Facebook data can be valuable for nonprofits targeting new donors, according to research published earlier this year in the journal Decision Support Systems.
The researchers constructed a model using certain Facebook data with high predictive value — data related to age, education, residence and interest in nonprofits — which can be used in targeted marketing campaigns to hone in one new donors.
From the paper:
Most nonprofit organizations view social media as a simple communication tool and fail to grasp the opportunities related to the analysis of social media. Some researchers (e.g. Quinton & Fennemore, 2013) have described this problematic but, to the best of our knowledge, none have actually attempted to offer a solution. In collaboration with Plan International Belgium, i.e. a well-known Belgian nonprofit organization, this research tries to unravel the true value of social media to nonprofits. This is done by presenting two methods for the analysis of the behavior and characteristics of Facebook fans of Plan Belgium. Specifically, this study consists of a cluster analysis on the one hand, and the construction of a predictive model on the other hand.
Nonprofits are confronted to the reality that online support almost never turns into offline donations. In order to respond to these issues, we presented two distinct methods for the analysis of Facebook data that nonprofits could incorporate into their social CRM efforts. More specifically, we focused on the Facebook fans of Plan Belgium. The analysis of Facebook fans enables nonprofits to gain knowledge about individuals who are interested in their organization and, who they would traditionally have no information about. This makes the presented analyses particularly relevant to acquisition efforts.
First of all, we demonstrated that Plan Belgium’s Facebook fans can be clustered according to the set of Facebook pages that they like. Hence, the presented cluster analysis empowers Plan Belgium to discover and discern different segments of Facebook fans. We find that the clusters of Facebook fans differentiate themselves in terms of geographics as well as personal interests. In addition, we were able to identify certain clusters that contain significantly more donors than others. Dutch-speaking Facebook fans are generally more likely to be a donor than French-speaking ones. Based on the differences in the relative number of donors, Plan Belgium could decide which cluster(s) to target. Subsequently, they could use the characteristics derived from the cluster analysis in order to develop segment specific marketing strategies.
The second part of this research focused on the construction of a predictive model for the classification of Plan Belgium’s Facebook fans into donors and non-donors. More specifically, the model generates predictions for a Facebook fan’s propensity to become a donor. By evaluating this model, we are able to assess the predictive value of social media data for the identification of potential donors amongst Plan Belgium’s Facebook fans. This model could enable Plan Belgium to detect the very few donors amongst the haystack of Facebook fans. On the other hand, the fact that very few Facebook fans are donors, makes the construction of a predictive model inherently difficult. Nevertheless, we found that, to a certain extent, individual specific Facebook data is predictive for donorship. Specifically, the categories of Facebook pages are, by far, the most important Facebook features. Their predictive power is not significantly lower than that of all Facebook features combined. Moreover, Plan Belgium could use this predictive model in order to improve targeted acquisition efforts.