Donors, governments, NGOs, and other stakeholders in the social sector are increasingly interested in how to spread effective solutions and innovations to everyone who can benefit from them. A growing literature on the topic is forming, as researchers and analysts describe general principles for building collective will and setting shared aims, designing scalable and charismatic interventions, and leveraging existing networks and institutions to have bigger impact.
In our view, however, there remains a crucial gap in this collected knowledge. While the attributes of effective learning networks (i.e. structured methods to support knowledge exchange and spread new behaviors) have been described in general terms, few attempts have been made to define and distinguish the approaches that have been applied to greatest effect. Building on our work in the last two decades, mainly in international health, we attempt to address that gap here by documenting ten approaches to orchestrating networked learning at various scales.
Our experience suggests that no single approach is preeminent. Organizations seeking the best way to “many” shouldn’t ask which is the “best” method; it’s much better to consider which path will work best for a given circumstance. The appropriate networked learning methods for spreading a complex care protocol to clinics in rural South Africa will differ significantly from the method to spread a simple practice to urban hospitals in the United States. (The table below enumerates key factors in determining which methods to apply, and when.)
Though the ten approaches we identify differ in important ways, several share notable characteristics:
- An emphasis on shared, quantifiable aims that ties together the work of participants
- A rejection of the idea that passive dissemination (through books, meetings, web sites, etc.) can lead to meaningful behavior change
- A commitment to high-tempo application of knowledge (i.e., daily or weekly testing of new practices, reviewing what’s being learned, and revising approaches accordingly)
- A commitment to connecting frontline actors to one another to accelerate learning
- An emphasis on providing just-in-time access to the practical “how to” details of applying an innovation or solution (also known as tacit knowledge)
- Provision of actionable data, arrayed over time (in formats like run charts and control charts) that give participants regular feedback on progress to inform learning
- Opportunism in taking advantage of existing networks and infrastructures (such as professional societies, associations, and interest groups) to mobilize participation (versus building networks from scratch)
- Willingness to use many technology platforms to facilitate interaction and learning
While not a comprehensive list, here (in alphabetical order) are the ten successful models that we have applied and studied:
Breakthrough Series Collaborative
The Breakthrough Series Collaborative, invented by the Institute for Healthcare Improvement, is a highly structured method used to bring a number of organizations (usually between 10-100) together to close defined gaps in practice. The collaborative asks all participating organizations to work together to introduce a group of evidence-based actions, frequently called a “change package,” over several months. Organizations are accountable both to the shared aim and to each other through regular reporting on progress toward common metrics and through discussions of challenges. This model couples group learning with immediate action in order to accelerate implementation.
Behavior-change campaigns aim to spread simple, turnkey practices through large distributed networks of local or regional field offices (“nodes”). Mobilizing national, regional, and local actors behind a bold, quantifiable goal, they seek to raise the floor on performance across large geographical areas, making sure that straightforward practices are introduced with high reliability. Nodes in these campaigns often apply other networked learning methods listed here (such as collaboratives and extension agent models, described below) to support local learning.
Communities of Practice
Communities of practice tend to focus on open-ended explorations of new knowledge, therefore differing from most other networked learning methods described here, which focus more on spreading established practices. Etienne Wenger defines three necessary elements of a community of practice: (1) a shared domain, (that is, a topic); (2) actively fostered relationships that create a community; and (3) a practice of applying and working on the domain at hand. Communities of practice have neither pre-determined membership nor pre-determined duration; rather, they evolve organically to suit the needs of their participants, often in the absence of explicit aims. In our experience this can lead to diffuse energy and limited impact, though there are exceptions. When trust between participants is deep and intentions are clear, more impact seems possible.
Extension agent models, pioneered by the US Department of Agriculture, rely on itinerant individuals who travel from site to site (e.g., school to school, nonprofit to nonprofit, clinic to clinic) to share innovations, troubleshoot challenges, and harvest best practices for further redistribution. This method is highly effective when the extension agent acts as a connective tissue between far-flung actors, a cross-pollinator of learning. This model can improve the effectiveness of existing networks by contributing an important face-to-face connection that deepens relationships and supports local implementation of new ideas in a hands-on way.
