Consider one resident trying to register the birth of a child. To finish, they need a hospital record, a municipal office, a national identity number, and a bank account for a benefit. Each agency has its own portal. None of the portals talk to each other. The resident becomes the integration layer — carrying paper between systems that were each, on their own terms, “digitalised.” This is the most common way digital government fails, and it rarely has anything to do with bad code.
Where projects actually break
Failure usually appears at the counter, not in the repository. A portal can be beautifully built and still defeat the person using it, because the underlying process was never redesigned. The World Bank’s guidance on digital government makes a version of this point: foundational systems deliver value only when they are paired with legal safeguards and institutional capability. A platform without the authority or the process behind it is a brochure.
The service journey, not the website
The unit that matters is not the page but the journey: every step, every document, every hand-off a person must perform. Mapping that journey is unglamorous and often reveals that three of its steps exist only because two agencies use different definitions of the same citizen. Fixing those definitions does more for the user than any interface refresh.
This is why some of the most effective digital-government work looks like plumbing. Shared registers, common identifiers and standard data formats are invisible to the public, but they are what allow a service to be completed in one place.
Legal identity and institutional capability
Behind the plumbing sit two harder requirements. The first is legal identity: a person must be able to prove who they are in a way the state accepts. The second is capability — staff who understand both the law and the system, and institutions empowered to change processes rather than only software. Where either is missing, projects stall after launch and quietly revert to paper.
Several countries offer useful reference points rather than templates. The United Kingdom reorganised its public-facing content around user needs with GOV.UK, treating plain language and a single entry point as core infrastructure. Estonia built shared data-exchange so that agencies could query records instead of asking citizens to resubmit them. Rwanda’s Irembo platform consolidated many government services into one channel for a population with limited fixed-line access. Singapore’s Singpass turned national digital identity into a login used across both public and private services. India’s approach bundled identity, payments and data exchange so that benefits could be delivered at scale. None of these is flawless, and each reflects its own politics and history — but each treated the journey, not the portal, as the product.
Why the counter still matters
Digital delivery does not remove the need for human help; it relocates it. People who are elderly, disabled, newly arrived or simply unfamiliar with the system will still need assistance. A programme that closes walk-in offices without providing an equivalent route to help has not become more efficient — it has shifted the cost onto the people least able to pay it. Good designs keep an assisted channel and treat it as part of the service, not a failure of adoption.
Questions worth asking
When assessing a digital-government project, the useful questions are concrete. Can the task be completed end to end, or does it still require a visit? Whose definitions had to change to make that possible? Is there a fallback for people who cannot use the digital route? Who owns the process after launch, and can they change it? A project that can answer these is more likely to survive contact with real residents.
Paper persists because it is a workaround for a broken hand-off, and because it carries authority a screenshot does not. A stamped document settles disputes in ways a database query often cannot, especially where officials are personally liable for decisions. Digitalisation that ignores this produces parallel systems: a portal for the easy cases and a filing cabinet for the hard ones. The fix is not to abolish paper by decree but to give digital records equivalent legal standing, so that a digital verification is as authoritative as a stamp.
Data protection is part of service design
Once agencies share records, the question of who may see what becomes a service-design decision, not an afterthought. A system that lets a clerk in one department browse another department’s records creates risks no firewall will fix. Purpose limitation, role-based access and audit logs are the unglamorous controls that make joined-up services safe to run. Countries that got this wrong have had to retrofit safeguards under public pressure — a slower and more expensive path than building them in.
The measure that matters
Success is not the number of services listed on a portal. It is the share of common tasks that can be completed without a repeat visit, and the time and cost a resident saves doing so. Those figures are harder to publish than a launch announcement, which is precisely why they are worth demanding.
This report draws on the World Bank’s digital government and GovTech guidance.
Image: Contributor(s): Queensland Newspapers Pty Ltd · Public domain · via Wikimedia Commons.