20 While digital solutions have many potential benefits, they are not an end in themselves and may not necessarily improve the work of government. Adopting new technology and introducing new digital tools and ways of working are expensive and time consuming, and experience from around the world includes costly, failed projects that have delivered few benefits.40 Before proceeding, government must consider carefully whether a proposed project reflects government priorities, is realistic, and offers value for money.41 Technology is ultimately a tool. The tool’s purpose and use is as important to success as the tool itself. The transition to a digital data ecosystem is important, in part, because it offers government an opportunity to reimagine existing ways of working. Doing the same things as before, but now digitally, is not the hallmark of most successful government digital projects. Instead, governments carry out in-depth systems analysis to reconsider their priorities, reevaluate workflows to better meet those priorities, and reimagine their work, choosing to digitalize as appropriate and deploying the right technical tools (see box D). Systems analysis also provides an opportunity to reconsider staffing numbers and essential capabilities. The starting point for the development of a digital data ecosystem is a focus on how data will be used. Government should work backwards from the priority policy problems they are seeking to solve, asking what specific data would be most helpful to understand these problems. Every piece of collected data should have a purpose, and departments and ministries with overlapping data interests should work together, wherever possible, to ensure that collected data serves their several purposes. By considering the needs of data consumers across government and at each administrative level, higher quality and more useful data will be collected. In the past, the purposes of the Union government defined what data should be collected. As Myanmar continues the twin processes of decentralization and people-centered development, the data needs of township officials should be given greater weight. Data collection can be expensive, so government should be circumspect when considering what data, and how much, to collect. In reforming the data ecosystem, data quality and trust in data should always be the priority. System reforms must always focus on data quality. Data must be timely, complete, accurate, consistent, and understandable (figure 1.2). If data quality is poor, evidence-based policymaking will be an illusion, and decision-making will not improve. For data to be used effectively, all those working within the data ecosystem must trust the data they work with and understand the data’s strengths and weaknesses. This requires a culture in which individuals can speak honestly about the strengths and weaknesses of the data they are presented with, and users can triangulate across different data sources before making decisions. Reforms to the data ecosystem should start small and develop iteratively. Bigger digital reform projects carry greater risks of failure, and big, ambitious initiatives that promise to solve multiple complex problems are less likely to achieve complete success. Instead, government should start small, with modular, integrated solutions, developed iteratively with the participation of stakeholders, that solve parts of the puzzle in a strategic and coordinated way. End-to-end piloting of all system components, both human and technical, before rollout is essential to identify unforeseen challenges and make adjustments before scaling. The early use of prototypes can help stakeholders iteratively develop a technological solution that best meets their needs. The sustainability of reform efforts should be considered from the beginning. Does the government have sufficient funds for the system’s intended scale and anticipated lifetime? Do government officials across the data ecosystem have sufficient motivation, skills, and knowledge to run the new systems without intensive support? Where possible, existing, proven technologies should be used, and the complete life-cycle costs of reforms should be identified. Systems should be designed and implemented so that participants can build needed capacity and benefit from the reforms. SYSTEM REFORMS MUST ALWAYS FOCUS ON DATA QUALITY. DATA MUST BE TIMELY, COMPLETE, ACCURATE, CONSISTENT, AND UNDERSTANDABLE.