There has been much talk in recent years about the problem of pilotitis in public innovation, with money wasted on innovations that don’t go anywhere. Back in 2012, Uganda placed a moratorium on mobile health pilots due to failures of integration. As recently as last year, a report on innovation in the UK’s National Health Service noted “no clear route from pilot to BAU [business-as-usual]”. This systemic failure is a direct downstream consequence of a failure of innovation governance.
A foundational misstep of public innovation was the indiscriminate adoption of the private sector R&D framing, with the idea of developing and piloting new products and services under controlled conditions before scaling out. This approach works for simple solutions that are easily substitutable with minimal costs or disruption, but it fails when confronted with many of the complex realities of public management. R&D innovation strips away context, but in complex environments, context is everything.
Context-sensitive innovation
The inspiration for this essay is VillageReach, a healthcare NGO I had the opportunity to engage with last year while conducting research on humanitarian supply chains. VillageReach recognise the value of context-sensitive innovation. Their entire approach is predicated on building local capabilities, working closely with communities as end users and target groups, and governments as a critical partner for long-term sustainability.
In Liberia, VillageReach and Last Mile Health partnered with the Ministry of Health and Social Welfare (MoHSW) around a shared vision to digitise medical supply chains. The team started out by recognising the importance of community health workers (CHWs), who would be absolutely critical to long-term embedding of any new tools. That meant engaging deeply to understand what their roles entailed and the challenges they faced in their day-to-day routines.
By taking a step-by-step approach, the VillageReach team were able to ensure that the CHWs were brought along the journey as co-developers of a new digital platform, that the level of technology was appropriate for the context, and that the end result was no more complicated than the needs of the present moment and environment demanded. In time, 1,428 CHWs and supervisors were using the new system, serving over 200,000 households.
Context-sensitive innovation unfolds one step at a time, starting with existing in situ capabilities and gradually probing what Stuart Kauffman calls the adjacent possible—all the ways in which the present might be reinvented through the novel repurposing and recombination of technologies, institutions, and practices. Moving in this direction requires three interconnected and overlapping shifts in how innovation is governed and managed.
Shift 1: Fund direction, not designs
The Cynefin framework for navigating complexity distinguishes between governing constraints that determine outputs, and enabling constraints that channel behaviour in a particular direction while allowing for emergence. The difference between what VillageReach gets right, and what often goes wrong, is in large part due to the ways in which funding models establish these different kinds of constraint.
VillageReach emphasise radical collaboration with government, built on funding models that establish flexible enabling constraints. This means identifying opportunities where the government is willing to act as “payer at scale” for new infrastructure, and support the co-development of new capabilities with flexible funding. As Wayan Vota advocated in a recent post for ICTworks, VillageReach design for government funding, not for restricted donor grants.
By contrast, traditional R&D funding tends to establish governing constraints. Because funding is often competitive and provided on the basis of up-front designs, innovators are incentivised to anchor proposals to the technological frontier, and these designs are then locked in for delivery. R&D funding based on up-front designs then constrains onward development, allowing only for “nearby” pivots. The space for open-ended exploration is reduced.
In complex environments, context-sensitive funding allows organisations to retain strategic optionality while channeling efforts towards lasting outcomes. It sets the direction and boundary conditions without mandating a fixed path. This means flexible grants based on alignment of values, core strengths, and strategic intent, ideally combined with advance market commitments from domestic sources that guarantee long-term financing for newly-developed capabilities.
Shift 2: Manage for emergence
In Lagos, Nigeria, a VillageReach project to extend the user reach of a telehealth platform found that trust was a key issue. This prompted a major pivot to develop a new service, working with kiosk agents in markets and pharmacies as trusted intermediaries to provide health checks and connect patients to doctors using the platform. This ability to learn, draw on existing capabilities, and make strategic pivots, is critical to context-sensitive innovation.
R&D innovation often relies on up-front research to design a solution before getting iterative feedback during development. It might result in an acceptable output, but the process treats feedback as a mechanism for tactical refinement rather than strategic discovery. Teams test hypotheses and iterate designs, but the final form remains locked in. Agile management is trapped inside a cage of linear product development.
In their article “Against Design”, Caryn Devins, Roger Koppl, Stuart Kauffman and Teppo Felin argue that in complex environments it is impossible to design an optimal solution to a given problem up-front because there are an unknowable number of variables, making future contingencies impossible to predict. Instead of focusing on up-front research and design, they argue for adaptive systems that allow capabilities to evolve organically.
The idea is to find what works for the specific place and time, not to pursue a universal solution. Enabling constraints set the direction and open-ended exploration allows for the emergence of new capabilities. In Lagos, the team discovered that trust, not access, was the real barrier to expanding the user base of the telehealth platform. Therefore, getting more users on the platform was not a viable or appropriate target. The next step was to explore an in-person service.
