Using Predictive Analytics to Cultivate and Convert | npENGAGE

Using Predictive Analytics to Cultivate and Convert

By on Nov 20, 2012


I checked my personal email inbox one day last week (as I do most days) and I perused some of the tens of emails I receive each day from organizations I have supported or do support.  But do I open them all?  Based on my interest in fundraising, I usually do, but we know that open rates on nonprofit emails are around 13% (give or take, depending on the study you look at).  As I perused through the emails, many followed the ‘book’ on creative and copy, including short, interesting subject lines.   Copy and creative are becoming less and less of a differentiator when it comes to marketing messages (especially combined with commercial marketing messages).

How to Stay Relevant with Analytics

One way to stand out is to classify your constituents and communicate to them with a message that is timely and relevant to them.

For instance, I donated to an organization via a peer-to-peer event that a friend was participating in.  The organization knew this information about me and used it to classify me in a different affinity group than the participant or a donor that gives directly to the organization.  The information they delivered was relevant, in that they reported back to me the success of the event that I had donated to, gave me a list of upcoming events, and made an attempt at converting me by offering me information about the organization that was interesting to me.  They may have classified me as a peer-to-peer event donor with an affinity to become a core donor based on other data collected about me and others (including potentially using cooperative databases).  I opened the email and subsequently gave a follow up gift.

What is Classification?

A classification model uses data where the category membership (peer-to peer with affinity to convert) is known and applies learnings from that data set to the unknown.

For instance, the organization I am referencing above may have data on first-time donors who gave an inception gift to a peer-to-peer event and categorized those into donors that gave subsequent gifts to the organization vs. those that have not (there may be more categories, like convert to participant, multi-year peer-to-peer event donor, but I will stick with a simple example here). Once the data has been collected, one should enrich that data with wealth, other demographic or psychographic data to help bolster the accuracy of the model.  There are a couple of tools you can use in building your model, including logistical regression and a decision tree (simple example below).  These tools will help to identify what variables are the strongest predictors and help to build the model with the minimum error rate or best pruned tree.  You can then take the results of this model and apply it to all new first time peer-to-peer event donors – allowing you to communicate with them in a more meaningful way and retain and upgrade a higher percentage.


Classification is a great way to leverage the power of analytics to best communicate to your constituents.  As you accumulate more data, you can revise and refresh the model to ensure that it is as accurate as possible.  Target Analytics has another great packaged way to use classification in their loyalty insights model.  It is critical in this time of unending communications that we continue to differentiate our message from competing messages.  Hopefully this will help spark some conversation and thoughts about how your organization can use classification modeling to help further your mission.


Leave a Reply

Your email address will not be published. Required fields are marked *