Introduction [00:00: 32] Sam: My guest today is Ralph Burns. Ralph is the Founder and CEO of Tier...
Michal Heiplik: Revolutionizing Public Media Fundraising
Introduction
[00:00:34] Sam: My guest today is Michal Heiplik. Michal is the President and CEO of Contributor Development Partnership. CDP develops data driven insights and provides fundraising services to more than 230 public radio and television stations across the United States. During its first five years as an independent organization, CDP has brought 500,000 new donors to public media and has facilitated more than 200 million transactions for its clients.
During our conversation, Michal and I discuss how CDP is deploying both predictive and generative AI to streamline operations and improve the performance of partner stations. We explore some of the limitations they've encountered with the technology, and the opportunities he sees for artificial intelligence to unlock even more value for CDP's clients. This was a fascinating look into the future of philanthropic giving, and Michal's passion for the company's mission is infectious. So, please enjoy my conversation with Michal Heiplik.
CDP's Unique Mission-Driven Approach
[00:01:33] Sam: Most businesses leaders have sat down at some point to craft a mission statement. I know that process for CDP was a little more unique than most, could you maybe talk me through the organization's origin story and how your mission and structure guide the way you operate today?
[00:01:49] Michal: Sure. Yeah. I mean, The story for us actually started like 12 years ago, not five years ago. We've been a standalone company for five years and we picked a type of a company that's very specific to us, which is a public benefit corp., because mission is really at the center of it. And the origin of the idea was from the need that public media stations, NPR and PBS stations across the country have had, which is when you start seeing decreasing membership numbers and support numbers, you’ve got to become inventive.
You’ve got to start using data in a different way. And that's kind of where our origins as a data company started, basically the idea was integrate as much data across all the various public media stations across the country and discern some interesting tidbit out of it that we can actually implement into a fundraising practice, right? So, we started even before any of the kind of big data models that exist today that we now know and talk about. We were trying to put that together 12 years ago when there were 600 NPR and PBS stations across the country. So, trying to bring all of their data together is a very interesting task.
So that's been our progression: start from the problem of challenging fundraising numbers and really trying to turn it around. And I think we've been quite successful in that in the last decade or so.
[00:03:01] Sam: I was looking at some of the numbers prior to jumping on and pretty impressive, 200 million transactions a year that you guys facilitate. So, an insane amount of data, that you're tracking, right?
[00:03:13] Michal: It is. I was most proud of, sorry, I’ve got to say it, but over a billion dollars in annual giving that we're able to track through this data. And when we changed that first letter to B from M that's pretty cool.
[00:03:24] Sam: I'm a way off that myself, but something to aim for. The strain on your customers must be immense, what are they going through right now? And how does CDP really help them to achieve their missions?
[00:03:39] Michal: I'll start the framing a little different because public media is an interesting beast, right? Because there's very little federal support to public media. Contrary to what people believe there's very little federal dollars. Most of it comes from individuals. When you look at the philanthropic landscape across the US, forget public media for a second, basically 80, 83, 85 percent of all the philanthropic giving comes from 15 million people. Notice I didn't say 15 percent of the population. I said, 15 million people, that's a super small subset of our society. And in the last decade, there have been 27,000 new nonprofits created. So, if I use the analogy of we're all after the same bowl of dog food, we are. Every nonprofit.
So, gaining an edge, looking different in that mailbox, inbox, text inbox, right. that's understanding the donor. And understanding the data behind that donor and what makes them tick, what makes them give, what makes them be a civic participant, if you will, that's really the edge that we're trying to gain for our clients.
So, our clients are stations, right? Have been facing that the change in broadcast. How much broadcast television do you watch Sam? Not much, right? I don't watch any, I don't even have an antenna or a cable anymore! Yet the entire public media system was built around broadcast, so moving it into the digital space and really understanding the individual, that's been at the forefront of what stations are facing and therefore what we have to tackle as a company.
Embracing AI and Incorporating it into the Business
[00:05:06] Sam: So how does AI help you guys do that? What was the initial spark that led to exploring the potential of AI and how it could supplement the work that your teams are doing?
