Data Science and Artificial Intelligence for Good

And in its free time, AI helps businesses increase revenue and customers.
Applications of AI and machine learning in the for-profit world are our signals to what’s about to happen in the non-profit world. If you haven’t seen the current applications in the business world, I recommend reading Eric Siegel's book Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die.

The most exciting technology, in this space, is natural language generation, which is exactly what it sounds like. Using text mining and natural language processing, a computer can create language, text, and narratives that read like a human wrote them. Companies like Google, AP, and the LA times have found ways to create narratives using facts and natural language generation. In the non-profit space, companies like Gravyty generate natural looking emails to engage prospects.

An organization that can combine these applications of AI and natural language generation to create small chunks of tasks distributed through a single, easy-to-consume platform will differentiate itself from all the other organizations.

Rather than asking people to look at reports and Excel files, if we can generate small tasks, true next best actions, we are likely to see more action on our insights. This HBR article by none other than Tom Davenport explains how Morgan Stanley is using the principles of next best action.

Let’s explore this "next action" idea using a real application in the fundraising world.

USC's Action Center

Using the idea of one action at a time and with the help of our fantastic Salesforce development team, we built an app at USC within our Salesforce mobile instance called Action Center, as seen in the images below. A fundraiser with active assignments sees some of these action items on his or her Salesforce mobile app.

The four main categories a user sees are:

1. Donor News
2. Prospect Recommendations
3. Gift Alerts
4. Proposal Cleanup

We have built a crawling engine that looks for assigned prospects who may be mentioned in the news. We then perform entity matching to ensure that the entity mentioned is the one we were looking for. Later, using all of the news items, we select and display news on the most important and relevant entities.

If a fundraiser’s portfolio is active with constant visit and qualification activity, we show them an unassigned prospect who they might be interested in qualifying.

These recommendations are based on the characteristics of the fundraiser’s existing assigned prospects. We try to recommend a prospect that is most similar to the majority of the existing prospects. We measure similarity using the prospects’ addresses, degree departments, giving likelihood, and other factors.

With the simple click of a button, a fundraiser can request an assignment or see more details on the prospect without leaving the interface.

If a donor from a fundraiser’s portfolio makes a recent gift, we show this to the fundraiser for an easy touch point with the click of a button. Often, if a donor has multiple giving areas, the fundraiser managing the relationship may be unaware of other gifts.

Currently, USC Advancement’s fantastic data folks (James Sinclair, Michael Pawlus, Jing Zhou, and Rodger Devine) are exploring network graphs of interests, prospects, opportunities, experts, funders and donors to create auto-generated proposals, leads or opportunities. It may look something like this:

Where do you think AI will take us in the non-profit space? Are you doing something exciting that you would like to share?