New research by DevEng Prof. Joshua Blumenstock and colleagues suggests that anonymized metadata from calls, texts, and data usage can be a promising way to measure mobile phone users’ — or a whole region’s — poverty or socioeconomic vulnerability.

Their study across Afghanistan, Côte d’Ivoire, Malawi, and Togo offers a promising new tool for governments designing social safety nets.

The findings are encouraging for data-scarce environments: You don’t need smartphone data or mobile banking records to get useful results. Blumenstock and his colleagues found that basic call and text logs carry enough signal to build surprisingly accurate poverty-prediction models. And while larger datasets help, governments can start with as few as 1,000 surveyed households to pilot the approach.

Their method, however, works best for identifying the chronically poor in diverse, nationally representative populations; it’s less reliable for measuring short-term hardship or reaching those without phones at all.

Mobile data won’t replace household surveys — the traditional gold standard for these measurements — but it may make poverty mapping faster, cheaper, and more responsive.