Sightsavers reactivates lapsed donors with Acxiom
Sightsavers International is working with Acxiom to reduce direct mail wastage and increase its campaign return on investment.
Sightsavers had amassed a large database of lapsed cash donors whom they had mailed en masse in an attempt to reactivate giving. However, the age and volume of this dataset meant it was becoming increasingly uneconomical to continue mailing in this way.
The charity therefore tasked Acxiom with helping it understand more about its lapsed givers and their propensity to respond to direct mail, with the aim of discovering how most effectively to segment this group.
Acxiom helped Sightsavers build a lapsed cash donor model which scored previous donors and established a cut-off point below which it would be uneconomical to mail. Before that Acxiom analysed data from a previous Christmas campaign using CHAID and regression analysis to identify eight variables that best predicted response. The variables were selected both from Sightsavers’ own transactional data and Acxiom’s mInfoBase™ Lifestyle Universe (ILU) database which provided demographic and behavioural insight.
Using these variables, Acxiom validated the performance of the model in a trial before using it on a live campaign. Tests on a mailing file of over 95,000 donors, showed a 120% uplift on return on investment (ROI) when the scoring system was implemented.
When implemented in a live campaign, Sightsavers was able to suppress 20,000 low scoring individuals, resulting in a mailing file of around 80,000. This represented a cost saving of £7,500 for the charity, whilst also achieving a higher ROI. Application to subsequent mailings has also delivered savings of over £20,000.
This was the first project of its kind for Sightsavers International. A spokesperson said that this project’s success “has now prompted us to work with Acxiom to see how we might be able to build other models to improve our DM activity, in particular to help reduce committed giving attrition rates.”
www.sightsavers.org