Can mobile phone data answer global development's call?

Community health worker Toumany Tigina enters a village girl's symptoms into a mobile application which facilitates early diagnosis and treatment of malaria. Photo by: John Bernon / USAID

Mobile network operators are gathering a tremendous amount of data every hour of every day. How they might use that data to support society is something they are still trying to figure out.

Many, both at mobile network operators and in development, have recognized that the data from mobile phones can help in tracking critical challenges — from the spread of disease, to migration patterns, to poverty rates — but they have now reached a critical junction where they need to translate that knowledge into action.

“In general we see we cannot leave this potential on the table and do nothing,” said Nicolas de Cordes, vice president of marketing anticipation at the Orange Group, a mobile network operator among the first to examine using data for social good. “We are really exploring now, asking the next question … not is it interesting, but how do we do to it in a safe and trustable way.”

De Cordes and others in the industry are quick to note that they’re early in these explorations. Questions remain, including asking who has access to the data, who owns it, what are the limits of the data and what role do governments, citizens and nongovernmental organizations play. Still, for those in the development community looking at how the treasure trove of data may inform policy decisions or programs — potentially in real time — current efforts may be a gamechanger.

Privacy, security and competition

It’s not easy for telecommunications companies to share cellphone data that tracks location and can be used to trace the spread of malaria, for example, or do this work themselves for a number of reasons, including privacy, security, competition and infrastructure.  

Preserving the privacy of mobile phone users is one of the greatest challenges in using mobile data for development. Cellphone data in particular is extremely sensitive and identifiable, which means that the ethical and privacy issues are of a different magnitude than with census or survey data, said Yves-Alexandre de Montjoye, an assistant professor at the Imperial College London Data Science Institute and a research scientist at the MIT Media Lab.

“The real great thing working with mobile phone operators is that a lot of them are really willing to engage and be part of the discussion and see it as something socially important to them to be part of and do,” he said.”The challenge is really how you do it technically in a way that preserves people’s privacy.”

Often data scientists will manipulate data to remove identifying factors in an effort to protect users’ privacy, but data anonymization or de-identification doesn’t work with metadata from cellphones because it is quite easy to identify users — it only takes four data points, de Montjoye said.

And mobile network operators are also concerned about opening up their data in a way that would allow their competitors to get too much information about what they’re doing or about their customers.

Even if those obstacles are tackled, there are others. Chiefly, getting local governments or development agencies to buy into the use of the data.

This work of using mobile phone data to examine development challenges is so new, de Cordes said, that he has often found that government officials have little understanding of the technology and how it may help. Statisticians are also sometimes resistant to the new methodology.

While early efforts are proving some one off examples of how data might be used, it will also be important, moving forward, to determine when mobile data can be most effective. Even if that is determined and there is a more robust system for analyzing and sharing data, there will still be questions about how to combine data from different sources, said Nuria Oliver, a computer scientist and the scientific director at Telefónica.

Despite those challenges, Oliver is quite hopeful: “I think the progress and the interest is just growing in the past few years and I think we are reaching a very exciting time right now because a lot of different factors are converging,” she said.

The OPAL Project

In an effort to address some of those challenges, a group of interested actors, including Orange, Telefonica, have come together to build a platform that aims to provide a broader, more uniform, more protected way of sharing data around key social or development issues.

The OPAL Project, as the initiative is called, is building a platform with software infrastructure that would give mobile network operators a way to safely allow third parties, such as national statistical offices, cities or NGOs, to use their data while still preserving the privacy of individuals.

Instead of trying to find ways to open up individual company’s data to researchers or government officials, it instead allows companies to run an algorithm and share the analysis — or answer to the question — rather than the raw data itself. Third parties could request certain information and the platform would use algorithms to gather it and share it with them.

With multiple telcos working together, it could also more accurately create a dataset that would be representative of a country or a region. In contrast to only using one company’s data, which could be extrapolated to reflect a broader population by calculating market share, a central system where all the information across mobile network operators could be shared would provide greater accuracy.

De Montjoye estimates that, depending on funding, a first version of the platform can be deployed in the next year and a half. Then the work of developing more algorithms to mine the data can begin in earnest.

“From a privacy perspective, if you really want to be able to use the data in an operational setting for development, you have to go through developing infrastructure,” he said.

Moving to action

So far mobile network operators have worked to prove the potential of mobile phone data. But now, they tell Devex, it’s time to build systems and find ways to use what’s been proven.

Partnerships will play a key role in that progress. Once a system for analyzing and sharing the data exists, mobile network operators will need to work with governments and nongovernmental organizations that have the ability to put the data to action. As mobile network operators and researchers develop new algorithms, local data and social scientists will need to help both inform how they work, interpret the results and cross check them with other data.

“We need to create this ecosystem of actors,” de Cordes said, adding that there is a lot of fragmentation today.

An important part of that ecosystem will be national statistical offices analysis; mobile phone data could complement official national surveys, which are often only done in five or 10 year intervals. But partly due to their lack of familiarity with the new technology those offices, and other development actors, rarely step forward to say what data they most need.

That paradigm will need to shift. Instead of telcos bringing governments or development partners a proposal or data, they should be working with them to respond to their needs, de Cordes said.

Mobile network operators will need to commercialize some of the work they’re doing around data to make it sustainable and profitable. For Orange, it needs to be profitable at some level to last for the long term, he said.

The company is patient, though, and is exploring models where private actors who could afford to pay for data analysis would do so; those funds would be used to offset Orange’s “data philanthropy,” de Cordes said.

“Like with any new emergent area there is usually a research period where new ground is being explored and set up, and once the ideas, the tech algorithms are more developed, then is the opportunity to try turn it into product,” Oliver said.

The potential of mobile data to inform development decisions and include a broader group of individuals in data gathering appears to be immense — but so, too, do the challenges. Reaching a point where trained algorithms can mine data to provide information on health, infrastructure and economics remains a promising venture and a daunting task, one that will require development actors to take a leap and gain the necessary skills to make use of the information.

Join Devex to network with peers, discover talent and forge new partnerships in international development — it’s free. Then sign up for the Devex Impact newsletter to receive cutting-edge news and analysis at the intersection of business and development.

You have 2 free articles left
Log in or sign-up to unlock all of the free news on Devex.

About the author

  • Saldiner adva

    Adva Saldinger

    Adva Saldinger is an Associate Editor at Devex, where she covers the intersection of business and international development, as well as U.S. foreign aid policy. From partnerships to trade and social entrepreneurship to impact investing, Adva explores the role the private sector and private capital play in development. A journalist with more than 10 years of experience, she has worked at several newspapers in the U.S. and lived in both Ghana and South Africa.