Critical Connectors

Networks have one additional quality that makes them especially interesting to OD practitioners: critical connectors. Stephenson (2005) patented specific mathematical algorithms that identify three types of critical connectors. According to Stephenson, these critical connectors comprise 5 to 10 percent of all nodes in a network and are disproportionately influential over the whole. By tapping into critical connectors and using them as conduits for change, OD practitioners can more effectively and rapidly generate change. The three roles are hubs, gatekeepers, and pulsetakers.

Hubs

In Figure 29.2, Karen is a hub because she connects with more people than anyone else. Stephenson describes hubs as having a hearty appetite for information, which can be gossip, technical expertise, product knowledge, or any other kind of information (2005). They tend to be multi-taskers who enjoy juggling activities and relationships. Hubs have great value in the network since they are compulsive sharers who spread information continually. OD practitioners can leverage hubs during change by sharing important messages with them. Because of their deep connectedness, hubs have the power to disseminate that information quickly throughout the network.

The hub also has what is called high degree centrality. Degree centrality measures nodes’ number of connections. With more contacts than anyone else, Karen has the highest degree centrality: 72 percent. In comparison, Aisha has only one connection. Her degree centrality is 9 percent. High degree centrality is not always positive. According to Krebs (2000), “What really matters is where those connections lead to—and how they connect the otherwise unconnected.” Karen has connections within her immediate cluster: people who are already connected. She might have too many redundant links and spend her time with people who cannot add information to her network, only duplicate it.

Organization Change

How many times do companies undertake a costly reorganization only to find that it fails to achieve the desired results? In these cases, the challenge often is that the informal system—the networks—has not changed along with the organization structure. People continue to draw on the connections they have in order to get work done. In essence, they draw on a shadow organization that maintains the status quo and undermines the desired relationships dictated by the new structure.

OD practitioners can use network analysis to help rewire the organization to support restructuring. By understanding internal networks, OD practitioners first deepen their understanding of how people operate in the organization and identify, for example, the constellations of connections that should not be interrupted by the reorganization since they work so well. Then OD practitioners have options: they can design the new structure to capitalize on the strongest and most productive network patterns. They can encourage new network patterns by shifting influential players in the networks to different positions. They can interrupt dysfunctional patterns by engaging critical connectors in new activities. These approaches can help minimize the time needed to realize the expected benefits of change by building alignment between the formal and informal organizations.

Network knowledge can also improve speed of integration after a merger or acquisition. By engaging critical connectors from both entities in joint activities, OD practitioners can accelerate integration. The theory proposes—and research supports—the idea that peers intuitively follow critical connectors, who are both highly trusted and adept at sharing information (Stephenson, 2005). As a result, rather than having to engage an entire workforce deeply, OD practitioners can focus attention on the critical connectors, who then naturally ease the communication and integration process due to their influence in the network.

Diversity and networks is explored in Exhibit 29.1.

Network analysis also provides insights into the nuances of successful team composition. For example, successful R&D teams share specific network qualities that, when planned into their structure, can improve the likelihood of success. Teams use technical-advice networks to access expert information on technical issues and organizational concerns, such as project scheduling and coordination (Rizova, 2006). They use organizational-advice networks to support them when they need “organizational competence and skills” (p. 50). By setting up teams with access to technical-advice and organizational-advice networks, members have access to both types of expertise needed to be successful.

Another important cause for success is a team’s ability to draw on their personal networks to support the team’s work. These teams take advantage of contacts from the members’ personal networks and integrate their own knowledge with capabilities from their networks (Cummings & Pletcher, 2011). A study performed by Cummings and Pletcher of 177 project teams involved in an organization team recognition program showed that finalists—teams that achieved product innovations, demonstrated superior customer service, and/or improved operational effectiveness—had stronger project networks than did non-finalists. A product innovation finalist, in comparison to a non-finalist, was able to draw on significant technical and marketing experts and engage more suppliers and customers. Their broader and deeper knowledge allowed them to create innovations that eluded other teams.

Networks and Leadership

Today’s leaders live in a different world than in the past. Rather than simply managing a vertical domain or a functional area, leaders are embedded in complex ecosystems: a type of macro network formed by the dynamic interplay between interconnected organizations that depend on each other for mutual survival (Moore, 1997). In these business ecosystems, building connections and managing relationships are critically important.

Effective leaders mobilize networks in specific ways (Cross, 2010):

• They manage the center: Leaders minimize bottlenecks, develop high performers, and acknowledge those who, while not individual stars, help their colleagues succeed.
• They leverage the periphery: Leaders bring in newcomers to the group and reengage high performers who have neglected or lost connections.
• They bridge silos: Leaders weave connections in order to improve collaboration at key intersections between groups or geographies.
• They encourage awareness of expertise: Leaders help those within the system identify and access people with expertise relevant to new problems and opportunities.
• They minimize insularity: Leaders bring in connections to external resources in order to prevent groups from developing tunnel vision and becoming isolated.

These five activities help individual leaders encourage success within their systems.

