There is a long history of experiments to align the actions and thoughts of many different organisations trying to achieve some kind of social change – better early years education; less violence; improved public health or urban regeneration. These have usually involved some combination of shared plans, targets, and commitments. They have had many names. Many books have been written about them, and many universities have run courses to make people better collaborators. People on the sharp end of social problems want those with money and power to get their act together and not work at cross purposes. Responsible funders naturally want to find ways to make their money go further, and recognise that this is bound to involve collaboration – pooling resources of all kinds with others.
Yet one of the oddities of the field is that there is not much cumulative learning. I regularly come across reports and articles, and outputs from consultancies, claiming to have invented a new way of doing this. They’re perfectly well-intentioned, and many are doing important work. But they rarely make much if any mention of any of past experiences and appear to be unaware of the lessons learned. Sometimes they seem rather less thought-through than their equivalents in the past. A field which should be visibly progressing can look like Groundhog Day. And some of the methods sound more like new ways of empowering funders rather than empowering people themselves.
In this short note I attempt a personal view on what collaboration and collective impact are, what has been learned, and how practice could improve, hopefully to prompt more argument and less unnecessary reinvention.
Why now?
The topic of how to achieve large-scale, cross-sector collaboration to deal with social problems is an old one. But it’s gained in importance because of greater connectedness and interdependence. Many big tasks require cross sector, multi-partner, multi-stakeholder collaboration, whether in a geographic area or in a sector. Many tasks also require some shared institutional capacity to coordinate and drive actions and this has become a strong theme of innovation, in the global south as well as the north.
The good news is that are thousands of examples. Some have been truly global, like the WHO strategy to eliminate smallpox in the 1960s, the collaborations on vaccines in GAVI, and the strategy to fight Malaria in the 2000s. Some have been national, like the National Community Development Initiative (NCDI) in the US in the 1990s or the various collaborations around coding and digital making in the UK this decade.
But most of the action in this space has been in cities: there are innumerable urban development variants with partnerships, joint boards, and an array of tools to make regeneration more holistic and systemic. A generation ago these were most commonly found in city centre partnerships trying to turn around de-industrialising cities like Cleveland in the US, or Manchester in the UK, and there are standout examples of sustained multi-sector collaboration like Bilbao.
The 1990s also saw many attempts to package up collaboration and collective impact under brands: one which was visible for a while was Communities that Care. Twenty years later there are innumerable live examples from Memphis and Cincinnatti’s STRIVE to the Harlem Children’s Zone in the US, McConnell’s work with many agencies in Montreal, to current plans to use similar methods in West London and Suffolk (one recent repository can be found here). There have also been very ambitious public-sector led programmes to align actions and resources, such as Total Place and the London Collaborative.
Although most of these are hard to evaluate because of the sheer number of variables involved, and often don’t last long enough for serious assessment, there are some good stories to tell (see for example the debates about the effectiveness of the HCZ and the long-term impact of holistic urban regeneration schemes). And there are many organisations trying to improve how it’s all done, from Collaborate in the UK and CSI in Australia to FSG in the US and elsewhere, to my Harvard colleague Jorrit de Jong’s work on organised crime in the Netherlands, as well as plenty of serious academic analysis, such as classics like Eugene Bardach’s work on cross-agency collaboration, as well as more recent books.
At Nesta we’ve been particularly interested in recent years in methods that combine methods to create strong relationships between partners; strong backing from system leaders; clear goals; intensive periods of joint work to achieve results; systems redesign and citizen empowerment. We set out some of the thinking in various papers on systems change, and have been successfully working with partners including the Rapid Results Institute to apply these methods to health systems, focusing on reductions in unnecessary hospital admissions.
One of the most interesting US examples was the National Community Development Initiative. In the 1990s this pooled the budgets of a group of foundations, banks and others to create an infrastructure of community finance across the US, aiming at a version of collective impact, both at a national level and in multiple poor communities. It directly committed $250m from 7 foundations and private partners, and leveraged around $2bn and created a network of hundreds of Community Development Corporations and national bodies LISC and the Enterprise Foundation (Rockefeller has kept its work alive in the Living Cities programme).
