Digital Public Infrastructure Adoption is a Governance Problem

Over the last 18 months, Caribou and Co-Develop have spent a lot of time talking with governments, civil society organizations, and others working on DPI. We’ve seen strong technical systems struggle to gain traction, and we’ve also seen how early decisions about participation can shape what happens later. Those experiences have led us to a simple conclusion: many of the challenges around DPI adoption are challenges of governance, not technology.

Digital public infrastructure (DPI) is spreading rapidly, but adoption is not keeping pace with deployment. Building the infrastructure is only part of the work. Governments also need to understand whether people will trust these systems, use them, and see them as worth adopting. That brings governance to the center of the discussion: who gets to have a meaningful say before the technology is fixed?

Adoption is often treated as a downstream problem: build the system, then work on uptake through communications, user-experience improvements, or outreach, but this instinct has the sequence backwards.

DPI delivers on its promise when people actually use it in their daily lives. Right now, digital identity systems launch, but few users pick up their cards. Data exchange systems go live, but hard copies of documents are still expected, still printed, still carried from counter to counter. The failure lives in the space between deployment and adoption in daily life, well past the procurement process and specific technology stack. 

Many initiatives that have taken this approach have struggled:

What stands out from these examples is how often adoption problems emerge from decisions made well before launch. The technical system may be functioning as designed. But the people expected to use it were not meaningfully involved in deciding what that system should look like. This is a governance gap—specifically the lack of multistakeholder governance—rather than a simple technology failure.

When design choices are made without input from the people most affected, adoption rarely follows, no matter how well the system works or how much money went into building it. Inclusion has often been treated as something to check after the fact, yet it should be a condition for the launch to succeed at all.

Building trust faces a similar challenge. Governments often treat trust as a reputational problem. Digital service delivery inherits distrust it did not create—of the state, of data collection, of systems that concentrate information about citizens in one place—and the reflex has been to respond with better messaging. That response gets the diagnosis wrong. Trust in DPI gets designed in at the point where decisions are made about who has a seat at the table and what authority that seat actually carries, or trust never fully arrives.

Brazil’s Pix Forum shows what trust built as a design output looks like. When Brazil’s Central Bank found the market unwilling to deliver interoperable payments on its own, it convened banks, fintechs, civil society organizations, and small business associations to shape what the plan’s goals, before the plan was fixed: zero transaction costs, parity of access between large and small banks, a broader financial-inclusion mandate. These goals had legitimacy because the organizations affected by the system had helped shape them.

Civil society participants in the Forum still say they want clearer, more structured pathways between their input and the decisions that follow, a concern worth taking seriously rather than treating Pix as a solved case. But the underlying design principle is that legitimacy comes from the process itself, from decisions people had a hand in shaping.

No government does any of this DPI engagement work alone. The requirements of population-scale adoption (distribution networks, delivery capacity, credibility with communities that have good reason not to trust the state) are not found in a ministry alone, however capable its team.

Ethiopia’s Fayda program addressed this challenge early. Regional experience made the risk concrete: biometric refugee registration elsewhere, done without adequate community engagement, had left tens of thousands of people effectively stateless.In our work around Fayda, we’ve seen how this history has shaped the program’s approach to engagement. Ethiopia’s National ID Program engaged UNHCR and the Refugee and Returnee Service before the system’s architecture was fixed, then built a formal interoperability gateway and data-sharing agreement that gave refugees the option to participate on their own terms, rather than automatically extending the national database to them.

Fayda, South Africa’s MyMzansi digital identity platform, Ukraine’s Diia app and Prozorro procurement system, Brazil’s Pix: different governments, different politics, different starting points, and the same DPI engagement pattern each time. The collaboration structures that worked were designed early, given a mandate, and properly resourced. Those that failed were consulted into existence after the architecture was already fixed, which is another way of saying they were never really designed at all.

One of the things that has struck me over the past 18 months is how widely this view of governance is shared across the DPI community. The core practical question now is how a government actually builds it. The constraints are concrete: a two-year political cycle, no dedicated convening budget, ministries with no mandate to share decision-making, and little institutional appetite to start. Which partners get formalized, at what stage of the process, and with how much genuine authority, is one part of effective multistakeholder governance. What it costs a government’s own standing to share power, rather than perform consultation with people who already know the difference, is another.

To answer the question of how a government builds in multistakeholder governance, Co-Develop and Caribou developed a practitioner’s guide that outlines the set of design choices that determine whether inclusion actually happens: who convenes, when, with what mandate, and what has to be true of that arrangement for it to survive contact with a change of minister. 

Having been involved with so many consultations and discussions over the last 18 months, we know that the governments closest to getting DPI right are the ones who have worked out, in practice, what building multistakeholder governance actually takes.
—

Read the Practitioner’s Guide and the Compendium of Evidence in full or watch the on-demand webinar panel with government, multilateral, and civil society leaders from the Fayda, MyMzansi, and IDB programs, moderated by Krisstina Rao (Co-Develop) and Emrys Schoemaker (Caribou). Governments and partners ready to move from principle to practice can also get in touch directly about implementation support.