Leaders need to nurture different types of networks. Ibarra and Hunter (2007) outline three networks that require active attention. The first, the operational network, consists of those needed in order to get work done efficiently, provide the functions required of their group, and maintain capabilities. These contacts include direct reports; supervisors; internal partners; internal players who can block or enable a project; and outside supply chain members, distributors, partners, and customers.

The personal network focuses on those who can help a leader grow personally and professionally. They can provide referrals to useful information, contacts, feedback on performance, and coaching. Personal networks support leaders during times of stress and uncertainty by giving them trusted resources and advice.

The strategic network includes people who can help address future priorities and challenges and help secure support for new directions. These contacts are future-oriented; they may be inside or outside the organization. The strategic network often needs the most attention by new leaders. Focused on building their operational and personal networks, many find themselves without the connections they need in order to achieve a future vision.

Even simply having some network knowledge can improve leadership performance. One well-documented case at Raytheon taught participants about network dynamics in a leadership program (Burt & Ronchi, 2006). In comparison to a control group, executives active in the business leadership program were more than twice as likely as the control group to far exceed expectations on performance reviews. They were three times more likely to be promoted to a higher rank and less likely to leave the company (pp. 1, 174).

Project Scope

The first step in performing a successful network analysis is to define the guiding question and population for the project. Typical network projects seek to answer questions about how people interact in order to get things done. Projects can focus on specific organization processes (like how the purchasing process functions), organization effectiveness (how people solve problems or collaborate), inter-organization relationships (how people cross organizational boundaries to complete work), or specific issues (how childhood nutrition programs are funded in a specific city).

The guiding question helps practitioners clarify the scope and boundaries for the project. Some guiding questions focus on a bounded population: a group that has clear boundaries about who is in and who is out of the group. A purchasing department, all project managers in an organization, and volunteer tutors are all examples of populations that are bounded. Of course, there are exceptions. The purchasing department may or may not include the administrative assistants who sit in satellite offices and who submit purchase orders, for example. However, the population has boundaries that can be fairly easily defined.

Bounded populations are ideal candidates for organization network analysis. A survey, usually web-based, is sent to all members of the population. A high response rate is needed in order to run analytics with any degree of confidence. Since practitioners know who has received a survey, it is relatively easy to encourage participation by following up with all respondents.

Other guiding questions focus on an unbounded population. In these cases, it is difficult to determine who is in and who is out of scope. For example, a project that focuses on relationships between funders of childhood nutrition programs includes so many different entities—government agencies, schools, private funders, hospitals, and so on—that it is difficult to place boundaries around the population. In these situations, a practitioner has a few choices:

1. The OD practitioner can employ a modified snowball sampling technique. In this practice, an initial set of participants is identified. They are invited to submit names of others they believe should be part of the project. This process continues—contacting participants and asking for nominations for additional participants—through several iterations. Once the project sponsors and OD practitioners have determined that the sample size is sufficient, they can launch the survey to all within the sample.
2. The OD practitioner can include in a network survey open-ended questions that ask people to identify who else should be included in the list of players relevant to the central question to be answered through the project. While those new additions can be invited to participate, those who have taken the survey prior to their inclusion will not be able to assess their relationships with the new group. Therefore, the analysis will be limited and unreliable. However, this approach can be used to expand an initial understanding of a core population and prepare for a more thorough, later investigation.
3. The OD practitioner can use a specific network mapping methodology that does not depend on identifying all members of a population. (See “Value Network Analysis.”)

Data Collection

Organization network data is usually gathered through a survey using a specialized network data collection tool like Network Genie (Hansen, Reese, Kelvin, Bryant, Bishop, Wynick & Dyreng, 2008) or ONA Surveys (Lee, 2011). The survey asks respondents to identify their contacts (Exhibit 29.2), indicate why they interact with their contacts, and estimate the frequency with which they interact with their contacts. Exhibit 29.3 shows a simplified version of the first step in a network analysis data collection: contact identification.

After completing this first step, respondents are asked why they interact with these contacts and how frequently they interact. Exhibit 29.4 shows a sample question screen. Note that the exhibit shows only those who were chosen as contacts by the participant in the first step, not all of the contacts on the master list in Exhibit 29.2.

After all participants complete the survey, the results are analyzed using mathematical algorithms. To conduct this analysis, many use software such as InFlow (Krebs, 2012), UCINET (Borgatti, Everett, & Freeman, 2002), NetForm (Stephenson, 2002), or NetMiner (Cyram, 2009). Information about various software tools can be found in “Overview of Common Social Network Analysis Software Platforms” (Monitor Group and the David and Lucile Packard Foundation, 2009).

Network Visualization

Networks are usually depicted visually. There are two common symbols on every network map: the node and the link. Circles or squares represent nodes or actors in the network. These nodes can be individuals, departments, organizations, websites, issues, or roles. Lines show links or interactions between nodes.

Figure 29.3 shows a sample network map of a small nonprofit organization. The guiding question that generated this map was: With whom do you exchange information in order to get your job done? The boxes designate departments, which are, starting from the box at the top center of the image and going clockwise: Admin, Development, Ed Sales, Production, Pub Services, and Systems. Within the boxes, each point represents a person. The lines show significant information exchange within departments. For example, within Production, the large box at the bottom, there are many lines between employees. This suggests that Production employees exchange information frequently; their work seems to require interaction, rather than isolated individual action only. There are also lines between departments, indicating that people work cross-functionally in order to complete daily work.