The tools
If you want to get the players in a city or sector to collaborate you have many tools available. All will involve some joint activity to diagnose the problem, which then leads onto commitments towards action to solve it, and then some way to monitor results and adjust. This isn’t rocket science, and has been described in umpteen very similar frameworks and diagrams. It’s fairly similar to the better standard policy design methods.
As I show later there’s also a wide array of new digital methods that can support collaboration. One recent example was the Washfunders initiative which combines 17 different data streams to produce an interactive mapping tool of funding in the area of water and sanitation. Others include social network analysis and systemic change tools.
But although on paper the job of deciding on tasks, defining metrics and following them through with performance management tools, can seem easy, in practice each of these steps brings challenges, and the greater the mismatch between the shape of the problem and the shape of the organisations with some power to act on it, the greater will be the challenge of coordination.
It’s a problem if the basic diagnosis is flawed; a problem if the actions committed to aren’t the right ones; a problem if the participants look at issues through radically different frames; a problem if the commitments are weak; a problem if there isn’t sufficient implementation capacity; a problem if street level cultures aren’t right; a problem if the metrics don’t capture what’s important.
All of these are the usual problems of any kind of complex problem solving. They require a heavy dose of humility along with commitment to work through them. They also require very particular skills. The ability to curate, facilitate and nudge a disparate group towards common purpose is extraordinarily important, requiring tact, wisdom, strength and an odd mix of charisma and quietness.
Light or heavy?
The tools then run in a continuum from the light to the heavy. The light ones involve mutual coordination: public commitments but without legal force, and some use of open data collection and shared metrics. This is roughly what the European Union does at a national level: the principle of open coordination is that explicit targets and transparent actions will encourage convergent actions. It’s also what most of the recent projects labelled ‘collective impact’ do. They are coalitions of the willing that work so long as the partners are willing.
The heavier tools formally combine budgets, targets, teams, reporting requirements over long periods of time and legally binding obligations. Many of the urban regeneration collaborations were heavy in this sense, with formal powers and rewards to incentivise sustained collaborations – including models like the Single Regeneration Budget, New Deal for Communities or Total Place in the UK. NCDI in the US was another example of relatively strong tools, with formally pooled money, as is the programme underway in Montreal led by McConnell. The many EU funded collaborations also tend to be strong in this sense (even if they are often in reality quite cosmetic).
A common lesson is that although structures and processes matter greatly, cultures matter more. This is often a blind spot for the more technocratic, or mechanistic, approaches to collaboration. So in one view the best collaborations are grown more as movements than as coordinated performance management; the main focus of work is on relationships and trust-building. Nesta’s recent work on systems change in fields like health has prioritised this sort of work – strengthening horizontal links and confidence between public sector staff, professionals, NGOs and others – as a vital complement to top down authorisation from leaders and the use of formal metrics. This was also an emphasis for the London Collaborative – building much stronger informal networks and trust to complement the formal structures and processes.
A generation ago it was assumed that these kinds of collaboration worked best if they were supported by new organisations – partnership boards &c – to act as intermediaries and coordinators. But experience showed that although in some circumstances these can help with achieving focus, just as often they create more problems than they solves, because they multiply boundaries and transaction costs.
Equal partners or dominant players – what role for government?
The tools also run in a continuum from government-dominated to wholly bypassing government. For most social problems government is likely to be better suited (in theory) to acting on the problem, having an order of magnitude more resources, more legitimacy and more information. But this primacy can become a problem if government is inflexible, wedded to old solutions, or poor at partnering. It’s also bound to be itself fragmented, especially where two or three different tiers of government are involved. So a lot of subtle knowledge has built up over the years about how to get the benefits of collaboration without government smothering everything.
There are many interesting examples of attempts to square this circle. One unusual recent example is Delivering Social Change in northern Ireland, substantially funded by Atlantic Philanthropy, but implemented by government, and aimed at tackling a range of issues including dementia and integrated schooling.
