Small loans and savings for people the banks overlooked, on any kind of phone.
Our client had a product people clearly wanted and almost no way to deliver it. They offered small loans and savings accounts to people the banks had written off, mostly in rural areas where the nearest branch could be half a day away. The economics were punishing. Every transaction cost too much to process, and because these customers had no formal credit history, deciding who to lend to meant a manual review that took the better part of two weeks.
We worked on three problems at once. The first was reach. Not everyone they wanted to serve owns a smartphone, so alongside the app we built a full experience that runs over USSD, the plain text menus that work on any handset. The second was cost. We rebuilt payment processing to move through several providers at once, choosing the cheapest reliable route for each transaction. The third was credit. We trained a scoring model on the signals these customers do produce, from repayment behaviour to how money moves through an account, so a lending decision takes seconds instead of days.
The response came faster than anyone expected. More than 100,000 people signed up in the first three months. 95 percent of loans were repaid on time, which told us the model was reading real creditworthiness rather than noise. Transaction costs fell by 60 percent, and within the year the client had carried the same system into 3 countries.