Guide Dogs

Guide Dog’s supporter management system provided automated weekly data feeds of all supporter and transactional data into a single customer view database. Data was then cleaned before being delivered to users through Apteco FastStats, who could then create and implement their data-driven supporter journeys. The project undertaken by Guide Dogs and their partner Qbase, had 4 stages – 1: Identify the variables that predicted pledger activity, 2: Create models to score supporters based on propensity to become a pledger, 3: Geo-spatial analysis to plot top segments and compare to known pledgers, 4: Action and implementation.

Stage 1 was to identify the variables that predicted pledger activity. Transactional variables were created and tested based on 4 giving products - cash, raffle, Sponsor-a-Puppy and Dogalogue. Profile reports were created, enabling comparisons to be made across each variable and to rank them according to predictiveness. This was done by comparing the profile statistics for known pledgers versus general supporters. Stage 2 of the project was to create Decision tree and Predictive Weight of Evidence (PWE) models to score supporters based on propensity to become a pledger. Selections were then made from each model based on the PWE score, and the resulting supporters were captured in Apteco FastStats to produce the final audience. Geographical analysis was then undertaken in stage 3, using Apteco FastStats integrated with MS Mappoint. Actual pledgers and the top segments of modelled potential pledgers were plotted, and the most suitable locations were identified, based on drive-times, profiles and pledger propensity scores. As a result of the extensive analysis, the implementation stage – stage 4 could be actioned. Invitations were then sent to the best potential pledgers inviting them to attend events held at hotels around the country. The top propensity potential pledgers were asked if they would like to receive a home visit if they couldn’t attend the event. The next propensity segments were telephoned if they didn’t respond. The lower propensity segments were sent a direct mail piece if they didn’t attend.

Project metrics were determined and used to judge the success or failure of the approach. Benchmarking was based on comparisons to previous years. With all KPI’s from the previous year being beaten the project was deemed a success and showed that the quality of the attendees was much better, with conversion rates being higher than had ever previously been achieved. During the telephonic activity, it became clear that many of the attendees wanted to attend and were interested in leaving a legacy but were unable to travel. Based on the telephone feedback, the strategy was then amended to include home visits to supporters with the highest propensity scores. The legacy team also moved many of the events to out of town-based hotels and into smaller venues that were closer to the attendees. This was reported to improve interaction and one-on-one conversations even further. Understanding the supporter experience and being able to proactively evolve the supporter journey had the knock-on effect that event costs were reduced through cheaper, smaller venues. A more personal service was delivered to supporters via home visits to the highest propensity supporters and minimising travel to events for the lower-scoring supporters.