Network maps arranged by department, such as Figure 29.3, are easiest to interpret since they mimic the organization’s structure, which is already familiar to most viewers. Much value comes from also depicting connections in a spider map format. This format arranges people using mathematical algorithms. Nodes with higher centrality are moved towards the center of the map. Nodes with lower centrality are placed around the periphery. By using this format, we can discern patterns that do not show up easily in a map formatted by department.

The spider map in Figure 29.4 shows expertise interactions in a technology organization. The map is color coded by department; all the black dots, for example, work in the same area. The four departments depicted are largely self-contained. The largest cluster consists of mostly one department with one person (Number 12) serving as the gatekeeper to another department. The small cluster on the top left includes members of two different departments. However, Number 55 and Number 59 are not connected to the other members of their own department, shown in the cluster at the lower left.

Another interesting pattern that can be seen easily in a spider diagram is the centrality of specific individuals within the network. For example, Number 13, a hub, has direct connections to ten people. Many of these people have no other connections. If Number 13 were to leave the organization, it would flounder until another person filled the function performed by the critical connector. To mitigate this risk, an organization can create a mentoring relationship so that Number 13 shares responsibilities with another and, thereby, prevents the possibility of work slowdown in the case of his departure.

Data Presentation and Action

After collecting data, just as in any other action research process, the data is shared and collectively interpreted. There are specific choices to be made around how to share the data in order to maximize the value of the network analysis.

One option is to share all data, with name labels, with the entire organization. In this case, everyone can see how everyone else connects. This option can be risky: people can misinterpret data and draw harmful conclusions, such as George has no connections, therefore no one likes him and he will probably be fired soon. All participants need network education in order to successfully interpret network data. Organizations that share data with name labels tend to be new organizations or mature organizations in which there is high trust.

A safer option is to share all data without identifying labels with the entire organization. In this case, everyone can see the flows between individuals. However, since individuals are not named, the focus is on macro-level patterns across departments or groups. This is the preferred option in many network studies since it allows collective data interpretation without the dangers inherent in sharing all information. That said, people are usually clever enough to identify some of the names on the chart. Care is taken to educate individuals about networks so they can be responsible and accurate with the information.

A third option is to share a summary of findings and no maps with the entire organization. In this case, a small working group and the ONA consultant analyze the data and identify a short list of learning based on their interpretations. This list is shared with the organization and used as a springboard for collective discussion of the learning, including conversation about the validity of the interpretations and alternative interpretations.

After the presentation, discussion, and collective interpretation of network data results, the process follows a standard action research model. Specific interventions are chosen based on the findings and implemented. Many organizations choose to conduct a follow-up network analysis six months or a year after the initial analysis in order to gauge how well the interventions have addressed priority network issues.

VNA Applications

VNA has been applied successfully in many contexts and for many purposes. Sometimes considered the ideal process-mapping tool for the complexity era, VNA excels at identifying and analyzing non-linear, dynamic processes. VNA also effectively maps business ecosystems: a dynamic structure of interconnected organizations that depend on each other for mutual survival. Other applications of VNA include:

• Project management. VNA helps clarify roles and expectations; clarify monitoring and quality control techniques; and improve quality and collaboration.
• Organizational restructuring. VNA helps identify current and ideal future relationships, workflows, relationships, and value production systems.
• Knowledge networks and communities of practice. VNA helps demonstrate the value provided by knowledge networks and communities of practice.
• Talent management. VNA helps shift companies from managing individuals to managing roles and pinpointing core competencies needed by mission-critical value production processes.
• Intangible asset management. VNA helps identify and quantify intangible organizational assets, such as reputation, brand image, customer and employee loyalty, and workforce competence.
• Workspace design. VNA helps companies design workspaces that are congruent with their workflows and align with their collaboration goals. (Allee & Schwabe, 2011)

Conducting a Value Network Analysis

Value network analysis follows less of an action research model and more of a dialogic systems model (Bushe & Marshak, 2013) in that it engages people in articulating reality as they experience it, forming a shared vision of the future state, and analyzing the system in the moment based on their collective picture. VNA is typically conducted in a facilitated meeting in which, in the course of three to eight hours, a group develops their network map. After creating the map, they ask specific questions designed to test the resiliency, effectiveness, and use value of the system. Typical questions asked during this stage include:

• How much reciprocity is there between roles? Are some roles giving much more than receiving? Are others receiving more than giving?
• Is there an appropriate balance of tangible and intangible inputs and outputs between roles and within the network as a whole?
• Are there particular roles that are carrying the burden of network interactions and, as a result, pose a risk to the network?
• Are there patterns of intangible transactions that might be supplementing the existing process in inefficient or ineffective ways?
• Where are the missing or dead links?
• Where does a value flow get stuck or break down?
• Where does a role fail to transform inputs to appropriate outputs?

After analyzing the value network map, two things typically happen. First, groups create desired future maps that show how they would like the system to operate. Second, groups make changes to the mapped system based on what they have learned.