Use Case: Donor Churn and Prediction Models

Donor Churn and Prediction Models

Name

Use Case: Donor Churn and Prediction Models

Identifier

UCNP01

Description

Nonprofit organizations can benefit a great deal from the analysis of their data. Using that conclusion, it is easier to forecast the factors that affect their donor membership and donation amounts. Thus, there is a great possibility for improvement in both.

By tracking donor churn, a nonprofit can achieve the following two objectives:

  1. They can intercept problems preemptively and prevent donors from leaving the donation process. Propagating loyalty is essential for the continued well-being of a nonprofit organization.
  2. They can introduce donor retention strategies when donor churn is triggered. This may include retaining the existing donor base as well as acquiring new ones.

Goal

One of the main aspects of donor behavior is churn. The term refers to the percentage of nonprofits’ donors who will choose not to contribute again. Some nonprofits are community-funded charities only. Thus, they must use their marketing resources in exactly the manner that will help them continue to thrive. Predicting churn and the likelihood of donation are two important ways of doing that.

Through data analytics and machine learning, donor data can be used to create a better plan, process, budget, and accurate forecast of donor engagement. The experts at 4sights Analytics can provide data analytics consulting to extract the following useful information from the raw data they receive:

  •     The actual percentage of contacts that are donors with ~96% accuracy or above
  •     The relative proportions of non-repeat vs. repeat donors with 80% accuracy or above
  •     The ratio between recurring donations vs. total contributions
  •     Whether there is a positive relationship between the donation amounts and frequencies
  •     Signs that allow the identification of donor churn preemptively
  •     Prediction of the amount donors will gift next
  •     Determining a way to score which non-donors can be encouraged to donate

Preconditions

  1. All the relevant information, such as that of donors and the amount they have given, will be made available and accessible to the analysts.
  2. The available data is in such shape that it can be directly pulled from the nonprofit’s sources into an analytics platform. It can then quickly be analyzed and used to build predictive models.  If data resides in multiple sources and formats data engineering effort is required before producing insights.

Highlights from our example use case

  1. Our engagement began with the discussion of the goals with the nonprofit leadership team. Spent time with nonprofit’s subject matter experts to understand their business practices and data collection systems.
  2. Data is sourced from HubSpot or Salesforce CRMs spanning 4 years.
  3. Some of major Data elements included are; Contacts, donors, donation amounts, donation frequency,  and their engagement levels in nonprofit’s events, website visits, subscribing to emails,  social media activity, and marketing campaigns.
  4. Data is mined to
  •   Identify and exclude data anomalies from the model
  •   Produce insights,  trends
  •   Come up with relevant factors based on the extent of their influence on target outcome metrics
  1. Multiple predictive models are built using decision trees (random forest), logistics regression, and neural network algorithms.  
  2. Presented our insights and recommendations to the organization’s leadership team.  Enabling them to employ the ready-to-use insights to formulate a campaign strategy.

Notes

  1. More accurate models and predictions are possible with additional information.
  2. A deeper analysis can be carried out once the data concerns are addressed.
  3. Integrating the results into CRM applications can give the nonprofit real-time visibility.

With much of the data underused in the world of nonprofit fundraising, nonprofits continue to lose donors to churning, leading to a loss in overall donor contributions.

This report shows that 79% of nonprofits lack the personnel or time required to focus on data. More than 40% of nonprofits consider they don’t have the tools to employ the data they have. Even less than half base their decisions on it regularly. About 40% of them don’t collect sufficient data for statistical analysis. A slightly higher percentage of nonprofits won’t store it in one place. The decentralization of the data makes extrapolation even more challenging.

Despite the barriers that so many nonprofits are facing, big data is an untapped opportunity; its insights are just waiting for deployment. Fortunately, the organizations in question do recognize the potential data has.

With accurate data, there is no limit to the insights that are available and can improve operations. An essential factor, however, would be the use of statistically sound and valid methods of analysis. Thus, it is vital that nonprofits only use reputable and dependable companies to provide the said insights. Contact us to carry out data mining that derives the most out of the data you collected! Additionally, our data analyst experts will indicate the shortcomings via an initial analysis, so that you can stock up the right kind of data in the future!

Leave a Reply