[00:05:18] Michal: Yeah, I would say the start and I would just recognize we're still at the very beginning of this. I really believe that we're in like, maybe not the first inning, maybe the second one, but we're still at the very beginning of this journey. We're still learning. So, I want to acknowledge that first, because by no means, I would say that we have figured out exactly how we're going to use and deploy AI and AI is still developing in a really interesting way.
But the initial spark was like everybody else, right? You read an article about ChatGPT and you go, you log in and ask some dumb question. That's basically how it starts, right? Or have an image generated or something like that. So, we as a team started playing with that, and then you evolve into the next phase of it, which is we were just working on a very complex technological project for last 16 months, right? Many people involved, a lot of money spent a lot of really, it just needed to happen on time. So, deploying AI to track large meetings, notes, action items, right? Super helpful.
As it graduates, you start getting ideas about could an AI engine write a fundraising email? Can it do it better than our in house writers? That's an interesting challenge and I'm happy to talk about some of those tests. And then, then really seeing, by the way, and probably the best application we've seen is believe it or not around major donor fundraising. and really spotting what I would call the diamond in the rough, finding that donor, that interest that really is worth investing a conversation with a real person as opposed to sending him a piece of direct mail, for example.
The Role of AI in Identifying Major Donors and Optimizing Engagement
[00:06:49] Sam: So, dive deeper on that. you have the software side of your business and also the hard interaction piece; is that more the predictive side of AI that helps you understand where the points of leverage are?
[00:07:02] Michal: So, before even applying AI we would apply a lot of machine learning, which would be high-end modeling based on large volumes of data, discerning trends and basically saying things like, being able to predict if somebody hasn't renewed their membership to their local station, right? We can, with a lot of data, do a pretty good job predicting if you are more likely to renew that membership compared to the person next to you who also has lapsed. And that's based on your giving patterns that are third party data, but you can basically predict if it's worth sending you a piece of direct mail or email or anything like that, if you are likely to respond.
We've done tests on predicting and modeling out how much money should we ask for. You can buy third party data of net worth income, and you can start modeling. Should I ask you for 60 dollars or 120 or a thousand dollars? So, we've done that, and the AI application then is really with major donors specifically, we actually partnered with a great company on this few years ago that enables us to do the research.
So, you basically mount the AI engine on top of a CRM and it's able to scan all of the data that's coming across it and figure out who is the next best one person to talk to. And I would just say this is a really powerful concept, right? Because before that in fundraising, you would do a lot of donor scoring.
You would send data to a third-party research company, and they would say, “Sam is, this is his income, this is net worth, this is the house he lives in.” You can buy all that data. But in the end, you still come back with say, I don't know, 3,000 prospects. How do you talk to 3,000 people?
It's really hard, right? so then, then when you go and look at human resource side of things, an average major gifts officer at any fundraising outfit, I don't care if it's public broadcasting or elsewhere, can handle actively about 125 to 150 relationships, So if I have 3,000 people, but I have one person whose capacity is 120, I need to make the best use of that 120, I need to assign the right people into that profile. That's where AI can be really spectacular. And the tool that we actually use not only helps us do the research and pick the next person, it actually pre-writes the first email that we should send.
So, you as a fundraiser says, “Sam, good morning, you should really talk to Julie today based on all of this data. I have done that research. Julie lives in such and such place. Here's an approximate income. Here is her giving history. And based on that research, you should send her this email…”
And that email says, “hello, Julie, I would love to meet up for coffee. It's going to be warm in the middle of week” – because it checks the weather – “we could meet at a coffee shop at a corner of Main and whatever” – because it checks Julie's address and can figure out that this is probably the closest coffee shop, things like that. And then when you send it off, basically we can prospect through that portfolio a lot faster and get to meaningful conversations with donors that they actually care about having.
[00:10:05] Sam: Our sales teams is exactly the same. You can spend all week making outbound dials. And you have a rough idea, right? Of people in this industry or in this region or at this time of year might want to hear from me with the service that I can provide. But the vast majority of them don't, or they don't pick up or they don't respond. And it's just that sheer amount of time, right? It's the opportunity cost of what could you be doing with those hours that you're spending? Like, where can you be more targeted?
