Climbing the Data Maturity Curve: Activating Enrichment, Segmentation, and GenAI

Welcome to Windfall's webinar series. Today, we're gonna be talking about climbing the data maturity curve and actions to take next. To kick us off, wanna give you a sense of who you're talking to today. I'm Kathleen Atkins, head of marketing here at Windfall and joined with our CEO and cofounder, Arup Banerjee. We love to keep these interactive and welcome you to pop questions into that q and a throughout, and, we'll have some polls and really looking forward to, engaging with you throughout today's, topic. And before we dive in, wanted show you a recent new thing that we've come out with. And perfect. So here's a new case study we have around save the children. My assumption is if you're on this webinar today, you'd love to talk a little bit about high net worth fundraising and some of the ways that you can leverage data around that. And you can see the results here are pretty powerful even thinking about database mining, looking at your own constituent base and saying, well, can we actually ten x that gift of a recurring donor? And how we think about major gifts from smaller donors and then exceeding goals for the team the prior year, like, that that's where all these things really pull together, and we have a lot of these great new case studies come out. We'll actually send you links in the chat, to take a look at those. And so we've got the Save the Children, one with University of Michigan, and Muscular Dystrophy Association. We'll pop those links right there for you. And, wanna give you a preview of our deep dive topic for today before I jump into the agenda. So sneak preview here on the next slide is, the data maturity curve for nonprofits that Arup is gonna dig into and the steps that you can take to reach the next level. Perfect. And thanks, Arup, for sharing those case studies. We'll also show you a sneak preview of our Windfall Academy, which can really help with putting this to action as well. So let's jump into today's agenda. We're gonna do a super brief overview of Windfall for those of you who are new to us, just enough to give context of why we're here today and talk a bit about more about why data matters more now than ever, how to actually assess your maturity, and we'll be doing that live on the webinar today, creating processes and workflows to move things forward, and a checklist for as you move into the next steps. And to give you an overview, today is really designed for nonprofit professionals and organizations who wanna use data. Like I mentioned, we're gonna be digging into that maturity curve and how you can describe it to your peers. We're gonna get into the four v's of data and why it's really hard and important in the age of AI. We'll talk through the steps to go up and down the maturity curve and including a real assessment of where you may sit on that curve today as well as a framework, for creating a trusted data driven program along with that checklist. And at the end, we'll have q and a time and a demo of how we can help. What we're not gonna cover today is a deep dive on mechanics or underlying data or algorithms, and any pricing or specifics. We're totally happy to talk with you about that after, so shoot us a note. I'm Kathleen at Windfall. Arup is Arup at Windfall. We'll also put our emails in the chat for you today if you have any questions. So as I mentioned, we love to keep things interactive. So let's go ahead and start with our first poll on why did you join this webinar today. Here we go. Give just another few seconds to get your responses, which are flying and fast. Looks like we're seeing a pretty good mix. Just a couple more moments, and I will go ahead and share this. Okay. Great. So you can see that there's a pretty good spread. Definitely a strong number who wanna separate AI hype from what's actually usable today. That's a topic we hear about everywhere we go, as well as a pretty good blend across the board. So thanks for participating in that. We'll be sure to tune as we go. So that windfall overview that I mentioned, so we'll jump in there. To really start there, we've been around since twenty sixteen, and our vision is to change the way that organizations understand and engage with donors. And, with that, we partner with over fifteen hundred organizations today across the country, supporting nonprofits of all different shapes and sizes. We have a proprietary dataset where we track more than a hundred trillion dollars in total wealth and deterministically match data from households of datasets to create more than a hundred million US households, of which more than twenty million are affluent, which we define an affluent household as one with a net worth of over one million dollars. And that is really the focus of that, high net worth, fundraising that we'll be digging into. And there's really three core ways that our customers leverage our data. There's many, but wanted to highlight, top three main use cases. The first is around identification or prioritization. So using a precise network figure, you can stack rank your database by net worth. Maybe you have a thousand records, ten thousand, hundred thousand. There's only so many hours in the day to go through those, and the prioritization can help you figure out where to spend your time. Second, we help with segmenting your data, which we like to call finding hidden gems. So imagine you get a ten dollar donation. Typically, you