Shift 3: Co-evolve capabilities
Pilot projects often succeed precisely because they ignore context, intentionally bypassing the institutional change required for the solution to eventually become embedded. Instead, they operate under controlled conditions with highly-motivated teams, extra resources, and shielding from bureaucracy. When attempts are made to integrate these solutions into business-as-usual, they meet the reality of legacy systems, varying human behaviours, and local politics and culture.
Successful pilot solutions rapidly break down at scale. The complexity circumvented by the pilot still exists, and when the mostly pre-specified design is finalised, the team finds they are stuck because local skills and institutional infrastructure haven’t co-evolved with the solution. With just a dash of hyperbole, it’s like spending all available capital on engineering a car in a place where there are no roads, no petrol stations, no car mechanics, and no one can drive.
R&D innovation incentivises working backwards from an abstract global frontier, creating a disconnect with local context. The result is coordination failure as the now-apparent requirements for investment in adaptation and infrastructure outweigh the appetite of funders. Overly optimistic expectations meet complex realities, and energy quickly dissipates. A bicycle might’ve been a better fit for our metaphorical mobility problem.
By comparison, context-sensitive innovation works forward from what exists locally. In Liberia, the VillageReach project began with co-exchange of knowledge around the work of CHWs and the importance of supply chain management. It proceeded by combining digital literacy training with co-development of a digital tool, gradually introducing new features so that CHW practices could co-evolve with the platform. At the same time, institutional structures evolved to integrate local data delivery into national decision-making platforms.
Escaping the trap of pilotitis
Traditional R&D focuses on “solutions” as standalone outputs that can be developed in one context and deployed in another with a bit of training and a few adjustments. But complex systems are relational. Context-sensitive innovation recognises that “solutions” are ephemeral. The core value is not found in a transferable “unit” of innovation, but in nurturing a co-evolving repertoire of technologies, practices, and institutions.
Pilotitis is a predictable result of R&D frameworks misapplied to complex social systems that require continuous adaptation. Escaping this trap requires shifting from a mindset of product/service development and scaling to one of managing for emergence and the co-evolution of capabilities, rooted in respect for the unknowable variables of local context. The aim is not to deliver a singular solution, but to leave a new set of capabilities to build on.
The persistence of pilotitis, however, suggests that the trap is hard to escape. R&D innovation holds the imminent promise of a global solution, making it look cheap up-front. As leaders look for attention-grabbing ideas and optimism bias takes over, pilot projects seem like a relatively inexpensive way to demonstrate learning and progress—even if much of it proves illusory. The hidden costs of pilotitis are fatigue, cynicism, and years of wasted momentum.
By contrast, context-sensitive innovation is locally-led and participatory by nature, requiring significant up-front investment in relationship building and early adaptation, and a relinquishing of top-down control. But this up-front investment secures a pathway to lasting impact, avoiding the development of over-specified solutions and the trap of pilotitis. The “slower and more expensive path” is faster, cheaper, and more equitable in the long run.
Sources
Aggrey, A. (2023) ‘A New Digital Solution to Support Liberia’s Community Health Workers’, VillageReach, 5 May. Available at: https://www.villagereach.org/2023/05/05/a-new-digital-solution-to-support-liberias-community-health-workers/.
Chamberlain, S. (2012) ‘Pilot-itis: What’s the cure?’, BBC Media Action, 20 June. Available at: https://www.bbc.co.uk/mediaaction/stories/pilotitis-whats-the-cure.
Devins, C. et al. (2015) ‘Against Design’, Arizona State Law Journal, 47(3). Available at: https://doi.org/10.2139/ssrn.2570030.
Greenberg, R. and Bertsch, B. (eds) (2022) Cynefin: Weaving Sense-Making into the Fabric of Our World. 2nd edition. The Cynefin Company. Available at: https://thecynefin.co/library/cynefin-weaving-sense-making-into-the-fabric-of-our-world/.
UCLPartners (2025) State of the Nation: Insights from UK Health Innovators. London: UCLPartners. Available at: https://uclpartners.com/landmark-report-reveals-hidden-barriers-to-nhs-innovation/.
VillageReach (2019) Goals and Strategies 2020-2023. Seattle, WA: VillageReach. Available at: https://www.villagereach.org/wp-content/uploads/2020/06/VillageReach_Goals-Strategies_FY20_FY23.pdf.
VillageReach (2024) Beyond the Health Facility: Strengthening Community Supply Chains. VillageReach. Available at: https://www.villagereach.org/wp-content/uploads/2024/10/VR_SupplyChainIntegrators_Technical-Brief_Liberia-Strengthening-Community-SI-2.pdf.
VillageReach (2026) 2025 Annual Impact Report. Seattle, WA: VillageReach. Available at: https://www.villagereach.org/wp-content/uploads/2026/03/VR_2025AnnualReport_EN.pdf.
Vota, W. (2026) ‘Please Stop Designing Humanitarian Solutions for Donor Funding’, ICTworks, 9 July. Available at: https://www.ictworks.org/designing-humanitarian-solutions-for-donor-funding/.