[00:10:40] Michal: The comparative of nonprofits is – I believe I worked at a nonprofit for, 20 years, right before we split this out as a separate company – and as a fundraiser, you have fiduciary responsibility to spend the donor's dollars wisely. If you entrust me your dollars – and it's a great moment when a donor steps up and gives you a check or gives you like, like that donation. Somebody is willing to part with their money because of the cause that you work at and that they believe in. It's a great moment. It really is incredible. I wish everybody could try it.
But my point is that when somebody entrusts me that dollar. I have to think long and hard about how I spend it efficiently to further the mission of the organization, not the administrative burden, if you will. And that's really powerful. And I think that's also why I think donors appreciate when you focus your conversation, when you talk to people that want, they genuinely want to talk to you. As a fundraiser, they're not evading. If they care about the mission, they want to know more about it If you're not bothering them, you are essentially enabling them to give.
[00:11:40] Sam: The efficiency of nonprofits, the idea that you can give a hundred dollars and only 25 of those dollars are going to the cause which you wanted it to go to - and that's a fairly standard amount, right – it’s crazy, it puts off giving in general.
[00:11:59] Michal: Why would you give?
Navigating the Challenges and Opportunities of AI Integration
[00:12:03] Sam: So, I guess you were already playing with some of the tools and the concepts long before ChatGPT came on the scene, but when that did come around and you started to see all these different opportunities where AI could impact various parts of the business.
You said you started out just by playing with it and testing out where it might be able to help. Now you're a bit further along in the journey, still maybe any second inning but have you put more of a framework around how you assess those opportunities and challenges and think about managing the implementation of AI at CDP?
[00:12:38] Michal: Yeah, I would say, the biggest consideration has to be data security. That's where it starts. Because picking the right tool, the right tech stack, making sure that, this is, we're dealing with donor data. That's sacred. That's a trust that you have with the donor. We need to have the highest standard of security at all times. And so you have to be careful about which tools you pick because you don't want the data to be out there exposed in some way. So that's, that to me is like, that's table stakes. That's a consideration we must all have. And I would say that probably at this time is slowing down or preventing us to even go crazy and try things. We have to be super measured about it because again it's the trust with the donor that we have, and we take that really seriously, right? Just as our client stations do. So, I would say that's probably the biggest consideration right now.
Where we're headed with it, if you will, or where it goes, I do think that there will be more tests like making your copy better, creating shorter emails, more to the point, faster testing. I think that all makes sense. I think that the biggest piece that is uncovered for us at this point is we have more data than we know how to use at this stage. Does that make sense?
You essentially run into a human limit pretty quick. And I'll give you an example. Before ChatGPT, before AI, before all of this. We did a test. This is two years ago. I think that we basically had some data that, we, again, research of donors, we identified people that have given money to station – this was with one station, a large station – have given money to a station. We've also checked their streaming pattern that we can track their content consumption on streaming platforms that we work with. So, they were watching a lot of science programming. Great. Then you bring on other third-party data and you say, “Hey, they're donating to science museums and science related things, they're watching science programming. So, it's very likely that I should ask my next ask should be about the work that we're doing in STEM, stations are doing in science, education of kids, et cetera, that will probably resonate.”
That whole process that I just described is a massive labor process because somebody's got to pull the data. Somebody's got to write the email, write the appeal. It's got to make sense. Like that prep is pretty, pretty big, so we did that, and we identified I don't know, 5,000 people that had that trait, and we deployed series of messages.
They were really cool, really well written. And by the way, they did really well, but the lift of that highly targeted message in terms of revenue return and fundraising and all of that did not outperform the amount of expense for human cost that we had in creating all of this. So, it worked better, it was a better experience for the donor, but the cost of doing all of that didn't make sense.
And so that's where I see the next step with AI, right? Helping us in making that lift a lot easier. There's going to be a lot of kind of natural language processing, querying and all of that stuff that kind of is happening, right? Being able to deploy that across the data set and say, “AI, write me an email that engages my science interested people and really talks about all the great content we do around science” and if that can be done that simply, then it starts making sense.
[00:15:51] Sam: Yeah, it gets very meta quite quickly, right? Like how many layers of AI do you need to do the various tasks? But yeah, you can see it, right? That's the interesting thing with AI. Like it's, opened the realm of what's possible.
