FIP 81: What mobile data can do for social enterprise with Guilherme Lichand

In the final episode in our data series, we speak with Guilherme Lichand of mGOV on the use of the “anti-apps” to establish a two-way communication channel with citizens, program beneficiaries, and customers.

On this episode, you’ll learn:

  • Guilherme explains the reason mGOV works with mobile technologies, but mostly leverages the power of the “anti-apps” – SMS, text messages, and automated voice calls (i.e. interactive voice response (IVR)) – which come pre-installed in every phone and are very powerful at the coverage level. The company evolved from using these tools for surveying to building active communications channels with target populations.
  • He talks about some of the most scalable aspects of their technology has been “nudge bots” which push users to make small changes to everyday activities, such as engaging parents in their children’s education, households toward healthier financial habits, etc. As he explains, it changes the architecture of choice in people’s minds which shifts their habits in a more effective manner.
  • He advises social entrepreneurs to try to allow its beneficiaries / customers to choose the method of communication that is most natural for them (noting connectivity barriers), as the message is actually communication platform neutral.
  • Guilherme also discusses how to “find” people for surveying, which depends on whether you have access to the audience and what your resources are. Some options for more generalized surveying of new audiences include random digit dialing, geo-fencing, heat maps, or public interest advertising.
  • He also shares some tips on mobile-based surveying, including keeping the survey under five minutes (i.e. 15-20 questions), having a “re-attempt” protocol to connect with busy people, providing incentives to response, and giving survey respondents information about the results of the survey.
  • Guilherme provides a series of open source tools that enterprises can use for surveys and data gathering, including RapidPro and
  • In terms of data privacy and protection, mGOV always requires informed consent, which can have a large impact on participation rates. They often try to get informed consent first face to face and provide people with the option to opt-out.

Links to Resources:

Connect with Guilherme:


FIP 80: The case for not collecting impact data – and what to do instead with Loïc Watine of IPA

I talk to Loïc Watine about evaluating impact in your social enterprise. Loïc is with the right-fit evidence unit at Innovations for Poverty Action. He helps organizations consider what else they should measure if they’re not in a position to make the most of a rigorous impact evaluation.

On this episode you’ll learn:

  • For this interview, we’re saying impact is the difference that is being made. So for farmers, the change in their yield or their income.
  • When feasible and when done well, RCTs are the best way to measure impact of an intervention. You can actually tell the impact attributable to the intervention and rules any other effect out, thanks to the use of a control group for comparison.
  • An RCT is a significant undertaking so the business model needs to be mature enough to warrant an RCT, such that any operational kinks need to be ironed out and the business is giving it’s best shot at the desired impact.
  • When an RCT isn’t the right choice, and it is more accessible than people imagine, you could instead focus more on the early outcomes and look at the existing evidence from similar programs elsewhere.
  • The first thing social enterprises can do when wanting to measure their impact is to draw out a theory of change, which is essentially describing what happens between the specific activities the business is doing and the outcomes you’re aiming for (e.g. more jobs, better exam results, etc.)
  • To decide what sort of impact studies businesses should be adopting, you can use your learning-cost ratio, which is about comparing the value of the learning you’re getting and the money you’re spending on the monitoring and evaluation element.
  • There are simple steps social enterprises can take to verify their impact without an RCT. Such as continuously verifying that the customers are actually using the product or service, which is a vital step for the impact to be realised.
  • Use the CART framework to prioritise what data to collect. Is your data Credible? Is the data you’re collecting Actionable and you know how you’ll use the data? Is the cost of collecting the data Responsible, compared to what other business benefits that money can be used for? And can the data be Transportable and inform your programming outside the specific context and time in which you’ve been collecting it?
  • The right-fit evidence unit is an advisory unit of IPA that works with other organisations to help figure out what data they should collect with their limited resources, if it’s the right time to do an impact evaluation, how to maximise their learning-cost ratio.
  • Organisations get tripped up by dedicating too many resources on the final stages of the theory of change, when they could be measuring more meaningful earlier outcomes far cheaper. Another mistake is to collect too much data without a clear reason or ability to analyse it, which is a waste of resources.

