On June 10, 2014, the 100,000 Homes Campaign announced that it had exceeded its four-year goal and permanently housed 101,628 chronically homeless Americans—one month ahead of its self-imposed deadline. That number reflected the collective work of 186 communities that joined the Campaign and represented an estimated annual taxpayer savings of $1.3 billion.
How do we achieve impact at a scale that meets today's enormous needs? Explore Bridgespan research, insights from leaders, and more on the Transformative Scale Resource CenterVisit the Resource Center
The Campaign didn’t get there by replicating a set program model. Instead, it created a geographically dispersed network of organizations with a shared goal (ending chronic homelessness) and a core set of design principles and operating practices (for example, starting by identifying every homeless person on their streets—by name). Participating cities shared data and insights, and each remained open to innovations that fit local needs. Borrowing a term from the Monitor Group, I’ll call this approach an “aligned-action network.”
To be clear, an aligned-action network differs from collective impact, another type of action-learning community, in an important way. Collective impact typically assembles organizations to work on social issues in a city or region, delivering local impact. In contrast, aligned-action networks address the same issue in multiple locations, aiming for impact and scale. Both approaches are part of a groundswell of interest in the power of networks (see Monitor Institute; Wei-Skillern), emergent strategy (see Mintzberg; FSG), evaluation as learning (see Patton; Eckhart-Queenan and Forti), and communities of practice—all of which point to the possibility of a path to scaling impact that looks different from the typical replication strategy of a single organization. (StriveTogether is an interesting example of scaling a collective impact process across geographies—a hybrid of local and national strategies.)
When I reflect on the 100,000 Homes Campaign’s experience, and that of others pursuing similar approaches (e.g., BRAC’s work with the ultra-poor; the Institute for Health Improvement’s “breakthrough” collaborative), I keep returning to a thought experiment. At the risk of over-simplifying, I’ve found this comparison helpful:
In a replication scenario, a program is highly prescribed and only those willing to sign on to its full adoption and embrace fidelity to it join the system. A relatively modestly sized central office manages the process of adding organizations and ensuring their compliance to the program design.
In another scenario, an aligned-action network brings together organizations that share a common outcome and similar programmatic approaches, but the focus is on disciplined learning and innovation—not strict adherence to a single model. A relatively small central office manages the process of adding organizations to the network, maintaining a common data platform, and coordinating opportunities for shared learning.
Now imagine each approach addressing the same social issue, say, chronic homelessness. Here’s my thought experiment question: In five years, which would have greater impact measured by the number of lives touched and depth of the impact on each life?
The emergence of aligned-action networks comes at a time when leaders in the social sector are increasingly frustrated by the limitations of program replication. Replication is the gold standard for spreading programs that work, having enabled organizations like Nurse-Family Partnership, Success for All, Youth Villages, Year Up, and many others to reach tens of thousands of people who benefit from their interventions. But it’s typically a relatively slow-growth strategy; in most cases over 95 percent of the target population remains untouched by the program—or any evidence-based program.
There is also the troubling pattern of replication sites’ results falling short of the initial site’s. No matter how “proven” the model, implementation on the ground determines whether we achieve impact at scale. Prescriptive models can make it hard to adapt appropriately to local environments, and the center’s tight control can weaken local actors’ motivation. Moreover, the lack of variation in the system can dampen the sparks of innovation that fuel improvement. (As a window into what’s required to implement with quality throughout a replication network, see Dan Cardinali’s recent article on Communities In Schools’ $50 million investment to strengthen its operation.)
Conventional wisdom holds, though, that without tight replication control, quality surely suffers. Models like the 100,000 Homes Campaign seem to be challenging this belief, suggesting that centrally-driven “tightness” may not be the only way to impact at scale. Interestingly, we’re seeing counter examples in the for-profit world, too, as companies like McDonald’s and Burger King now accommodate a much greater degree of variation in menu and restaurant appearance—and have shifted significantly towards more franchisees versus company-owned units in their systems. Less control, in fact, seems to be good for business. Is there a lesson here for the social sector?
So, what conditions might need to be in place for an aligned-action network to outperform replication? Several come to mind:
- Alignment around a common outcome and a shared measurement and data collection system that all players participate in (technology opens up new possibilities on this front, which is in part what has propelled the shift noted above in for-profit multi-site service industries);
- Participants that are committed to impact, have incentives for improvement, possess the capability to use and analyze data, and share a learning mindset;
- A platform (in most cases in the form of an intermediary organization) that has the capability to capture, make sense of, and share insights from the data in actionable ways. (While this may sound simple, it is rare to find entities that blend robust research capabilities with the capacity to help actors implement and test innovations.)
