Abstract Non-take-up of means-tested benefits is a widespread phenomenon which undermines the effectiveness and fairness of social policies. The digitalisation of the welfare state creates new opportunities for proactively contacting people who are potentially entitled to benefits, but do not take up their social rights. In this study, we report on how new data flows were used to reach out to potential beneficiaries of the Increased Reimbursement of health care, a programme targeted at low-income households in Belgium. By randomizing the period in which potential beneficiaries were contacted, we were able to identify a three- to four-fold increase in take-up among those contacted as a result of the outreaching activities. Households that did not respond to the intervention, the never takers, have lower pre-intervention healthcare expenditures. This suggests that non-take-up was reduced primarily among those who would expect to benefit most from receiving the Increased Reimbursement. Exploiting the combination of rich administrative data with experimental evidence, we also find that early responders are mostly older and have higher historic health expenditures than late responders. Furthermore, results point to the need for balancing well the inclusiveness of the intervention with an increased number of applications by ineligible people.