Propensity Modelling

Analysing A Donor’s Propensity To Donate

Donor propensity modelling analyses a donor’s propensity to donate – allowing you to target more strategically and efficiently, eliminating budget wastage on those who are less likely to donate, and thereby making your marketing strategy more effective.

– Defining new best-practice in the NFP industry

– Donor-centric mentalities and principles

– First steps in building a more sustainable NFP business model

– Differing significantly from traditional static segmentation

– Incorporation of a series of variables in any combination

– Scores the propensity of an individual, not a group or segment.

– Incorporates machine learning & collaborative insights with over 9 million NFP transactions from over 60 charities across Australia


Mail Makers Integrated has recently worked with a Not-for-Profit organisation on their Tax and Spring 2018 Appeals to ultimately test the efficacy of Predictive Modelling to increase revenue while reducing costs through better targeting.

Out of the 28,516 high and medium score donors that were mailed during Tax 2018:
(donors who received a model decile between 10 and 20)


of Responses came from top 3 decile bands!


of Total Campaign Cost ($12k) could be saved whilst still generating 86% of total campaign revenue ($380k)

Key Findings from High Scoring Donors:


Return on Investment


of Campaign Nett Profit


of all HS donors are those with a last donation date that is older than 24 months

70% of the Campaign’s profit came from those that would have been selected as Lapsed or Very Lapsed Donors via RFM
(Last gave >12 months)

There were 271 donors who had a last donation date from 2012 – 2001 (>5 years)

A donation total of $23,340.85 would have been missed if our client had mailed according to traditional RFM selections (up to 5 years). It works to be about 5% of the Total Campaign Income that would have been excluded.

Spring 2018

For the Spring Appeal, our client usually mails to 30k donors, however due to a follow-up Propensity Modelling analysis, they mailed 16,730 for 2018. On top of that, they also did a Wave 2, mailing HS donors who had not responded to Wave 1 (6,676).

Total mailed for Spring 2018 was 23,406 records.

Comparing the client’s 2018 Campaign (both Wave 1 & Wave 2 with Predictive Modelling) to 2017 Spring Campaign (1 wave only without Predictive Modelling):


Less Mail Packs Sent


Less Money Spent


More Nett Income


2017 Responses

In 2017’s mailing, the majority of the mailing to 15,407 records only produced 103 responders.


2018 Responses (Wave 1)

In 2018's Wave 1 alone, the majority of the mailing was to 8,017 HS records, producing 1,860 responders.

The High Scoring segment consistently produced:

Highest Gross Income

Largest Average Gift

Most Number of Gifts

Consistently in both waves,
Model Score 20 (highest) had the Most Number of Donations.