This model harnesses the techniques of games (i.e., pursuit of an objective in a competitive environment, clear rules, ranked performance) and applies them to tackling a collective problem to increase engagement and accelerate learning across a group of participants. Games draw in participants with the possibility of fun, rewards, and a sense of accomplishment, and have the added benefit of fostering group discovery when the individuals and organizations that play have to solve shared challenges in order to perform better and win. Online games are also scalable, requiring fewer resources than methods that require active instruction or in-person exchanges.
In this approach, members of a given group or community take responsibility for mobilizing others in pursuit of a goal in their shared interest. They start by making close one-to-one connections with other individuals in the community, surfacing shared values and common purpose. As individuals are recruited to the cause, they then commit to recruiting others in the same way. This method can build large, tightly connected networks of volunteers who rapidly exchange knowledge, experience, and tactics in pursuit if their goal. In this way, it bolsters many of the other strategies listed here.
An incident command system (ICS) is a management approach widely used in response to emergencies and natural disasters in the United States and other countries, but it can also be used for planned initiatives. Employing a strict chain of command and sometimes suspending standard, deliberative decision-making processes, it can mobilize a large group of actors to take rapid, coordinated actions toward a specific objective. It relies on a crisp, cross-organizational structure in which the roles of all stakeholders are well defined. As a method to spread learning, incident command usually requires a highly attendant audience working to address a problem or apply a solution under tight timeframes. It is often appropriate as a temporary approach in a larger initiative, or for initiatives with limited timeframes.
Innovation competitions, which share some things in common with gamification approaches, issue a challenge to a diverse group of people or organizations to figure out how to solve a shared problem. They incentivize problem-solving through recognition (e.g., monetary awards, free coaching, celebration) and, at times, access to expert input. Usually participating individuals or teams do their work independently, but increasingly competitions allow them to study each other’s ideas and even collaborate in successive rounds of competition.
“Wedge and Spread”
When the innovation or practice a group seeks to promote is delivered by separate but interdependent organizations (for example, a health system consisting of tertiary, secondary, and primary care sites, or a multi-site school district), the “wedge and spread” (or “wave sequence”) method can be appropriate. The “wedge” describes one representative slice of the integrated system in which piloting is done to perfect the service or practice in question; the “spread” occurs when actors in the “wedge” teach peer organizations what to do.
These projects—applied in both for-profit and nonprofit sectors—use a highly structured approach to develop, test, and refine a promising concept, determining its readiness for broader uptake or confirming its failure over 90 days. Such efforts are often undertaken by multiple organizations working together to leverage one another’s experience and resources, and to more rapidly develop knowledge. To ensure that the organizations act on what they’re learning, this approach calls for decision points scheduled at regular intervals over the 90 days and immediately upon the project’s completion.
The table below offers examples of each of the ten methods.
Examples of Networked Approaches to Learning and Improvement:
Factors unique to an organization, a movement, or the moment in time should guide the approaches change leaders select, and these leaders should feel empowered (not overwhelmed) by the variety of options at their disposal. Change leaders also should not feel constrained by any single option; many successful large-scale initiatives have brought elements of several approaches together to create effective hybrids. For example, initiatives can use the 90-day Project method in parallel with other larger-scale methods (e.g., a campaign) to allow a vanguard group to deeply explore of an emergent challenge while simultaneously ensuring adoption of simpler, established practices among a broader audience.
The table below attempts to describe the conditions under which each method defined above might be appropriate, using six factors (that draw in part on the seminal work of Everett Rogers):
- Nature of the intervention. Is the intervention simple or complex? Well-documented and evidence-based, or under development?
- Stage of development. Where does the work lie on a continuum from early concept through broad culture change? (I.e., Is the effort focused on concept development, prototype, pilot, spread, movement, institutionalization, culture change?)
- Dimensions of the target audience. How open or opposed to change is the group of people being targeted for change? What is their readiness and experience level? Do other contextual factors (e.g., geography) matter?
- Level of control. How much influence and authority do you have over the target audience?
- Resources. Which are available?
- Timeframe. What is a realistic expectation for how quickly change can happen? Is the time pressure acute?
Conditions Under Which Each Method Is Used:
Execution, Above All
While selecting an appropriate learning network structure is critical in designing a large-scale change initiative, success depends on how the learning network is brought to life, week in and week out. Successfully executing any method requires an operating culture wherein a rapid rhythm of learning is established, improvements are constantly sought and made, improvisation is the norm, and recognition of small victories and joy fuel the work.