Interview with Linus Bengtsson – Leveraging telco data in developing countries
Data from telecom operators is highly sensitive, but also extremely valuable. Currently a US/Germany-based startup called zeotap tries to capitalize on that, aiming to build a “global telecom data analytics platform, which helps different industries to be more innovative and efficient”. Mobility and transport, internet of things, financial services and mobile advertising are among the industries, where the startup expects great potential. However, beyond commercial interests, the utilization of telco data is especially valuable in situations, where it could actually help to save lives. We had the unique opportunity to talk with Dr. Linus Bengtsson, one of the pioneers in the field of utilization of telco operator data. His work is the enabler for a better coordination of aid and emergency resources in times of natural disasters and for improving targeting of scarce resources in development.
When Linus Bengtsson started his research in Haiti in the aftermath of the 2010 earthquake, the term big data was not commonly known. “It was at a time when the telecom operators hadn’t considered how their data could help”, Mr Bengtsson explains. The Haiti earthquake lead to casualties of tens of thousands, severe human tragedies and displacements – and a large-scale cholera outbreak. Bengtsson and his team got unique access to data from Haiti’s largest mobile phone company, Digicel, which allowed for detailed tracking of population movements. Building upon data of more than 1.9 million SIM cards, the research team could establish valid proof of a close-to-real time tracking mechanism for population movements in response to natural disaster, which provided insights for operational response tactics for responding agencies. What’s even more significant though, Bengtsson and his team continued to fine-tune their model and merged it with insights from epidemiology. As a result, they were able to show that mobile phone data patterns allow for enhanced prediction of epidemic spread. Mr Bengtsson, who holds a PhD from the renowned Karolinska Institute is the Co-Founder and Executive Director of Flowminder, a non-profit with the mission to leverage the use of data to improve public health in low- and middle income countries.
With the massive potential of telco data in various spheres of application, caring about unlocking its value in developing countries becomes societal duty. These are the two main issues, which need to be tackled on a path to fully utilize telco data for international development:
1. Access to data
“After the Haiti earthquake the CEO at the time, Maarten Boute, realized what we could achieve together. It was the first time it ever happened and the process was fairly quick.”, Linus Bengtsson reflects. Usually it takes long and tough negotiations about the specifics of access to cell phone data. “As we have worked in various countries and with multiple operators over the years, we have now build up reputation and credibility, which makes it easier for us in negotiations”, explains Bengtsson. Furthermore, Flowminder is recognized globally as the go-to organization for data-driven work, being supported by prestigious organizations, such as the Gates Foundation or the Rockefeller Foundation. Generally, researchers from many fields might express an interest in telco data, but there is no standardized process as how to apply for access. Startups like zeotap try to capitalize on this aspect and make the data accessible for various parties, while themselves functioning as the clearing house. This however, remains limited to commercial analyses and cannot be expanded to other purposes, like public health. Nevertheless, it seems feasible to provide non-profit organizations with preferred access to the data. To streamline these processes, however, a globally recognized (and non-profit) clearing house would need to be established. “Many have tried, but failed – with given regulation it might not be feasible”, says Bengtsson. Until then, operators will be defining who will get access to the data.
2. Operationalization of data
While telco data might be used for all kinds of insightful analysis, e.g. to outline certain sociological patterns and document reactions to drone strikes in Yemen. But the real value of data is only unleashed, if clear operational implications exist. “We always work closely with UN and other reputable government agencies, because the operational aspect is key for us”, underlines Bengtsson. Nevertheless, this remains one of the most challenging parts of data science. Especially in fields of application like development and emergency response, the operational implications need to be precise in order to be actionable. Thus, the elimination of biases in telco data is key. “That is precisely, where we put most of our effort on”, says Bengtsson. Biases in mobile phone data could originate, for example, from phone ownership and usage patterns, which could lead to overestimations of mobility and relocation flows. As part of the efforts to improve the models, careful blending with survey data could help, but in developing countries reliable and on-going surveys are oftentimes challenging enough. The issue thus remains one, where academic rigor and scientific precision are needed to enhance value creation from data in low- and middle-income countries.
Could motives be shifting?
With data access and operationalization being priorities on the agenda of researchers and non-profit organizations, one could take the emergence of startups like zeotap as indicators for potentially shifting motives on the side of telco operators. As awareness about the value of their data is undeniable, monetization of this asset could become crucial. “For telcos in emerging countries, analysis of data for e.g. advertising is not yet a big issue, but will surely become”, says Mr. Bengtsson. However, for many of the development applications we work on, the case to monetize is much weaker and the stakes are high if you get it wrong”. Eventually though, the topic might arise and put the issues of data accessibility in the spotlight. In such a case, Mr Bengtsson, his team and Flowminder are well positioned to effectively lobby for continued access of telco data for the developmental cause. Their success stories from various countries, at least, serve as a splendid example of the power of data, academic rigor and direct local impact.