Links to further resources:

Connect with Loïc:


FIP 79: Lessons from a data driven social enterprise with Mina Shahid

This is our first in an episode series on data and social enterprises. Today’s guest is Mina Shahid of Numida who shares his thoughts on how to effectively incorporate data into social enterprises.

On this episode you’ll learn:

  • How Mina and Numida started to incorporate customer-provided data to both develop digital tools that their customers themselves can use to improve their financial literacy and serve as the information backbone for Numida’s credit decision making.
  • In terms of steps in building a data driven enterprise, Mina highlights:
    • First, figuring out what type and quality of data an enterprise needs to make its business model work (which may be a trial and error process).
    • Second, the importance of focusing on what channels you are going to use to get that information. Mina notes access can be a huge challenge, as a company may need to create new data sets in order to be successful.
    • Third, the need of having a strong product development team and being very rigorous in the design process. He imports that a lot of app or technology features that Westerns think are intuitive are not actually “easy to use” for first time smart phone users. From that perspective, Numida feels like it is really at the forefront of creating the market for digital tools for African small business users.
  • Building on the last point, Mina also talks about all the information that Numida tracks for its enterprise customers and how to make those reports easy to understand and actionable for its clients. In terms of developing and improving “usability”, every action in the application is tracked – meaning that Numida can see how many steps or how much time it takes for its users to use different features. For example, they shaved off a minute of time from their process to register their new transaction flow. Mina underscores that this laser focus on app “usability” can be key to getting and maintaining customers in a digital-first business.
  • Mina highlights how being a data driven company has in the end helped conserve precious resources by allowing the company to focus on what is most important for its business. This includes quantifying the impact of its products on its customers, further supporting the value proposition of Numida.
  • Mina also touches on privacy – noting that a lot of people in the developing world are not really focused on privacy – and that Numida does not use data mining practices that it finds unethical. They are very upfront with their clients on the type of information they are using and why they are asking for it.
  • He notes that while Numida still has some manual tasks, making it clear that not everything has to be entirely digitalized from day one to be efficient and scalable.
  • Finally, Mina’s advice to other social entrepreneurs that want to leverage data in their business model is to just do it and take the opportunity to be data driven if it is there. He highlights the importance of really thinking about how you want to use data to make decisions in the future and to make that a cultural element in the company from day one.

Links to resources:

Connect with Mina:

FIP 015: Pro-bono Data science with Jake Porway

In this episode, Jake Porway talks us through DataKind, the organization he founded that helps connect data scientists with non-profits. He was previously with the New York Times as a data scientist so is a great person to have on the show to unpack how data can help non-profits make better decisions on the allocation of resources.

Some of the things you’ll learn on this podcast include:

  • A quick overview of data science and how big data, statistical modelling, machine learning, and artificial intelligence all fit together.
  • How DataKind brings together data scientists with social organizations to work on projects together, and how they facilitate developing the scope of the project, the design and the process of working together.
  • How to start right, which means coming up with the question to answer, rather than the data to use.
  • How the Red Cross worked with data scientists to to predict where fires were more likely to happen across a city.
  • How the Gates Foundation worked with data scientists and satellite imagery to help prevent the spread of wheat crop disease affecting subsistence farmers.
  • How more and more Corporates are sharing their data of human interactions to help with social problems, like a financial transactions of refugees or social media activity.
  • The six components that DataKind brings together for a successful project: a smart problem statement, data sets, data scientists, funders, subject matter experts, and the social actor who will use it at the end of the day.
  • How a crisis textline organization connecting teens in crisis with councilors, used data science to predict which texts were the most urgent, so resources could be allocated more effectively and lives saved.
  • How 14,000 data scientists are standing by with DataKind to help tackle social problems around the world and how you can take advantage of this incredible resource.


Connect with Jake