Turning Donor Data Into Decisions With HubSpot's Data Agent

Eric Housh and Zack Terry dig into a real implementation with a university fundraising team sitting on a database of over 100,000 donor records — and no reliable way to know who to call first. The problem wasn't missing data. It was that the data lived in silos across HubSpot, making it nearly impossible to surface the right relationships at scale.

Zack walks through how FSM used HubSpot's data agent to pull together contact history, giving records, family relationships, and event attendance into a single scoring output — high, medium, or low — so the team could stop guessing and start making targeted outreach decisions. The result is a system that applies senior staff expertise automatically across thousands of records.

This episode is for any nonprofit, development team, or relationship-driven organization wondering whether their CRM is actually working for them.

For the full episode video, show notes, and related resources, visit: https://loom.ly/vi7hTbc

Transcript

SPEAKER_00 0:09

AI in Action is brought to you by Fast Slow Motion. Our team helps growing businesses put AI to work with practical, scalable solutions. To learn more about how we can help you implement AI in your business, visit FastSlowmotion.com/slash AI. Hello and welcome back to another episode of AI in Action. I'm your host, Eric Hush, joined as always by Zach Terry, Director of AI at Fast Slow Motion. Zach, we talk a lot about using AI to make businesses run smoother, but today we're looking at a sector where the work is deeply personal, uh, higher education and nonprofits. So we're going to be talking about a university fundraising team. Uh, these guys were sitting on a database of over 100,000 records. Again, an incredible network, but the goal was to find better ways to highlight the people who care most about the school's mission within those disorganized records.

SPEAKER_01 1:01

Truly a volume challenge in this case, but it's volume combined with very specific data about who is going to be most likely to be a great donor. So you've got an institution like this that has a wealth of information, but it's scattered across a bunch of different silos across the organization, even within their HubSpot org itself. This data was located in a bunch of different places. And so while I can go and look at a contact inside the system and I can eventually figure out their donor history or anything else that that may be going on that could lead to them being a good potential donor, it was pretty difficult to go in and find that from one unified location. So the opportunity here was really to move from assuming who might give towards actually knowing who to call. So take it from conjecture and trying to figure out where I should spend my time to really identifying that at scale and pointing people in the right direction. So the team was really spending a lot of time on the administrative side of things, sorting through this data. It was limiting capacity for actually doing the donor engagement. So reaching out and cultivating those relationships. If we're spending a lot of time just trying to identify the right people as opposed to actually cultivating relationships with those people, you can see how that could be a problem. So this project was really about proving how just a little addition to what they were already doing, adding a little bit of AI into their existing HubSpot instance to help them actually work with the data that they already had. It was able to turn that information into better, more clear, actionable steps for identifying and going after the right donors.

SPEAKER_00 2:40

Well, let's start with the problem, uh, some of the hurdles this team was facing. Why was it helpful for them to look beyond just the standard software settings when they already have a system like HubSpot in place?

SPEAKER_01 2:51

This is really any CRM. You have it for a long time, you put a lot of data in there, you start building out customizations, and those customizations are all well intended, but over time things can sort of start to get a little bit separated, things can start to get moved into silos, and your data can get a little bit disconnected, right? So, for example, you might have a contact record that could show a bunch of past gifts or specific event attendance. I mean, it's it's all recorded, but it's not easily accessible, or at least not in a way that can help you make a decision when you're looking through thousands of records at a time, right? So it could be located across different objects, different properties, not easily visible from a single place. So we looked at a few ways to improve this. One is the ability to sort through those records. We wanted to move away from the team having to spend time qualifying these potential donors manually. We wanted to organize relationships. So look at the way that contacts are associated with different family members inside of the organization so that we can track and surface true donor relationships. So, for example, if somebody is attending a university, well, the best donor may not be the student at that time. It may be the parent. And then the student, years and years later, may become the best donor. And so it's all about how do we track that information in a way that's going to allow us to track that relationship over time and determine who the best donor is going to be and when to reach out. And obviously, we don't want to reach out too soon. We don't want to reach out too late. We want to find the sweet spot. So, how do we use data to help identify where that is, right? And then there's the systematization of the experience. So we really wanted to take the high-level expertise of some of the senior staff and apply that logic across these thousands and thousands of records automatically. So when you think about people that have been doing engagement for a very long time, how are they thinking about high potential donors? What is the process or the system that they are using? Probably just something that's in their head, every single day to think about how to take thousands and thousands of data points and find the 10 that they're going to use that day, the 10 people that they're going to reach out to, that they're going to have conversations with, that they are going to cultivate those relationships with. So it's really kind of combining all of those things to say, hey, how can we build a system on top of the HubSpot platform, the platform they're already in, to help streamline this process and give them more actionable data?

SPEAKER_00 5:16

And you and your team tackled this by building something called a data agent. What is that in plain English?