The Power of Mission for Attracting and Retaining Talent
[00:16:04] Sam: You probably have a much higher than most constituency of employees that are tech savvy, and maybe even computer scientists and people like that. But is that a limiting factor that you see for actually doing some of these things is the expertise either in house or that's available externally to be able to work on some of these problems.
[00:16:27] Michal: I would think so. We have a very tech forward workforce, I would say like everybody on our team is very much focused on innovation and new technology. I almost have to apply the brakes because of like, “Hey, we have to check for data security first. We have to do all these things before we dive into some of these tools.”
But yeah, it is a limiting factor because finding people that are really skilled with data and data architecture that's going to become – it already has become – a very high demand job. That's not easy.
I would say what I absolutely love about the work that we do is the fact that we have a mission and that is how we differentiate as a company. Because you have a lot of people that can work on data in some really large organizations, but it's gonna be a grind. It's gonna be a little different, right? The fact that you can actually then go home and say, “I've actually contributed to education in the United States” or making news better or whatever. The portion of that mission you pick as yours is really powerful. So, I think that gives us the edge of finding really motivated, really talented people that actually also care about doing good in the world as opposed to just doing. So that's important.
But yeah, it's a highly sought after skill at this point and harder and harder to hire and also very hard to discern from all of the offers out there about what's true and what's not. There's so many people come in to approach us, companies with the latest and greatest inventions and snake oils and whatever, and it always all works in a PowerPoint, right? When they show you the PowerPoint, it always works, but there are plenty of failed tests that we've seen in the past.
The Impact of AI on Workforce Dynamics and Skill Development
[00:17:59] Sam: How do you see AI impacting jobs, the workforce? You talked about being a data company and a very tech forward business. How do you think that mindset within the company has impacted the way AI is viewed by your employees? And I guess, are people concerned that it will come and it will take their job and everything's going to be done by a robot?
[00:18:26] Michal: It's possible. you know, I would certainly say that it's jarring, right? When we, for example, we tested a fundraising appeal, having AI write it and having a person write it. And then split the list down the half, send out both and see who does better. By the way, in that one instance, AI actually slightly edged out the writer, but that was just a test.
But it was jarring at first for people on our team to even think that way. Because if you're a writer, if you're somebody who creates these appeals, there's absolutely still a need for your skill and the humanness of it. AI is not at the level where that can be replaced. It definitely is a good assistive tool. And I would say, if people are kind of afraid that AI will take their job, I read this quote recently and I'm going to butcher it totally. The essence of it was that “there's no need to be afraid that AI will take your job, but a person who understands AI certainly will.”
And that I find is a big difference because it's being afraid that AI will take your job. I think that's BS. But sharpening our individual skills in understanding this technology, utilizing it in the right way, responsible way, secure way, and all of that, I think that's a real threat to a job of people that will bury their head in the sand and not want to understand it.
So that's where I see the difference. And that's what I try to communicate to our team. And I don't think that we have a bunch of people that are sitting around going, “Oh my God. Robots are coming. Robots are coming.” I Think that we have the right dose of that curiosity to say, “Hey, I should really explore this. I should test this, but also, let's look at the data. Does it actually have the result intended or is it just smokescreen?”
That's really how we've tried to model our entire company of the hype is great, but in the end, let's let the data guide us. And if it actually can improve our work, the results, the impact on the mission for our clients, then hell yeah, we're in, we're going to try it. If it doesn't, it needs to go.
Data Security: Prioritizing Donor Trust in the Age of AI
[00:20:18] Sam: You said data security is paramount, number one. Are you developing closed system, in-house tools versus “Hey, go and use chat GPT to write this letter to donors”?
[00:20:33] Michal: Oh yeah. Absolutely. And there's a difference between letting AI loose and asking ChatGPT a question about, “give me some copy.” There's not a lot of data security that you need around that because it's just a copy, like, here's the theme that I'm going for. Help me be a better writer” versus, “Hey, I crunched a bunch of donor data.”
That's different. There's a different level of security that we would see. we're working on how we would develop that into our own data lake that we have built and how that will apply. But again, we're taking rather a cautious approach at this stage, just really understanding it first and really making sure that we don't, as I said, that trust with donors around donor data is so high for us that I want everybody on my team to be sufficiently paranoid all the time about data security.
