Designing with data — 4 principles and a cheat sheet

You might have heard the analogy that data is the new oil, a resource that can be exploited, extracted, and used to produce limitless wealth, with nothing to be heard of its consequences. Of course, this analogy is incorrect and a reduction of all that data can be. Rather, data is less of a resource to be extracted, and more of a garden, requiring careful attention and revisiting. Tending to data is similar to caring for plants: one must choose what one wants to grow, must create conditions where those plants can grow and thrive, returning often to clear weeds and debris, and after the harvest, one must continue to care for the land and ensure it is not depleted for any future use.

At UNHCR’s Innovation Service, we try to carefully cultivate a garden that anyone can walk through; we try to anticipate the paths through which you can walk easily and enjoy the fruits of collective labour. In practice, this means that we support our colleagues to intentionally incorporate data in their work. We want each and every person in the organization to understand how data might unlock innovative solutions and approaches, as well as how to use it in a safe, ethical and principled way.

We want to help people understand what they are harvesting, because often people see data as a confounding and inexplicable source of stress, rather than something that can illuminate our work. At UNHCR, data has helped us improve services in refugee camps, anticipate when more resources may be needed in an operation or even improve provision of urgent humanitarian services, via sensing water tank levels remotely. Data will always be generated and collected by people, and it will be people who determine whether or nor that data can be useful, appropriate or principled.

At UNHCR, the Innovation Service undertakes design processes with a focus on human rights and ethics to ensure the link between people and data is understood, recognized, and respected. All data is measured around humans and our interests, and if we lose sight of that fact, we risk making data inaccessible and removed from the contexts in which we live and work.

And context-specific data is especially important, as it allows us to understand the needs and ethical considerations behind data projects focused on different areas and challenges. For example, seasons are different depending on which part of the world you are in. If we were to model data based on the assumptions of incorrect seasonal patterns, we may produce conclusions that don’t make sense and cannot be used to create a realistic and representative dataset and model for the climate in that part of the world.

Most people today want to use data (or at least say that they do), and that, in general, is a positive shift in how we work. However, the Innovation Service has been in the process of collecting and structuring it for so long, that people incorrectly assume that all of it has to be used, and that in using it, they will unlock never before seen secrets to productivity, foresight and possibility. This is especially true with regards to AI, which people often mistakenly think can automatically solve all challenges. Our organizational culture sometimes lacks the ability to be intentional with how we are using this data, and ask critical questions such as for what, and for whom.

In our experience, many projects that seek to use data as the backbone are never finished, and the ones that do cross a finish line very rarely actually give something of value back to the team. Why is it often this way with data projects? It’s because people are reluctant to slow down, ask questions, and interrogate their own assumptions about what data can and should or should not do. And yes, innovation can be slow, valuable and nurtured beyond the move-fast-and-break-things narrative. Data, especially in relation to innovation, should be wrapped up around creative processes and informed decision-making to create value.

When the Innovation Service works with UNHCR’s operations around the world, we often find ourselves coming back to certain ideas that help us to refocus and centre our work on people, refugees, and ethics before any product is developed. These ideas are central for ensuring data can be utilised for social good and grounded in the values that underpin our work within the Innovation Service. Shifting our understanding of why data is of value, and how it can create value, will open up new understandings for how it can be tested and experimented within our work. Along the journey, and in tending to our many data gardens, we’ve learned a few things on how to build these principles into our work and collaborate meaningfully, and intentionally.

Here are four principles to lean into if you are thinking about exploring data innovation solutions or simply laying the groundwork for collaboration around technology:

1. Principle One: Always start with a question. You need to be asking questions or else the data isn’t aligned with a deliberate purpose. The questions can be as simple as “what do I specifically want to achieve in this project?” or can be as varied as inquiring about how data was collected and what its limitations might be. Asking questions and instilling curiosity should be the centre of any data project, as it steers the project and creates conditions for sustainability after its completion, while having everyone working towards the same goal. In our work, data holds no inherent value, but information does and it’s worth it to understand the flow of the information and the use case it can inform.
2. Principle Two: Data is your compass, not your answer. Data is a journey of exploration and inquiry. Look at all the possibilities: You can ask for more time and should probably collaborate with more people. You can plan exercises to bring everyone on board, to define the scope, the research questions, to test different tech and non-tech solutions. At the Innovation Service, we launched UNHCR’s Big Data and AI Working Group to do just this. We wanted UNHCR colleagues to feel comfortable using data to guide exploration and understanding in their work, rather than expecting it to solve every problem for them. The group is also crucial for building a cohesive community that shares knowledge and understanding about data, important principles, inspirations and warnings from other data projects. We hold public webinars on important topics, share updates on what is happening in the UN system, and allow people to share how they use data in their own work. The working group sets out to explore the topics around data and let people know how to use their data compass.
3. Principle Three: Use the past to inform the future. Research what has already been done before you launch into grand inquiries and questions. Who else has thought about this before you? You’ll find that many people have either done or attempted to do something similar; they may even have the data you need for your project or perhaps the software licence you need. It’s important to learn from those who came before you so you can avoid duplicative efforts and have a stronger sense of the initiative’s possible configurations. Take the time to socialise, crowdsource, and use this as an excuse to break silos and to learn about the successes and failures of others.
4. Principle Four: Check back on your data. The race isn’t over once you’ve crossed the finish line. Be sure to check back on the users, and the data, and make sure the project is achieving what it needs to. Be humble and return to collaboration and exploration if it isn’t (or even if a team has ideas of how to do it better)! It’s important to check in on how data was used, explore the ethics and make sure the project is inclusive and transparent, now and in the future. This might well require planning for resources to be dedicated to the project.

We’ve also found that there are common phrases and language used when data and its project potential are discussed, and we want to make it easier to understand what those phrases really mean and how you can respond to them. Think of the following phrases as a cheatsheet for giving you space to work on a data project more deliberately and expansively in your day-to-day. Understanding and responding to these phrases can help you to manage expectations and communicate related challenges:

• The data is scattered. There are often challenges of identifying and combining relevant datasets, as not all of it was collected with the intention to specifically answer and address the things we need to be answered. Here is the time to describe the data, its limitations and biases.
• I am working on homogenising the data. We don’t live or work in silos. The datasets could have been collected by different actors for different purposes, and don’t necessarily speak to each other. The relations between datasets are not always obvious, as the sources are unlimited and the appropriate scheme (aka blueprint) needs to be shaped, to bring forward the connection between the datasets, and this is the chance to highlight this need.
• Let’s redefine the research scope. Sometimes we tend to overestimate what we can answer, because let’s face it, data cannot answer everything. This can and will happen, and it’s not the end of the journey. The best path is to focus on which questions the data can actually answer. It’s important to iterate and learn as you go, constantly reframing your inquiry as you move through the process.
• I need a licence for this type of analysis. None of us is born with programming skills and unfortunately, trial licences (aka the way one can temporarily access different software or platforms) are not a permanent solution. It’s important to not reinvent the wheel as there are many technological wheels out there. Explore what type of licences are available for you to experiment with your data and analysis — see how they can provide an important approach to your data toolbox and offer complementary ways of supporting the project.
• The complexity of the analysis can vary. And don’t worry, you couldn’t have anticipated this from the beginning of the project. If you are having mixed types of data, and they all require different techniques, there are a lot of decisions and iterations to make before you will get the answer to your question. Successful analysis requires building a pipeline or path to answering your research question.

The most important question to ask before, during and after a data project is why. Why do we think this? Why do we want to find this answer? Why do we think showing the data in this way is the most useful? Why is this approach better than any other? Asking questions starts the process of challenging our assumptions and helps the project exhaust all alternatives before progressing for the sake of using data.

We cannot think of data as a stand-alone product, without acknowledging its relationship to people and context. We must constantly explore our questions and goals, and assume that no solution is perfect or permanent. At the Innovation Service, we have been lucky enough to see the creative ways data can be leveraged in UNHCR’s work, but have been even luckier to be part of the creative process of finding principled approaches that happen to involve data. If we continue working on making data more accessible, we all might be a little wiser about when and how we can gauge the potential of data in our work.