SPEAKER_01 5:22

That is a part of Data Hub, which is a newer hub that HubSpot has that has a bunch of data processing capabilities. It allows you to aggregate data and visualize it in different ways. And then the data agent specifically allows you to actually do some transformations that involve large language models and things like creating prompts and hydrating those prompts with context and data about your customers. So in this case, we were able to pull all of that data together, the contact details, past gifts, school history, recurring donations, things that are really important to consider when determining who is going to be a really great donor, who you want to cultivate those relationships with. We were able to kind of pull that together using data studio. So creating sort of like a smart list that includes not only the information about the contacts, but all of that other related information. So super key information that is often stored as multiple related records. It can be relational history, it could be donor history, it could be the time they spent working with an educator or whatever that might be. All of that is relevant context. Then the data agent allows us to provide instructions or a prompt to the system, and we can hydrate that prompt with that data. And what's really cool about the data agent in HubSpot is it's not just the information that lives on the contact record, but we are able to actually pull some of the aggregate data from the related records and provide that full context to the AI. So we've got this custom set of instructions. It is being hydrated by the contact data and all of the related data that we have in the system. And all of that provides rich contextual information that's located all within Data Studio that allows us to pull together all that history. So what does that mean? Why do we need that history? Because then we can take that and create a simple output. So something called a smart column in HubSpot. Imagine you just have a spreadsheet and you have a special column of that spreadsheet that's able to use AI and all the context that you have to output some result. Now, this could be a summary. This could be some rich text that you want to output and maybe you want to read the story. But in this case, we needed something simple. We needed the ability to quickly understand who is going to be the right person to reach out to today, right? So we didn't need a bunch of text and a bunch of information. We wanted to distill all these data points into a simple high, medium, low, right? Reach out to today, cultivate for tomorrow, don't bother. And if it's don't bother, maybe that's because there is a signal that is indicating that they have said, please do not contact me about this anymore, or something along those lines, right? And then for medium and high, it's aggregating some of those data points to say, hey, how do we know when we're looking through our contact database? How do our senior people know when to spend more time reaching out and cultivating relationships versus not? So that's where we're taking that information, that process and systematizing it using this data agent capability to hydrate a prompt, provide exact instructions for what we want to output, and then ultimately just providing a very simple high, medium, low output that allows somebody to scan that list and understand what's going on. And then they can take that and they can turn it into marketing segments, lists. You can take the information and save it as a property on the contacts in the system. And so it's not just that point in time, but you're able to actually do things and take action with the data and the output that you're generating using the data agent.

SPEAKER_00 8:59

Seems super powerful. Let's talk about impact. Once this data agent is up and running, uh, what's in it for the university? What are some potential outcomes here? You're getting more targeted outreach.

SPEAKER_01 9:09

So you can instantly find the best prospects to reach out to, ensuring that you're saving personal time for the most promising relationships sourced from that historical data without having to actually go through and find all that historical data yourself, because this tool is able to go and synthesize that for you. It's able to find missing opportunities. So maybe it's finding people that have slipped through the cracks, people that have been overlooked, maybe someone who attempt attended a campus program years ago and has the capacity to be a major donor, but it wasn't surfaced directly to you. And so uh it has the ability to synthesize that information so that you're not missing out on opportunities. It's reducing manual review. So that administrative burden that we talked about at the top of uh this episode, it's it's significantly lower with a tool like this because it's able to handle that research at a speed and scale, which would be very difficult for a single person to replicate. And then it leads ultimately to some more leadership clarity. So now management has a single source of truth that helps them decide where to allocate their staff's time and resources. And it allows them to easily report on this information, to surface all the donors in the system or the people that are most likely to be donors when they're going out and they're they're actually trying to create additional engagement.

SPEAKER_00 10:26

So, Zach, for the fundraising VP, the development director that may be tuning in, listening to this, uh, this is piquing their interest. What's the main takeaway?

SPEAKER_01 10:34

And so it's it's taking what your employees and your team are already doing day to day, but figuring out how to put that into a process. And then you're able to take that process and scale it across the organization. So you probably already know what a perfect or an ideal donor looks like. And if you're not in a donor organization, this is your ideal customer profile. And so um it could really apply kind of across any business, whether you're a nonprofit or not. And so then you're able to use AI as a worker that can apply that expertise, but you've you've got to be able to extract it first. You have to be able to systematize it. So this was really a great proof of concept for this client that shows how starting with a specific goal, in this case, it was kind of about lead scoring and donor scoring, that can build the trust that's needed to expand that automation across other departs. And so now that they have this tool in the system, it's working, it's helping to identify these very high potential donors. Now they're all thinking about, hey, how can I apply this to the work that I'm doing? Because they've seen how successful this can be.

SPEAKER_00 11:40

I love that. And I love that connection with ICP. That's something we've been thinking a lot about here internally. Uh, again, very powerful stuff for the leaders out there that are listening to this, wanting to make this sort of move toward this kind of relationship-driven strategy. Uh, what's what's their next move?

SPEAKER_01 11:55

I'd say reach out to us. Reach out to Fast Slow Motion. We can be your second set of eyes on your setup. Often it really is just a few small changes that can make a big difference. So maybe it's how the data is stored or it's how you track families or households in a donor organization. Just evaluating and making a few small changes can really unlock a lot of potential there. So I would say this week, if you are a nonprofit or a donor organization, take a look at your donor list and ask yourself if you have a unified view and you have very clear signals that point towards where you need to focus your time. And if you don't, there's a huge opportunity there to potentially automate some of that research and uh aggregate those signals into something that's really usable. And then, of course, I mean, we would say that the AI is a great way to help you do that. So uh reach out and let's see if we can have a conversation and find some opportunities to use tools like this to really streamline these areas of your business.

SPEAKER_00 12:48

Good stuff, Zach. Thank you for uh sharing the details of this project and this story with us. Uh, great reminder that AI can turn a database into a relationship engine, uh, really surface some key insights to help you uh sort of manage that database more effectively, manage those relationships. Uh for more info on this, visit fastlowmotion.com/slash AI. If you are interested in engaging us, you can do that through the website as well. Uh Zach, any final words?

SPEAKER_01 13:12

I think this is just a great example of how a great AI use case could just be as simple as changing the lens through which you view your data. So that can enable you to add insights that you didn't have before. So really, I would say no need to overcomplicate it, right? Keep it simple and just think about ways that maybe your data can reflect back some information to help you make decisions. Awesome stuff.

SPEAKER_00 13:33

Thanks for listening, everyone. We'll see you next time.