would send maybe a thank you and move on. But by overlaying Windfall's data, you could realize that this person has net worth of ten million or a hundred million, and that can help you tune what you ask in your next steps to deepen engagement and stewardship, and and get more results out of out of those hidden gems. And then third is around engagement. So we talked about that net worth figure. In addition to net worth, we're also, providing our twenty five additional triggers, which are things like recent moves, marriage or divorce, liquidity events. Do they own a boat? Do they own multi multiple properties? Knowing more about that person and what they're interested in, philanthropically and how they might be using their free time, you can craft the best opportunity for a really great conversation. And how do we do this? The foundation of that is our data. So transitioning into our core topic of the day, why data matters more now than ever, we'll dig into that, and we'll start by, walking through what we really hear in the market today. So we get a lot of questions, whether it's, APRA prospect development where we were about a week ago in Chicago or on any of our calls. We get questions about, how can how can we use data in our models? Who's gonna actually donate in six months? How do I speed up what we're doing? And most importantly, how and most commonly, how can I use AI and data to accelerate the results? And that's the thing that we hear the most in market today. So to set the stage for that and the data, you probably hear a lot about data decay or steal data or the great transfer of wealth. So I wanna give a little framing to that. Data about people changes constantly. But like I mentioned, people change jobs, get married, have liquidity events. And then when we look at the economy, there's even more movement. So we're in what's called a k shaped economy, meaning the top earners are rapidly gaining wealth, and the lowest earner share is getting smaller. And when we focus on the affluence at the top, there's a shift due to that great transfer of wealth already happening and other economic conditions causing a a look at wealth to just really be a point in time. And if you're too far past that point in time or change, you might be looking at a version of someone that doesn't even exist anymore. So that's a key topic about data and change. And when we really think about the impact of that, the fact is that data decays at twenty five percent every year, which means that data that was right, it becomes wrong fairly quickly. And in the age of AI, what does this really mean? So a lot. We first when I think about data and first party data alone, it can be hard to know who to prioritize. You know who's given before, but you don't know who can give more or who's most likely to say yes right now, and there's gems in there. But the process could still be based on guessing, and you look at a database, and you could see it's hard to know what to look at and where. And third party data can provide that missing lens where you're able to structure and organize and no longer sorting by what someone did last, but you're sorting by the characteristics that make someone more likely to convert or respond, in the way that you're hoping for. And by leveraging data and AI together, we can help isolate and identify those consumers and the donors where we want to spend time. And so on on this one here in particular, I'll I'll step in for talking a little bit more about the rapid change of data. So I think a lot of what was mentioned even in the first poll was how much do we actually have to care about and, like, what is the elements around AI? And so we're based in San Francisco. We have a lot of these conversations pretty pretty frequently. And at the end of the day, why is this a challenge? Well, there's always been a lot of data. We'll talk a little bit more about the evolution of AI over the last, you know, fifty years. But, ultimately, again, it is about the foundational level of, like, what you're looking at. So the four of these data have been around for a while. The first one, though, is around volume. So there's a lot of data. There's a lot of historical data. Just think about your CRM snapshotting information. Every day, you're going to basically replicate that database as frequently as you're updating it. Well, that makes it exponential in the size of growth that you're actually dealing with in terms of analyzing that information. Well, on top of that, even if you look at your CRM, you might have notes in there versus a column that says their net worth and a specific number. So you have different varieties. Right? And a computer, whoever else needs to synthesize that information to understand how do we wanna structure it. And you can take unstructured data as an example of notes and look at sentiment analysis, or you could determine, you know, how long the call was. There's different ways to extract meta information from unstructured sources. Then we think about the speed of data. So Kathleen mentioned a little bit earlier where if your data is incorrect and you're basing models off of it, how do you know that it's gonna come up with the right outcome? So if you do a wealth screening once a year or if you do it once every four years, the world's very different today than where it was in the past. So, ultimately, that velocity actually does impact the ways that we think about AI or just leveraging data