According to the World Bank’s Digital Dividends report, fewer than 20 percent of digital government projects are successes. Particularly in developing countries, these numbers are often associated with a number of challenges: limited funding, stretched implementation capacity, and political instability, to name a few. Yet, even in developing countries, despite similar conditions, some projects seem to fare better than others. Why is that?

The projects we have worked with in the global south have followed a similar pattern. While there were successes, many projects have failed. We have learned a few things along the way, that we think relate directly to the success or failure of digital government projects. These are not scientific conclusions, they’re personal impressions based on what we’ve seen and experienced.

A basic function of digital government is the provision of actionable information concerning public services, be they online or offline (e.g. opening hours, documents required for services, and so on). Even more so in developing countries, where most public services are in-person, paper-based, and often involve multiple steps. Yet, fueled by international rankings and benchmarks, governments are often eager to skip stages in their digital journey. This leads them to attempt, and often fail, to provide transactional digital services, before they can even learn how to offer basic information about these services. The first step in effective transformation should be offering information to users in a simple and accessible manner. Done well, that forms a good foundation for the next step: delivering digital services.

Remember that public service delivery follows a power law distribution: a small number of services account for the vast majority of transactions with government. Which these services are will vary according to country, level of government, and models of public service delivery. When the time comes to decide where to start, don’t rely on cookie-cutter lists of services to be digitized. Instead, find out which ones are the most used, and will have the greatest impact. Start with the ones that can be delivered faster, and that are most likely to make users’ lives easier.

The fact that a process exists doesn’t mean it’s a good process. Transformation is an opportunity to radically rethink how things work. We’ve seen examples including, for instance, requiring multiple copies of a single document, or imposing more procedures on women than men to open a business. When there is inefficiency in a service, map the bottlenecks and think about how to streamline the process. Don’t just digitise the bottlenecks, they will keep on being an expensive problem. Resist the temptation to digitise things that should not exist in the first place.

Civic tech projects are those aimed at promoting transparency, citizen participation and accountability. To many, including us, they are one of the most rewarding areas to work on when it comes to digital government. However, many civic tech projects are being sold for the right reasons, but bought for the wrong reasons, like over-the-counter drugs sold in a pharmacy. Some governments are willing to engage in token digital reforms and a thin veneer of openness, even though ultimately, everything remains the same.

RFPs work well when your government is buying roads or bridges, but they’re not so good for procuring for digital goods and services. We suggest writing them in collaboration with multidisciplinary teams. Avoid contract-speak. Write for humans who do not have the time to go over hundreds of pages. Aim for smaller contracts and encourage smaller firms to bid, supporting your domestic digital economy. Good RFPs aren’t 100-page+ documents, but a way of stating what your goals are and visualising your roadmap.

Every digital government person has already been in a room where the meeting was about features specified a year ago, without any user research, for a project that hasn’t even begun yet. Technology advances daily; software that’s 6 months old is software that’s already going stale. Developing countries can’t afford to fail big. Try to sail the ship with the minimum requirements and improve the journey incrementally. Technology projects need breathing time to pass on testing, iterating and improvement. Avoid large, rigid waterfall-style projects: they are hard to halt and change, and end up becoming museum pieces sooner than expected. Reconsider your approach, and look at the problems you can solve while you still have time and budget.

Government technology projects require changes at a policy and institutional level, especially when you start dealing with things like privacy or procurement, etc. Innovation won’t be delivered by an old mindset with old regulation. Involve your stakeholders in the journey, deal with it as a public policy discussion. Make way for disruption in institutions and create new ones to support long lasting change.

Don’t take for granted that everyone has online access, because that’s not the case. Developing countries usually have pretty poor, slow or expensive internet services, so the usage isn’t evenly distributed. Those less likely to be connected are those who are mostly likely to need public services. Design for inclusiveness, and to what extent it’s possible, make services available online, via phone and in person. Also, when thinking of ‘data-driven’ policies, remember that social media should not be your main source for data. Consider the amount of time that people spend online and what they are doing, and the price they pay to stay browsing. This will make you realize that you shouldn’t ask citizens to spend their internet time on governmental bureaucracy: they won’t download your app.

User research is essential for the design and development of good digital services. Yet, in developing countries, this is even more so because the poorer a country, the less likely public services are designed by people who use them. In that case it is common to find, for instance, designers of transport systems who use private drivers and health experts who never received treatment in a public hospital. The result is that services in developing countries are built with an even larger empathy gap, conceived by people whose frame of reference is unable to capture everyday users’ experience. And while we are at it, remember: “stakeholder consultation” is not user research.

Several governments have recently engaged on a service numbers race. The performance of digital services is thus advertised by the number of services offered, sometimes by the thousands. There are some exceptions, but this approach can easily turn into a race to the bottom, with the number of services provided (“output” measures) inversely proportional to their quality and impact on users’ lives (“outcomes” measures). Performance measurement of digital services is more likely to provide meaningful results when focused on the usage of services and their added value. Start with simple measurements, such as service uptake, time savings and cost per transaction, and gradually transition to more customized measurements that are aligned with the goals of each service.

These suggestions are based on our experience from the field. Digital government faces a new wave of development in the future with the arrival of new technologies such as artificial intelligence, but we have to fix the basics first. Developing countries can use digital government to boost development and service delivery, making the most of the opportunity to sail on the wave of the 4th revolution.