The best large scale example of collaboration – with explicit targets, large scale public engagement, and commitments from many agencies – have been associated with governments rather than private foundations. In the US the Oregon Progress Board and indicators, was a model for many years. More recently South Australia’s State Strategic Plan has been a model (if anyone knows of anything as serious and systematic, I’d be interested to know).
Structure, process and culture (and why actions may be a better starting point than principles)
The general lesson from most of these exercises is that structures help, but are less important than processes, and that these in turn are less important than cultures. Announcing the collaboration and getting the principals to turn up is (relatively) easy. So is the paraphernalia of shared metrics and reporting. But lasting impact depends on much deeper rooted habits of collaboration, for example between key professions on the ground. This tends to take a lot longer. The joined-up-government method has thrown up lots of options for all of these – joint boards, teams, reviews, policy making, implementation teams, targets, budgets, double and triple keys. But these tend to have most impact when they are given enough time to become embedded in everyday cultures, and are reinforced by the main drivers of behaviour, from how leaders act to appraisal and pay systems.
Another common lesson is more counter-intuitive. Organisers of collaboration often want the participants to get back to first principles, and then move logically on to actions. A great deal of experience, and political science, shows that this is mistaken. It is much harder to get a disparate group to agree on underlying principles and values than it is to get them to agree on actions.
Why collaboration is hard – and not always desirable
Any kind of collaboration can be frustrating, imperfect and time-consuming. The literature which makes it sound neat and logical can be misleading. Herding cats may well be easier than the job of getting disparate organisations, with radically different cultures and accountability arrangements, to work in harmony. But there are also more fundamental challenges in collaboration, barely mentioned in the recent literature.
The first is that collaboration brings costs as well as benefits. Silos exist for a reason, and reinforce themselves for good reasons too: it’s not possible for everyone to talk to everyone all the time. As I have shown elsewhere, any anti-silo measure brings its own trade-offs, and its own new handovers. In the 1990s the joke used to be that the only thing worse than organisations never talking to each other, was organisations doing nothing but talk to each other. The critical question for any horizontal collaboration is whether it adds more than it subtracts, since it inevitably creates new boundaries, and new problems of coordination.
A second critical issue is that when radical innovation is required, bringing incumbents into a collaboration can slow down necessary change. Imagine a cross-sectoral collaboration to improve the book retailing business in a city. If all the bookshops and libraries had been included it would almost certainly have tried to crush new innovations like Amazon. Similarly a collaboration of existing banks to improve financial services would be highly likely to exclude competing peer to peer and other alternative finance tools. So collaborations work best when there aren’t fundamental clashes of interest.
The third challenge is democracy. All jurisdictions will have an existing legitimate authority – a municipal or regional government. Some collaborations are interwoven with democratic power, and defer to elected leaders. Others try to circumvent them. Often community leaders claim a competing legitimacy to that of elected politicians. Since collaborations often bring in unaccountable philanthropic money this can create any number of tensions and problems. Giving private foundations greater power is, for fairly obvious reasons, fraught with problems – often empowering the white and wealthy relative to everyone else.
A final challenge concerns argument. Any kind of partnership encourages people to form a consensus. That can be very healthy. But often the best thinking depends on vigorous, robust argument. A common finding of partnership boards is that they brush difficult issues under the carpet and don’t cultivate the kind of hard argument that is found within the best organisations.
Making collaboration easier
Several recent technology trends could be making large scale coordination easier.
- Open data – allows for more systematic scrutiny of funding decisions and actions. Nesta is part of the 360 degree giving group in the UK which is promoting provision of machine-readable data on all grants, their purposes and locations. Washfunders points in the same direction. This sort of openness is still very far from being the norm in most countries. But it is a significant help to serious coordination because it allows more independent, and rigorous, assessment of whether collaboration is really happening in terms of actions.