holistically. Finally, we have veracity. This is the big one. And, again, probably for folks who are looking at wealth screening for predictive modeling for, you you can guess whatever. Is the data correct? And, in fact, we'll be talking a little bit about this over the next, like, three or four slides in particular. But if you're not sure your data is correct, then how can you actually validate that somebody else's data is correct as well, meaning the underlying email, phone number, etcetera? So this is a really hard challenge. And in today's world, and why most folks are here is saying, well, is this actually something I have to deal with? Because the world is telling me that I should become a data expert and AI expert, but, like, is that really what is happening today? So with that, I'll hand it back over to Kathleen to talk a little bit more about what's happening in your CRM and how you can actually try to look at it in the lens of AI as well. Thanks, Arup. So as Arup walked through with the data, and I mentioned before that we're now squarely in this world of AI, some are using more than others, but it's happening now, and its growth is inevitable. And the effectiveness of AI is only gonna be as good as the data that it sees and is leveraging, and that's why we really focus in on this so much and have a bit of a deep dive to show you on that. So as we take a closer look at this in the prospect research space, tools like ChatGPT, Gemini, Claude, and Perplexity are becoming more common. And instead of going to multiple sources, you might think, why not just ask an AI tool and synthesize everything in one place? And in some cases, these tools can save time or offer insights that you might not have on your own. But as many of you know, these tools also come with a disclaimer, and the results may vary. And you you really need to verify the information, which leads to important questions of what can you really trust. If you run the same search in Gemini and ChatGBT, will it give you the same answer? Will they cite the same sources? What happens if you enter the same prompt twice? Will you get the same results or something different? And these inconsistencies really matter, especially when you're dealing with wealth insights where accuracy and trust are really core to everything. And that's really what the the bottom of it is is that AI is only as good as that data. And it's just like a predictive model where if you feed in bad data, you're gonna get bad results in the kind of garbage in garbage out, way. And, like, the key point here is that these tools don't have access to the insights that they need. So if we were to ask them what does Windfall know, they can't tell you because that's not public. And, that's why we've launched our own AI Copilot as an example, which is designed to help fundraising and research teams leverage AI, but with that trusted right foundation. And wanna show you an example of this in the real world. So I'm gonna put Arup in the spotlight here. So we asked chat GPT about Arup Banerjee, who specifically is Arup a good prospect for nonprofit giving? And Arup Banerjee is actually a fairly unique name, but surprisingly, there's actually fifty or more people named Arup Banerjee in the US. And we what chat GPT returned was actually pretty interesting. It gave us three results based on that simple prompt. However, when we looked closely, none of the results matched the exact spelling of Arup Banerjee's name. And one of those results wasn't even for someone in the US, which raised a question. How confident can we be that that information and the tools that provided is actually about the right person? So if you're a skilled prompt engineer, which means, like, you can refine and rephrase your questions and queries, You might be able to get closer to the right information, and that's exactly what we did. We spent time working to prime and get get some tuned answers. So when we think about why is this happening and what what's really under this umbrella term of AI. It can mean a lot of different things depending on the context. It includes technologies like machine learning, which is a type predictive AI used for things like lead scoring and natural language processing, which is used for translation and classification. However, you're probably most familiar with generative AI, which is the tools like ChatGPT. And AI has a really significant history in the nonprofit, sector evolving from early rudimentary data analysis to the sophisticated tools that are fundamentally changing how fundraising organizations operate. And initially, it was limited to predictive where basic algorithms helped identify donors and predictive giving likelihood, but this early application was a major step forward for manual research. And as technology matured and became more accessible, AI's role expanded. It began to power efficiencies by automating tasks from donor thank you notes and email campaigns to social media management and even helping with grant writing. And with the emergence recently of generative AI, there's really a new era of personalization possible, allowing nonprofits to create unique tailored content for individual donors, and really improve engagement in this really increasingly complex landscape. And when here's an example of looking at that trend. So what you're looking at here is called a Google engram view, which is really the frequency a phrase has been used in publications over time. So it shows how AI has