When I was a kid, my parents enrolled me in a program called “Science by Mail” through the Museum of Science in Boston. The Museum would send me a kit. Once I received a box containing balsa wood and glue with instructions to build a bridge that could hold as many pennies as possible.
Fast forward 20 years, and I am pretty much doing the same thing. For the last two years, I have been working at different non-profits and companies, trying to figure out how to use data to maximize effectiveness. While I think relying on data too heavily without taking into account externalities that make us human risks making bad decisions, I do believe that data is a powerful tool to make good decisions.
So, with that in mind, here are a few things I have learned about using data in the context of non-profit, social enterprise, and all other endeavors at the base of the economic pyramid.
1. Choose a Problem-Solving Framework
McKinsey, the management consulting firm, has an approach to problem-solving called MECE. It stands for “mutually-exclusive, completely exhaustive.” The idea is to first come up with a question you want to answer, and then deconstruct all of the elements of that question to its most fundamental elements. Then, once you have those bite-sized questions, you approach answering them methodically. This is particularly useful when you have a big data set and a wide scope of questions you could answer.
When I was trying to figure out how Bridge could expand our reach to as many students as possible, I started breaking the question down to its components. How do get more students in school? You can do it two ways: attract more students, retain the ones you have, or both. I decided to focus on retention, since the data I had could more easily be used to answer that question. I broke off a sub-question – why do students leave? Well, that can be broken off into preventable reasons – poor quality education, parent preference – and non-preventable ones – money, moving upcountry, etc. You can continue breaking these questions down to make your analysis more manageable.
The visual representation of this process is a decision tree, which is typically used when you have a series of binary, fork-in-the-road decisions. There are other ways of looking at problems – scenario analysis, etc. Breaking down complex problems this way is an easy way to make these questions a lot more manageable.
2. Start with what you want to know
Sometimes, you don’t have enough data. Sometimes, you have too much data. Believe it or not, the latter can sometimes cause more problems than the former. A couple months ago, the World Bank and the Kenyan government launched “Open Kenya,” an online database containing every bit of government data that could be digitized. As the first African government to open its data to the public, Kenya was considered to be at the vanguard. And, working at the iHub, I had good fortune of seeing the World Bank’s “Open Data Evangelist,” Tariq Khokar, speak to software developers about how to use this Open Data website for good.
He cautioned the group of mostly researchers and developers about a problem that many people who love data don’t often think about: what to do when you have too much data. You can easily get lost if you try to boil the ocean. One approach is to look at the data, see what you have, and decide which questions you want to answer. Another, more enlightened approach, according to Tariq, is to step back, think about what you want to do, then seek out the data you need.
This is good advice. Data can’t tell you everything. But it can tell you how to optimize a process, think about the most lucrative market, the most cost-effective way of doing things, etc. This is particularly important when you are trying to balance two dimensions of creating products and services for the base of the pyramid market: affordability and quality/utility. There is a point where these two lines cross – the maximum people will pay for a certain value – that can be identified by taking a look at the data.
3. Take It With a Grain of Salt
The current financial crisis has at least some of its roots in financial engineering, also known as computation finance, which tries to “precisely determine the financial risk that certain financial instruments create.” The problem with this goal, of course, is that it is completely impossible. What distinguishes human beings – sentient beings with feelings and emotions – from, say, gravity, is that sometimes we behave irrationally. What you end up with are a bunch of mortgage-backed securities and derivatives with AAA ratings from Moody’s and Standard & Poors that become collectively known as “toxic assets.” Another case of when keeping it real with data goes wrong.
My point is that data can only tell you so much. I happen to rely on it quite a bit in my job, but I know that, to understand the nuance behind the data, I need to speak with people who understand what is happening on the ground. Parents living in the slums and earning $60 a month have a much different idea of what “quality” means than a parent living in a posh suburb of Nairobi. And, even more importantly, parents living in Baba Dogo (a slum in Nairobi) have a different concept of quality than those in Mathare (another slum in Nairobi). Unless you understand what motivates those parents to make education decisions and ask them why they choose to send their students to one school over another, your data will be useless.
These are three things to consider when thinking about how to use data in running your social enterprise or non-profit. Go to work.
Develop Economies’ Music Recommendation