- Digital tools for collaboration. Perhaps the most powerful aids to collaboration now come from the digital world. There are now many more online tools to help dispersed groups work together, from simple ones like Google Docs to complex ones like Genius. Some rather older ones still look impressive – such as the Rosabeth Moss Kanter/IBM tool for helping a whole system to navigate change (which wasn’t used much in practice, and was probably ahead of its time). We’ve learned that these online tools are complements, not substitutes, for face to face engagement. But they’ve improved a lot and they can bed in the habits of collaboration.
- Digital tools for self-assessing collaborations. I was influenced in the 2000s by social network analysis which seemed the perfect for mapping the reality of partnership by asking the several hundred people working in a system at city level – eg the system delivering secondary education or cutting domestic violence – who was helpful to them. Some of these methods are described in the Nesta report ‘Transformers’ which I co-authored. These methods remain uniquely well-suited to showing the reality of collaboration; but they are very rarely used, perhaps for that very reason.
Better collaboration could also be helped by other, non-technological, trends. One is the spread of common languages for evidence. The temptation for any programme of this kind is to use self-generated data, which then allows all the partners to pat themselves on the back for the numbers of jobs created or skills learned. But without independent validation, and a common, rigorous framework for thinking about evidence, these rarely stand up to scrutiny. The wider use of common standards should improve the performance of partnerships. A related trend is the spread of cross-cutting leadership development programmes. There are now lots of organisations skilled in training leaders from across sectors to cultivate a shared sense of responsibility. In the UK Common Purpose has been doing this for decades. Quite a few universities now offer tri-sector leadership courses.
One of the biggest problems in Europe has been the failure of elite structures to represent a rapidly changing population. Birmingham in the UK for example is now a majority BME population, young and female – but has tended to be run by white, older men. Cities like Marseilles in France or Malmo or Hamburg. In the US, it’s now clear that social mobility went into reverse some years ago.
Uprising has attempted to address this in the big cities of England – and to create a cadre of leaders well prepared for the kinds of collaboration described in the rest of this paper. It offered a yearlong training course for 18-25 year olds, mainly from minority backgrounds, focused on the skills needed for inclusive leadership. Some of the learning is classic pedagogy – how power structures work, how to communicate. Some is about self-knowledge. But a core of the programme is running a real-life campaign, learning by doing. The model drew on the ideas of US community organising (Alinsky et al), India and Brazil (Freire et al) and attempted a synthesis better suited to the 21st century.
Problems
What can go wrong with collaborations or ‘collective impact’ projects? Apart from lots of talk and not much action, or waning commitment, for example as the leadership changes in key organisations, or a major change in the environment (like a recession), several typical problems often affect collective impact projects:
- Too much focus on what’s easily measured rather than what matters
- Actions to cut bureaucracy that end up increasing bureaucracy as they create new committees and processes while the existing ones continue in parallel
- Creating new, competing forms of multi-functional governance – but without the direct democratic accountability of the municipality.
- Excess technocracy – a tendency to neat powerpoint diagrams, dependence on consultants and insufficient growth of capacity on the ground
Where next?
It has to be right for organisations with power and money to collaborate, to focus on outcomes not activities, and to make themselves accountable for results. But this is a field that needs to become a lot better at learning – with critical case studies; sharper analysis; and more hypotheses to be tested.
It also needs to be more democratic in spirit – too many examples have increased upwards accountability but done very little to grow downwards accountability to beneficiaries (and there is almost no mention of this in many of the recent articles on the subject).
A survey of the recent literature suggested that the field hasn’t moved forward much in the last two decades, and may have even gone backwards. There appears to be little sense of what’s been learned from problems and difficulties as well as successes.
Precisely because collaboration is obviously a good thing it risks being talked about uncritically. But bad collaboration, and cosmetic collaboration, could be worse than no collaboration at all.
There won’t be any neat formulae to guide the future collaborations of governments, funders and civil society. But this is space where we should set our ambitions higher, and be less accepting of prescriptions that haven’t bothered to draw on what’s already known.
What do you think? Share your thoughts in the comments.
