When the Start Analysis button below is clicked, a synthetic donor dataset is generated using the configured settings, RFM segmentation and churn-risk prediction are run on it, and a downloadable PDF report along with a CSV dataset are provided. No file upload is required — the entire process runs automatically.
About This Demo
This app is a simplified, publicly shareable version of a donor segmentation and churn-risk analysis project originally developed for an organization. In the original project, an institution-specific, action-oriented strategy document was also produced alongside this analysis; that strategic component is not included in this public version, which presents only the analysis and reporting pipeline.
How Is Churn Risk Calculated?
Rewinding Time (Cutoff Logic): The model picks a specific point in the past (e.g., the last day of a month) as a “decision moment,” positions itself as if it were that date, and looks back at the donor's history up to that point.
Defining Eligible Donors: Not everyone is tracked — only donors meeting certain criteria (e.g., at least two donations in the last 180–365 days) are included. The goal is to focus on protecting the active donor base we already have.
Labeling by Looking Forward: A forward-looking window opens from the decision moment (e.g., the next 180–365 days). If the donor gives again within that window, they're labeled “retained”; if not, they're labeled “churned.”
Monthly Checkpoints: Each month-end is treated as its own checkpoint, so the same donor can appear multiple times in the training data across different months — giving the model more examples to learn from.
What Is RFM Analysis?
RFM is a classic segmentation approach that scores donors on three dimensions: Recency — how recently did they give, Frequency — how often do they give, Monetary — how much have they given in total. One important detail: RFM is sensitive to the selected time window; the same donor can land in a different segment depending on the date range chosen. In the original project, a separate strategy was developed to address donors' segment transitions — such as declines or upgrades — over time.
🤖 Demo Data Parameters
Create synthetic donation and donor data with your chosen parameters.
Leave this unchecked to get a brand new, random dataset every time. Check it and type any number below to recreate that exact same dataset whenever you want.
⚙️ Analysis Parameters
Reference end date for churn risk analysis.
ℹ️ The analysis will run on data generated from the parameters specified above.