1. Not all data is useful – only that which tells a human story.
Agencies often have an overwhelming amount of data at their disposal, but not every piece of information is useful. To cut through the noise, consider what data will help tell the story of the people who will benefit from the public service. This starts with asking:
- How will this service benefit its recipients or society as a whole?
- What metrics should we use to measure success?
- What data do we have or need to identify at-risk or vulnerable individuals?
- Which other government agencies touch the lives of people who need this service?
- Do we, or any of these agencies, already collect the data we need?
- How will we bridge any gaps?
By starting with the understanding of how people can help themselves and where government can usefully intervene, and thinking about the data that would tell this story, agencies can shine a light on accountability across the decisions, investments and impacts in the system. It’s a great way to step back and see how agencies, singularly or collectively, contribute to beneficial change.
Once all agencies understand how the system fits together, steps can be taken to reduce duplication and fill any gaps between supports. This leads to a structure that will help all agencies in the system to learn, improve, innovate, and collaborate. Effectively, data becomes a shared language to help agencies agree on how to tune system performance.
2. The most useful data is often the hardest to get, but perfect is the enemy of good.
Sometimes the information agencies want in an ideal world is hard to get. For example, it’s challenging to extract granular client voice information and collate it at a system level. Often acceptable proxy information is sitting with other agencies who hold important pieces of the data puzzle. Or a well-designed system can draw conclusions from an anonymised view. For example, data about the people and places that interact with a system gives invaluable insight into:
- Cohorts with similar needs and accessibility drivers
- System-level changes to better find and serve
- Connections, overlaps and boundaries within the service system
- Pathways into and through a service system
- Changes in service utilisation over time
This proxy information can be compared with other information sources, such as research and front-line feedback to test the conclusions being drawn from the proxy data. By undertaking this sort of approach, service provision and design can be refined and tested for on-going effectiveness in improving desired outcomes.
3. A system-level view requires a system-level response.
Once governments have an overarching view of interacting issues, the system as a whole, needs to respond.
In a siloed system, an agency may own a ‘problem’ but be unable to move the dial because the solution does not lie under their auspices. Take welfare as an example. In a prevention sense, success could be defined as reducing the number of young people coming on to benefit. This requires young people to have the skills to gain jobs that are available in New Zealand, so much of the solution lies within education and industries.
When agencies share information and use it to design systems to solve problems together, government can make progress in addressing complex issues. Data helps system stewards spot what’s not working, target interventions where they can have a proactive ripple effect through the system and align the collective actions of all players in the system toward the overarching goals – rather than separately undertaking point interventions and reacting to service-level concerns.
4. Focus on the action.
Once agencies understand what’s actually happening, the imperative is to act. No matter how unwelcome the news is. No one wants to hear that an expensive initiative isn’t moving the dial. But programmes should be seen – and talked about – as experiments. The purpose of the data is to provide an understanding of performance of services, individually and collectively. It is then a matter of what to change and what to leave the same. Important questions are:
- Which services are working for whom? And which are not?
- Where performance is not as hoped, is the issue in the design, the implementation, or the timing of the implementation?
- What have we learned that we can use to improve other programmes – or share with other agencies?
- For services that are working, what could further improve them? Should they be rolled out more broadly?
Being clear on what data you have and how you’re going to use it forces decision makers to confront uncomfortable trade-offs in the system. Going back to welfare, we may know that lack of skills is one of the main drivers of young people entering the system. But there may be other candidates – at-risk children, victims of domestic violence or mental health barriers to entering the workforce. In a system with limited resources, who takes priority?
The only way to make this type of tough decision is to demonstrate the evidence and choices behind it. If at-risk children are given priority, the data can tell a compelling story that demonstrates why this decision has been made.
Humanity, social acceptability, and ethics must always sit alongside the use of data to improve public services. The best decision makers see the realities of the people behind the numbers and are transparent about the trade-offs they are willing to make.
Better questions for data-informed public service system design
As agencies consider the data and systems that support public services, it’s important to get collective agreement on the answers to the following questions:
- What data do we need to tell the story of the people we serve?
- If we can’t get the data we want, what proxy information can we use?
- Who will we share our data with to enable a system-level response so people on the front line can make better informed decisions?
- Which cohorts does the community want to take priority wheemen assessing trade-offs?
