Tuesday, April 16, 2013

Where is the data? Analyzing customer footprints for better product design

by guest blogger Jacobo Menajovsky, Senior Data Analyst – Grameen Foundation / @jj_menajovsky

These are my daughter’s old shoes.

We just took them out of the closet to pass them on to her younger brother who’s recently started walking, but when I took a closer look at them, I wondered if their best days weren’t behind them. Call me crazy, but I immediately started decoding all the signs and indicators of their usage. Yes, to me, data is everywhere.

We are constantly gathering, interpreting and acting on data. Think about it. Every time you walk into a new situation, your “decision support system” starts to process past data to help you adjust to the new experience. Your brain is actually modeling those signs and symbols (data), building connections and classifying them into categories.

What if you wanted to understand how these shoes were used? Do you think you could reconstruct the past simply by looking at them? There are lots of signs and indicators: a broken ankle wrap, a heavily-rubbed toe cap, and many holes.

Now let’s move from data gathering to data modeling. When we put all this data to work we can build a great profile of how the shoes have been used. It looks like they went through a lot of kicking and dragging, and plenty of crawling.  If you look at the soles, though you’ll see that they’re unworn. So, it seems the upper parts of the shoes were used more than their bottoms. This observation might even give you a few ideas about how to improve shoes like these and make them more durable. This is exactly what we call improving your product using customer footprints. In this case, the footprints are literal!

As a data scientist and microfinance practitioner, I am always searching for signals and indicators that show how poor people are using products and services. I believe the best way to understand their behavior is by analyzing their footprints. This data can come to me in various formats (e.g. digitally or on paper) and platforms.  I often have to put in a lot of work before I begin analysis, but if done correctly, it gives me a lot in return.

In an ideal world, records would have unique customer IDs and information about the products that each uses as well as past transactions. If you are really lucky you may also have some socio-demographic information like age, gender, rural/urban indicator, branch or location, household composition, family size, and poverty level.

The more data the better.   Your data set can help you answer some key business questions: What’s the penetration of product A at different locations? Is this affected by poverty level or household composition? What about understanding our customers’ lifecycle? Do we see differences in outstanding balances at different customers’ tenures? Our recent study on implementing data analytics provides an exhaustive list of business questions and analytic approaches.

At Grameen Foundation we are working towards helping pro-poor organizations crunch their numbers, understand their customers better, and make more informed decisions. We are also refining our data collection process, using the right set of mobile data collections tools and state of the art analytics to better understand the challenges and needs of the poorest. After all, it’s only by gaining better insights that we will contribute to the development of more tailored products and services – from baby shoes to microfinance products– and that’s essential if we want to improve the lives of the most vulnerable people in our planet.


  1. What a great analysis of using data. I love the beautiful analogy of the tops of the shoes being used more than the bottom as a way to indicate what part of the shoe can be strengthened in the future because that is a perfect description of product development in the microfinance and microsavings industries. Sometimes you think a product needs to have its strength in one place and it turns out you were completely wrong. With data analysis such products can be improves. This blog entry is eloquent and gives excellent insight into data collection and use.