This is not about letting a thousand flowers bloom—it’s more like a well cultivated garden. (Extending the metaphor, replication strategies might be viewed as a greenhouse of orchids which are highly sensitive to any change in conditions.)
Does an aligned-action approach ultimately perform better than program replication? Myriad factors, of course, might play into the answer—the nature of the intervention, for example. But it’s an important question that needs more clarity. An affirmative answer would signal a need to figure out how to get more of these networks up and running—and call for a deeper understanding of what kinds of platforms support excellence and how best to fund them.See more transformative scale blogs
Rayshawn and Anna & Dana - Thanks for your thoughtful additions to the thinking about aligned action networks (and for the work on many of the original notions about them from Monitor's earlier research.)
Your comments point to things that should be considered when conducting the thought experiment -- or ultimately, hopefully, an actual natural experiment -- that might compare replication to aligned action networks approaches to scaling impact. First, the experience of the local actor in these two settings can be quite different -- perhaps as a result of the kinds of people attracted to them, the incentives, the opportunities for visibility created, etc.
Long ago I did research on franchise systems and found dramatic differences between company owned and franchised stores (can be found in the book Franchise Organizations.) Despite efforts to make company managers feel and act more like "owners", the difference between them were stark and the levers affecting their performance were very different.
Anna and Dana, your points about the dimensions of impact that might be triggered by aligned action networks is thought-provoking (and hinges on some of the different kinds of behaviors enabled and supported by such networks.) So maybe two experiments are needed: one with a relatively simple intervention or practice and a second with some set of activities addressing a wicked problem. As you note, it is very easy to see why the aligned action network might be best in the second instance. It feels like a bit of an open question about which one might outperform on the first experiment. In any case, more data needed, which I know people on this thread (and others) are working on.
We are excited by the points you make and the questions you raise about Aligned Action networks and scaling impact—all important issues for the field to explore more intentionally. For too long, scale has meant programmatic replication—and considered the path to impact rather than one of the many ways to get there. The fact is, we need strong organizations with the right amount of reach and also we need to work in and through effective aligned action networks where impact can be achieved in different ways.
Over the years, we at Monitor Institute have been developing Aligned Action networks that mobilize disparate groups of key stakeholders—from a handful of organizations working in the national capital region on youth development issues to close to two hundred organizations working on reimagining learning in this country. These Aligned Action networks become robust platforms that can launch a portfolio of interventions and simultaneously pull many levers for impact. They are bound by a shared narrative for change, fueled by a set of principles for what matters and motivated to measure, learn, and adapt along the way.
You raise a critical question around whether an Aligned Action approach ultimately performs better than program replication. Aligned Action networks pursue a broader range of impact than individual programs, mainly because they're focused on systemic outcomes. Impact certainly means the "beneficiary effects" of number of people served, percentages of "x" increased over "y" time frame and other tangible measures. In that sense, a network's impact can be compared directly to that of a program. But it also includes impacts around innovation; the spread and adoption of ideas (through broad culture change efforts or by specific policy mandates); changes the network has promoted in the development of a field (shifts in funding streams and other infrastructure resources); and, the development of network's own structure, membership, connectivity, and activities, which lay the groundwork for future growth.
This broader impact aperture matters because Aligned Action networks are better suited to tackle wicked problems that are cross-cutting, systemic and adaptive than technical problems where the problem is defined and there is a clear right answer. Of course, this makes measuring the impact of networks more difficult, and it will require that we advance our notions of measurement in significant ways, but that is a challenge that we are certain this dynamic field can overcome.
Thank you for sharing your thinking on this Jeff. Your assertion that "No matter how "proven" the model, implementation on the ground determines whether we achieve impact at scale" resonates strongly with me and raises another question "what are the drivers of success during implementation?"
You've highlighted autonomy as the first driver of successful implementation. I believe there may be two other important drivers of successful implementation.
The second driver of successful implementation is incentives. In order to assure outcome and learning fidelity across the aligned-action network the central office has to nail incentive design. The 100,000 Homes Campaign seems to have done that with their homelessness audit. It's one thing to not achieve an outcome for a homeless family, I will feel badly but that's a statistic, when I'm not housing Joe Thompson and his two kids I'm letting down one of my neighbors.
