The non-profit world is changing a lot, and that means that new data sources show up all the time. Some of them are a flash in the pan, and some wind up being core to the organization’s work.
I had a great conversation with some organizations recently about how to approach and think about new data sources. We talked about a couple of important questions that can guide your organization’s relationship with a new type of data. The idea is that as you experiment with a new kind of data, you can work on figuring out the answers to these questions. When you have answers to these questions, then you’re ready to build things around this new data.
As you figure out the answers to theses questions and most importantly, after you’ve answered them, revisit them regularly and with a lot of different people. That helps to keep everyone on track, and helps everyone say no to rabbit holes that look tempting along the way. Remember that while this data may be something you’re up to your elbows in every day, for other people you’re talking to this is something they don’t think about for days at a time. Remember to frame the conversation for them each time you talk, and to point them to a reference where they can get the information they need.
How mature is our relationship with this data?
Some kinds of data you’ve been working with for a long time – for example, donation data. Everyone is probably pretty clear on the most important things you need to know about each donation and about donations in aggregate. You’re probably always making some adjustments, but your relationship with this data is mature.
Some kinds of data are new and constantly changing and it isn’t entirely clear yet what everyone wants from it. I’d call that relationship experimental rather than mature. Experimental data is exciting and may wind up being really important, but while we’re experimenting, we’ve got to resist the urge to build something big around it, and we’ve got to communicate clearly about how this is different than data that we have mature relationships with. It can be really tempting to have people “integrate” new data with some existing system, but if you do that before things are settled, the results are probably going to be terrible.
If you communicate broadly about the experimental nature of what you’re working on, it will be easier to manage everyone’s expectations and to keep your options open. The CEO won’t be surprised if you say different things about this data each month if they know it is experimental.
What questions do we hope this data can answer for us?
Early on, your experimentation of the data may be to help you figure out what questions you’d like it to ask, but you probably have some ideas. Write down what you do have, and communicate that broadly and regularly. Remember that what’s obvious to you isn’t obvious to everyone in your organization, and that people need to be told things several times.
If you don’t know what those questions are yet, or if that list is changing constantly, it isn’t time to finalize much of anything about this data. Keep things loose and communicate clearly and widely about where you really are with the data.
What actions do we hope to take based on what we learn?
If there’s data you spend time and effort on, it should be supporting actions you take that can make a difference for your organization. With new data, you may be figuring out if there are actions you can take, but if you figure out that there aren’t, it is time to stop spending time and effort on that data.
As you work with a new type of data, be looking to see what actions the information could support. Until you have some clarity on what these actions are, it will be impossible to build the tools you’ll need for the longer term.
Some easier questions
These questions are easier but just as important – if you start building things or integrating this new data without answering these questions, you’ll be unhappy with the results.
- A list of stakeholders who will use the data
- A list of tasks each stakeholders will use the data for
- A list of systems that generate the new data
- A list of systems that store data related to the new data, but that don’t generate it
- What level of detail does each task require?
- How does each system understand a constituent?
- Where does identity reconciliation fit in?
- How often do we need this new data to be refreshed for it to be useful?
Moving Forward
Being purposeful about an experimental phase with new data can feel like slowing down, but you’ll save a lot of time and effort in the long run if you do!
Artisanal Data 