So, that needs to be kind of the first step. As exciting as this technology can be, we've taken a measured approach because we've also tried a few things that just didn't work. So again, just because a salesperson is telling you that it's the greatest thing, it doesn't mean it is.
Forecasting the Future: AI's Expanding Role in Nonprofit Strategy
[00:21:37] Sam: To wrap things up a little bit, how do you see AI, the role of AI evolving in your business? If you were to set your stall out say five years down the line, what goals do you think AI could unlock for CDP? And then, flip side of that question, what are the constraints around that, that are preventing you from maybe fully capitalizing on that potential?
[00:22:03] Michal: Yeah, that's great. I’ll start with the future. One, I would love AI to help us stop making stupid decisions. And I'll qualify this.
If you deploy the data model correctly, or that understanding that intelligence I guarantee you at many nonprofits, including what we do, there is a time when we're going to make a decision where we send a renewal letter to a donor, asking them for a hundred dollars where they really want to give us 50,000 dollars. I would love some intelligence to tap me on the shoulder and say, “Hey, don't send that letter, do this instead, or consider this” that would be freaking awesome.
Because by the way, as any fundraiser, you see this. I've had a donor when I was in Houston, she was a wonderful lady. She was sending us 40 dollars each year and then all of a sudden a check for 750,000 dollars showed up. We were only asking for 40, but she just felt that was the better amount. We had no idea.
So just saying that happens to every nonprofit. And making that right decision and really connecting the donor with the right passion for the organization, that would be really cool. And I think that will take many forms that will take forms in the content that you provide back in the report back on the mission.
It differs between public media and other nonprofits. But if you're supporting a nonprofit that, I don't know digs wells or whatever else. There's so many great causes, how you can report back on that impact and really make the donor connected to the actual mission and put the fundraising in the background. The fundraising should not be in the foreground, right? It should be the mission of the organization. And so, I think that's where AI can help. And I know that's somewhat a broad answer, but I think that if we do it, then that's how it will feel is that you won't even know that it's there. And that's the point.
And I would say right now, what's kind of stopping us is just, which target to shoot at first, there's so many exciting opportunities. You just can't do everything at the same time as a company, there are limited resources, and I would say that's, what's probably the most applying the most breaks, it's just picking your shots and saying, “I want to try this and I want to fully test it.” That's probably the hardest, just limited resources. And I would love to deploy all the tests at once, but once you test everything, you really test nothing. So that I would say is the biggest limiting factor at
Advice for Other Leaders: Lose the Fear
[00:24:29] Sam: That's a big piece of strategy it's as much about what you don't do as what you do. Okay. And then the one final question - if you were giving some advice to a business leader who's just getting started thinking about implementing AI into their business, what would you say?
[00:24:45] Michal: Start slow and be curious. My very first interaction with somebody asked me to write a letter of recommendation. Don't tell him I did that, but I had ChatGPT help me write it and I edited it. And that was like a moment for me because I hate writing. I just don't like it. As a person and just that kind of supplementing my lack of skill, if you will, that was like a powerful breakthrough moment for me.
So, I would just say, start slow and listen to people that are super curious about some of this work. Find allies that have a knack for exploring this. None of us will be able to avoid AI or working with AI. I really believe that it will be part of everything that we do. It already is without us realizing it. The best deployments of AI and machine learning are those where you just don't even know that it's there. And it's happening to us every day. Just check your email, log onto your Gmail. You'll see plenty of examples of data being applied to nudge you in one direction.
So I would just say, lose the fear. It's going to be part of the society, just everything else that we do now. Nobody thinks twice about searching Google anymore. I'm sure that it was scary. The first time people were doing that, what, 10 years ago, how old is Google? Not that old.
[00:25:56] Sam: Yeah, I'm older than Google, so it's not that old.
[00:26:00] Michal: I am too.
[00:26:01] Sam: All right. Awesome. Hey, Michal, thank you so much. I really appreciate you sharing everything with us and good luck with everything you're doing at CDP.
[00:26:09] Michal: Anytime. Thanks for inviting me.