Backtest of Projection Model Analysis

Backtest of Projection Model

How to use the Backtest of the Projection Model analysis

This analysis allows you to retrospectively evaluate the accuracy of your donor projections by comparing forecasted donor counts and revenue by tier (Top, Middle, Bottom) against actual results for each of the past ten fiscal years. Using your gift transaction history and defined segment thresholds, it simulates what your forecasts would have been if you had only used data available prior to each target year, creating a robust backtest of your projection model. The analysis takes into account donor transitions between tiers, retention and lapse rates, and the influx of new or reactivated donors, delivering detailed metrics such as transition probabilities, average giving amounts, and error percentages for both donor counts and revenue.

You can use this analysis to understand not just how many donors you retain, upgrade, or lose each year, but also how accurately your model predicts these movements. By breaking down performance by donor segment and tracking the share of new/reengaged donors, you gain actionable insight into the underlying drivers of your annual fundraising results. The side-by-side display of forecasted and actual outcomes allows you to pinpoint where your projection model is strong and where it may need refinement—critical for strategic planning and resource allocation.

Key metrics include forecasted retained donors, projected and actual revenue, transition percentages (such as upgrade, downgrade, and lapse rates), and the rate and value of new/reengaged donors. These measures are essential for understanding donor lifecycle behaviors and the effectiveness of your stewardship and acquisition efforts.

You should review these results as part of your Constituent Intelligence routines, especially at fiscal year-end and during strategic planning cycles. Fundraising directors, analytics leads, and those responsible for annual fund growth should use the insights to adjust forecasts, refine segmentation strategies, and tailor stewardship approaches. For example, persistent underestimation of new/reactivated donors may signal an opportunity to invest more in acquisition, while high lapse error rates could prompt targeted retention campaigns.

To measure the success of using this analysis, track improvements in forecast accuracy over time, monitor increases in donor retention and upgrade rates, and assess whether campaign strategies based on these insights lead to higher actual revenue and engagement. By integrating these findings into your ongoing Constituent Intelligence practice, you can make data-driven decisions that strengthen donor relationships and drive sustainable fundraising growth.