Looking for the lapsed
UNHCR Spain (ECA España con ACNUR), part of the UN’s global refugee support mission, relies on regular donations to fund its life-changing work. But with donor churn a growing challenge, they approached us to help spot the signs of supporter disengagement. We were to build a smart, data-driven model to predict when donors might lapse, so the team could step in early with personalised communications. A proactive approach to retention would ultimately help protect this crucial income stream – and strengthen UNHCR’s ability to deliver on its promise to refugees, displaced people, and those without a state.
Chats and check-ins
Stakeholder engagement sits at the heart of everything we do. With UNHCR Spain, that meant building trust and staying in sync from day one. We collaborated closely with their skilled data analyst, sharing detailed methodology, exploratory insights, and regular check-ins. Our “double check-in” process ensured full alignment on variable selection, avoiding costly surprises later. To futureproof the work, we created readable, modifiable code and detailed documentation – all in Spanish – and trained the internal team to own the model long-term. This led to a sustainable, empowering handover built on transparency, collaboration, and a shared commitment to lasting impact.
Incredible insight
The outcome was a bespoke machine learning model that scores each donor’s likelihood of churning, using data from UNHCR’s warehouse. With this predictive insight, UNHCR Spain can now identify at-risk donors early and target them with tailored retention messaging. Though the model is only a month into deployment, the team is already preparing A/B tests to optimise communication strategies. This early adoption marks a big step forward in securing stable, reliable income – essential for planning humanitarian support. It’s a shining example of what happens when technical expertise meets strong relationships and purpose-driven collaboration.
