What Stock Picks Can Teach Us About Prospect Research

Recently we came across an excellent article that described how the financial research team at FactSet used Natural Language Processing to analyze the most recent Tesla earnings call.

They were focused on the ways in which Tesla founder Elon Musk's tone changed during the course of the earnings call. Most earnings calls have an upfront section of prepared remarks that pretty much follow a boilerplate for every company and then that's followed by Q&A where financial analysts are given an opportunity to ask questions of company leadership.

In this call, the FactSet team zeroed in on the change in sentiment - how positive or negative language is - as Musk moved from the prepared remarks to the Q&A section. I'll let you read the article to see why things got negative.

How can we use sentiment in prospect research? Why would it matter? And where would we get it from?

Let's start with the fundamental source that drives NLP: words. There exist dozens of sources for linguistic data regarding our constituents, but the stuff directly from them is the best. To gather this data, you'll want to think about the following sources:

  • Gift officer visit notes
  • Phone interaction notes
  • Responses to emails
  • Social media posts and comments
  • Content consumption

Next, we have to think about what we want to get from sentiment data. While an historical review of donor sentiment can be a good starting point, I'm not sure there's a lot there that's going to change a gift officer's behavior. However, if you create a sentiment score for each donor and then look to see how it's changing, you've really got something.

If sentiment starts to dip down, that's a warning sign to dig in and figure out why things are going south - especially with a key donor prospect. If the score starts to gain positive velocity, timing could be optimal for making an ask.

Something we hear a lot is that NLP is sexy and exciting. You bet it is. We also hear a lot of curiosity after the excitement: "what will we do with it?" Start simple. If you can pass gift officer notes through a sentiment scoring model and get a sense for which direction things are going, you're already ahead of the pack. We keep pushing here into deeper and deeper applications of NLP and we won't stop hustling down the path of Artificial Intelligence for fundraising with our Isaac fundraising and engagement platform. If any of you are leveraging NLP for fundraising, tell us about it!

Nick ZecketsComment