The formal structuring of work has been found to be particularly useful in coordinating virtual collaboration (Hoch & Kozlowski 2014; Steinhardt and Jackson 2014). A fundamental approach to structuring work involves breaking tasks down into discrete, manageable chunks. This notion of task decomposition is at the root of everything from the division of labor to the work breakdown structure process in project management (March & Simon 1958; PMBOK 2001). The principles of modularity, which involve breaking up of work into well-defined chunks with specific interfaces, are particularly critical for managing complex science tasks in virtual collaborations.
Research into the many forms of virtual collaboration emphasize the power of modularity. For example, modularity is critical to successful distributed technological innovation, outsourcing relationships, and distributed software development (Baldwin & Clark 2000), including open source software projects (MacCormack et al 2006). The key to modularity are the two components: (1) break up work into discrete tasks, and (2) create well-defined interfaces between these tasks.
Task decomposition
The first step involves hierarchical task decomposition - to take the complex task and divide it up into smaller tasks. In project management this step is often referred to as creating the work breakdown structure. Take large tasks and break them up into smaller tasks, and break these up to even smaller tasks until you find the appropriate unit of work. Hierarchically enumerate these tasks. In project management practice, one then typically takes this work breakdown structure and organizes it according to a schedule in a Gantt chart (PMBOK 2001).
Of course, this process sounds simpler than it actually is in scientific projects. In scientific research, goals and outputs are not well-defined in advance and often change throughout the course of the product (Turner & Cochrane 1993). Simply, it is not quite apparent how to breakdown a team’s work in advance. A best practice for dealing with this is to identify key deliverables and the information requirements for those deliverables (Turner 2000). In an exploratory situation one can never perfectly predict things like the information content nor the duration of tasks, but one can predict the sorts of information that will be needed for subsequent steps. Deliverables of data collection involve the data. Deliverables of analysis may involve lab reports or statistical analyses. These deliverables can be further broken down.
The role of deliverables
An important note about deliverables is that they do not just mark the end and completion of a task, but that deliverables are used for many and multiple subsequent tasks. If one thinks about deliverables as the interface between tasks (i.e. all subsequent tasks) then the conversation moves to structuring the sorts of information - or the information requirements - of all subsequent tasks. It is critical to align leadership responsibilities, communication and incentives (such as reward systems) with deliverables to drive better performance (Hoch & Kozlowski 2014).
Even if work breakdowns can only be projected out for a short period of time, beyond which tasks are unknowable, work breakdown can be worthwhile for its side-effects. This is because working to accomplish simple tasks in a way that is visible to others in the initial stages of a collaboration can provide trust and knowledge of each other’s skills that can enable the group to adjust their work over time (Iacono & Weisband, 1997; Mitchell & Zigurs, 2009). Thus establishing a set of simple tasks whose accomplishment and deliverables are visible to each other is worthwhile almost regardless of the specific content of the tasks; certainly visible, structured, work is a promising alternative to long meetings early in a project. Further, science teams will have difficulty developing an accurate project schedule from a work breakdown structure, but the exercise of creating and targeting dates and date ranges for deliverables assists in coordination of a project. The importance of goal setting is perhaps the most well-established factor for better performance among innovative teams in the existing body of research (Hoegl & Parboteeah 2003), and the act of breaking down work, defining deliverables, and determining initial schedules are the fundamental activities in goal setting.
Meetings
So how does this all translate into structuring a virtual collaboration? First, it is well-established that face-to-face kick-off meetings are particularly important for virtual team success (Hertel et al 2005; Ferrazzi 2014). Kick-off meetings help build interpersonal relationships and trust that will be invaluable throughout the collaboration. But, as we discuss further below, the kickoff meeting should not be unstructured. During that kickoff meeting it is important to structure the work and to involve project participants in goals setting. Identify key deliverables, breakdown these deliverables into their lower level tasks, identify information requirements of tasks and codify them as standards for deliverables and assign teams to the different tasks.
Focus the subsequent virtual meetings and documentation on deliverables with participants from the appropriate teams responsible for these deliverables. Team members do not need to participate in meetings, conference calls, or email chains, unless they are specifically associated with a given task for a deliverable. To the extent possible, schedule meetings around deliverables and include only those assigned to particular deliverables for those meetings and included in those tasks.
Limits of modular work
It is important to note that complex tasks like science projects can rarely be considered to be perfectly modular. They inevitably involve unforeseen interdependencies with other tasks in unanticipated ways. Nobel prize winning economist Herb Simon referred to this as the “near decomposability” of complex systems (Simon 1962). So it is important to realize that a perfectly modular work breakdown is likely unattainable and no team will get it right the first time. So teams should occasionally coordinate across modules. This cross-module coordination involves periodic cross-team meetings, but these cross team meetings should focus on deliverables and information requirements to reduce unnecessary discussion. Broader meetings across groups (i.e., meetings across groups responsible for different deliverables) should be scheduled before, during, and after deliverable handoffs, and when interdependencies become evident.
A note about scale
An important note is that the coordination requirements of virtual collaboration scales with the size of the virtual team (Boh et al. 2007, O’Leary & Cummings, 2007); bigger teams, teams with more diverse expertise, and teams from different organizations and in different time zones all require more structure to avoid the inefficiency of too much ad-hoc, discussion-focused coordination that may be fine for smaller, homogenous teams (Cummings et al 2013; Nguyen-Duc et al 2015).