Reducing Attrition with AI
A Regular Giver lapsing is costly for Not For Profit’s (NFPs) and often difficult to counter. The Data Science team at Adroit set out to produce a fast, low-cost attrition model using AI. This model was designed to predict lapsing supporters within any NFP’s supporter base. The new model could immediately adapt to new data. The new approach can provide instant results, allowing NFPs to react quicker and significantly speeding up the overall time.
To build learning, the project used over 100 bases comprising accurate data from 14 large Not-For-Profit (NFPs) organisations covering 30 countries globally, including UNICEF, Save The Children, Plan International and Oxfam. The datasets for each organisation varied but were generally in the millions of records. Our Data Science team leads the project to provide Machine Learning and AI development for our clients in many sectors.
The solution was designed specifically for NFPs, needed to meet three principle aims:
- Speed: a key aim was to apply the process to new datasets and reach full completion the same day.
- Minimal loss of capability: While developing specific models on each dataset may produce a slightly better result; it would require much more significant time. The aim was to create an effective general model capable of good results.
- Replicable: The modelling process should be able to be applied to relevant datasets without modification.
In a broader sense, the project’s purpose brings Machine Learning capabilities to the point where they could be applied quickly and efficiently (on-demand) for many organisations. As a result, the process is less demanding or taking significant resources.
In a broader sense, the project’s purpose brings Machine Learning capabilities to the point where they could be applied quickly and efficiently (on-demand) for many organisations. As a result, the process is less demanding or taking significant resources
The resulting application can be applied to new datasets comprising millions of records quickly, with minimal effort. Producing results are swift; practical, and on-demand. The project using AI has radically transformed the costs, time and effort required to predict lapsing supporters successfully.
The streamlined process allows Machine Learning technology to be deployed in various applications, opening up Machine Learning technology for real-time, everyday use within NFPs. In addition, this gives NFP’s the ability to reduce attrition rates among Regular Givers’ with a significantly reduced effort.