The final driver of implementation success is attention. When I think of the importance of the central office, capturing, making sense of and sharing insights is important. The central office's ability to put an affiliate on a pedestal is also crucial. I once witnessed a flurry of rapid prototyping and iteration in a rural Ghanaian village after a farmer was recognized in a community meeting for improving on the design of a cellphone charger that ran off of D-cell batteries. Shortly after this meeting several famers presented versions of the design that were either more efficient or less expensive. I talked to the village chief and suggested that some of the farmers must be obsessed with technology. He corrected me pretty quickly and said "it's not the technology they're interested in, it's the attention."
This may be a concept worth exploring as Bridgespan continues to develop an understanding of what allows leaders and funders from across the social sector to attain impact at a transformative scale.
Jeff and Mark - Thanks for the comments and additional ideas. As you say, Jeff, the core issue is "spreading practices" -- which sometimes might occur in the context of "programs", other times through intentional sharing/learning, and yet others through simple mimetic dynamics (person X is doing it, I respect them [or I see they get rewards for it], so I am going to copy them.) Some of the most interesting writing done on the diffusion of practices is by Atul Gawande in the New Yorker (see "Slow Ideas" -http://www.newyorker.com/magazine/2013/07/29/slow-ideas.) Highly recommend.
Mark, your last point is very thought provoking: I agree that the cost factor is a vital one to consider in comparing replication and aligned-action networks/CI, and indeed, in some cases CI may offer a better deal. But it is important not to underestimate the costs involved in influencing practice across locations in many-- and I suspect most-- cases.
Perhaps more vexing is the impact question. While RCT's have many limitations (documented well elsewhere), the method helps give confidence in the impact of certain interventions and/or programs. It seems to me that there is still a lot of work to do to demonstrate empirically the actual impact of the aligned-action network/CI pathways in many cases. Granted, it is much more complicated to measure than a simple single service being spread, but before assuming it outperforms we need a deeper look at its impact (including the extent to which it is embedded and endures in its context, which when working well may be one of its strengths.)
My main point is that your statement "Sharing of information, inspiration, and ideas is fast and free" may not tell the whole story. "Sharing" doesn't necessarily lead inexorably to impact, and often ideas left to their own devices actually travel quite slowly (e.g., handwashing in Atul Gawande's article which took decades to spread or high impact home visitation practices/programs today.) Some Bridgespan colleagues are doing work on "demand", as it is startling how many things that "work" do not actually spread naturally.
The work the two of you have done with your organizations - and many others noted in the original blog -- is critical to us finding new pathways to impact at scale. Understanding which of these pathways have impact, what organizational scaffolding is required to support it, the cost dimensions of the different approaches, and the conditions under which one or others might "work" is critical work for us all to do if we are to find ways to address some of society's most important challenges. All of this makes me think we need to convert the "thought experiment" sketched in the original blog into a real research project, which doesn't stay at the abstract level, but finds ways to compare replication and aligned-action networks more rigorously to get at these questions.
I completely agree with your observations that aligned action or collective impact can be a powerful alternative path to scale compared to replication. I would only add a couple thoughts:
Collective impact (CI) is often applied in a specific region, but it also works at a national or global scale. The Global Alliance to Improve Nutrition and Roll Back Malaria are two of several examples cited in our SSIR article on global partnerships that fit the CI model while operating in dozens of countries around the world.
It is true that there are many competing frameworks for collaboration, and the three bullet points you describe as key to aligned action are also elements of CI. There are differences among these approaches, however, and in our experience, all five elements of CI matter to its success.
Finally, when comparing CI to replication, it is important to note that CI works with existing organizations and resources. Many of the solutions that arise through CI have no additional cost or even save money. Scaling through replication requires substantial investments of new capital in order to create new entities or expand existing ones. Scaling through the sharing of information, inspiration and ideas is fast and free. Even if the two approaches to scale were equally effective in the number of lives reached within five years, the impact per dollar spent may be substantially greater through CI. And given the scarcity of philanthropic resources, that matters a lot.
This is an outstanding and thought provoking piece, Jeff. Thanks so much for codifying this for the field. One of the concepts we struggle with conveying and that I think you have been able to get right is that collective impact really shouldn't lead to people working together to scale isolated programs. We actually try to encourage a shift from "scaling programs" to "spreading practices" as a way to get people out of the traditional mental model that a single program could ever reach every child that needs it. The idea of Aligned Action Networks could give us a way to communicate how important it is for partners to adopt, incorporate and improve practices to achieve better results at the population level. Thanks again and we will do our best to share this broadly.