaning that different types of habits and practices have accumulated over the years.
When looking at his company, Gulas saw it as more tribal than global, with business units and regions operating in relative isolation.

The IT department was no exception in this regard. There were 185 employees spread over four offices in the US, two in the UK, and two more recently opened ones in New Zealand and India. When MWH had been smaller, such a regional organization had not been much of a handicap. But as the organization grew, best practices were rarely adopted in a systematic way. Offices tended to function independently; for example, the team in Pasadena, California responsible for upgrading the ERP system in the Americas worked in a virtual silo, in part because it reported to the finance department.

Gulas’s mission was to take this IT department, and create a new organization baptized iNet. iNet was to set an example in global connectivity and collaboration for the rest of the firm. As Gulas saw it, achieving this goal entailed implementing the right structure (see box on functional organization) but perhaps more importantly creating the right relationships among organization members.

The right relationships
Serendipitously, around the time of taking on the iNet challenge, Gulas attended a session on analyzing networks in organizations at a conference held by IBM’s Institute for Knowledge-Based Organizations. While network mapping was relatively common in sociology and anthropology, its application to business organizations was relatively new. Gulas felt that this form of analysis, known as organizational network analysis (ONA) would be very helpful in rendering explicit the ties that did and did not exist within his group.

The first step in such an analysis is the design of the survey. Gulas prepared a web-based questionnaire which contained questions about which people the respondents turned for important information in their jobs, about their level and tenure at MWH. 90% of the 185 employees responded.

After processing of the survey results, Gulas had at his disposal a series of network maps and data tables. The maps use line and nodes to represent connections between members of the network – they provide a visual representation of the strengths and weaknesses in the network (see below for the MWH map with the different node colors corresponding to the different offices). The data tables, of a more customary format, provide quantitative summaries of network connectivity. Professor Cross insists on the visualization the maps and tables allow: “Think of network analysis as providing an x-ray of the organization. What was invisible before can be visualized. The manager, like a doctor, can see where the strengths and weaknesses are in the network and now knows on what she should focus.”

The maps and tables allow managers to address a number of issues. For example, how connected are the different sites? Given that the IT group had been organized along geographical lines, Gulas expected a fragmented network and the network map showed him just that. The data tables confirmed this (see table below). In particular they showed that the two newer sites, Christchurch and Pune which MWH was hoping to gear up had very little connectivity with the other geographical locations. Low connectivity between offices could explain duplication of tasks. Connectivity between Denver and Pasadena was low  (6%), providing an explanation for the fact that when HR requested a program to process employee address changes, the Denver office created an application not realizing that Pasadena’s ERP application already provided a solution. Gulas was going to need to work on the connections between different sites.

Another important question concerns connectivity within a site. Within the IT organization, some sites showed seemingly low internal connectivity. For example in the biggest US office, Denver, only 20% of possible connections existed. Obviously the bigger the office the lower the connectivity you can accept. While 80 % connectivity in a 100 person office would be a drag on the organization, 100% connectivity in a ten person office is a much more manageable goal. But Gulas had evidence for inquiring into the Denver situation.

The Walnut Creek figures pointed out another potential problem. While intra-office connectivity was high (48%), connections to the other locations were low, indicating the possibility of a go-it-alone, stick-to-ourselves attitude. High intra-office connectivity is fine as long as it goes hand-in-hand with high inter-office connectivity.

Network analysis also allows one to examine connections among hierarchy levels. Gulas’s survey results showed that there was relatively high level of collaboration amongst the most senior employees but less connectivity among lower level employees. This low connectivity was troublesome in two regards: the front line employees were not sharing expertise to solve client problems; such isolation from their colleagues could create unhappiness and ultimately result in higher turnover.

A further benefit of network analysis lies in its ability to identify workers who are either underutilized or overburdened. Gulas needed to pay particular attention to two types of individuals, those with potentially insufficient network sizes and those with potentially excessive network sizes. A network size of only one or two connections points to potential isolation. Such peripheral people are often less satisfied in their work because of their lack of connections and to the extent that they have skills worth keeping within the organization, need to be tied more strongly into the network.

Second, Gulas noticed individuals with more than 25 people coming to them for information on a frequent basis. These individuals were potentially overloaded; they could be creating a bottleneck and what would happen if they left? Spreading information delivery over the network would improve the situation. Bringing the peripherals in and decentralizing somewhat the centrals are tasks that the maps and tables allow managers to initiate.

Network analysis can look at channels of connection through which people connect to each other. Does the information come though face-to-face meetings or through e-mail or over the phone or through some other means? In MWH’s case, the analysis revealed that 37% got information from unplanned meetings versus 5% from planned meetings, thereby raising the question for Gulas of how iNet could improve information delivery in its formal meetings, which were going to be more necessary as the company undertook more complex global projects.

Professor Cross teaches this and other cases related to network analysis and summarizes his students’ positive reaction: “The students are amazed by the way the maps and tables allow them to visualize the organization. It’s like having a brain scan of the organization to work from. Organizational change ceases to be opaque and becomes tractable. You’ve got images and data to work from.”

Reference:
UVA-S-0112
“Building a Networked Organization: Restructuring the IT Department at MWH”
University of Virginia – Darden School. McIntire School of Commerce, Batten Institute
Rob Cross, Vic Gulas, Amy Halliday, Andrew Parker, Teri Vigneux

Published September 2009