been growing in time since, like, nineteen sixties or seventies, where we can think of AI has been around as a broad term for that long. As an example, Excel's goal seek function was considered AI when it came out because it could solve a problem, and machine learning and deep learning are subsets of this with Gen AI built on top of deep learning. So today, with powered of power powerful generative AI, so much more is possible. And with that, Arup is going to dig in a little bit more to the different types of AI. Yeah. Thanks thanks, Kathleen. So I think Kathleen mentioned this. And so for those folks that are out there and you're thinking about, like, AI, can you be as an expert, or is there, you know, what what's what's real here per per our first poll question? What Kathleen just mentioned with Excel being AI, and you're saying there, you're like, no. It's not. Well, actually, the true definition of it is just a program that can sense, reason, act, and adapt. That's just the simple definition of AI. Everything else that we take a look at is a subset of that. Thus, when we look at GoalSeek and it was interesting because I had a teammate, who have never used Excel and never used that function. But GoalSeek, if you, remember it, it's just brute force where, effectively, you're taking an equation and you're saying, hey. Find me the right value for that. And Excel it goes through and does every single iteration it possibly can in order to find you that right answer. Well, that's AI. Right? The the computer is actually trying to help you solve that problem without you interfering with it. Now a lot of the world that we live in today has been in these two considerations, machine learning and deep learning. So machine learning, again, is predictive AI. We're gonna talk a little bit about how we have an academy on that course. But generative AI really takes a look even further beyond that, which it's looking at creating content effectively from a lot of the artificial intelligence that's learning. Now when we come into this world, there's been a lot of webinars from folks in the industries around, like, what is their solution, how do they fit in, things of that nature. We're not necessarily gonna talk about how we necessarily fit in, but we'll talk a little bit more about agentic AI versus generative AI and, like, just making sure we're all speaking the same language on the on the call today. The first thing is Gen AI. As we take a look at it, I just mentioned, it generates content. So that can be text. Those could be images. It could be documents. It could be a lot of different things, but it create it's created from a prompt. And Kathleen mentioned that. So you're entering something into your computer, and it's a one shot prompt potentially, or you're engaging with that chat. Right? So that's generative AI. It's like a chat orientation associated with it. Now you have AgenTik AI, which then effectively takes that and puts in to get together different tools. So you might prompt something, and it calls different functions from different places. It will get that data from Windfall. It will then get your CRM information, then it constructs this lovely, you know, email outreach program for you. Well, that's more agentic AI versus what we think about as agents. Agents are actually combining all of what I just mentioned historically, but they're accomplishing the tax by themselves. So it's not a human being clicking a button, but it's that actual agent connecting the dots and saying, hey. I think I have enough intelligence. Let me go ahead and go through the next thing. Now most programs today or most SaaS offerings today offer a Genetic AI, not necessarily the agents themselves. And we have to be somewhat cautious around the agents that you deploy with guardrails. Imagine that it's a new employee at your company, and it's issuing payroll, right, as an example. You might not necessarily trust an AI agent to write payroll for you. He might wanna double check that, and we'll talk a little bit more about where AI and some of these things come into place. But humans are still the primary judge in any of these factors whatsoever. Okay. Cool. So with that, I wanna talk a little bit more around, you know, specifically where we're are at on the data maturity curve. So let's go ahead, and we'll launch our next poll here. And by the way, this is around where your system is set up to enable what I just mentioned a little bit earlier. So how often are you currently creating data in your workflows today? So we'll go ahead and launch this poll. It's a multiple select, so you can figure out if you do all of them, feel free to do so. But this goes into pulling lists, segmenting on history, screening for wealth on a consistent basis. Like, again, not necessarily AI, but specifically, you know, the technology workloads that you've adapted as an organization instead. So we'll we'll wait a couple of moments as folks are are filling this in. So I will share this, but as as the results are coming in, I'm seeing we segment on giving history. We screen on a set cadence is, like, number two, and we pull list by hand when someone asks. Okay. Cool. So those are the top three at the moment with segment on giving history and pulling list by hand as being, some of the workflows. So I'll share those results. When I think about a workflow of pulling list by hand, I would go ahead and claim that's not a workflow. Right? Like, that's a task that you do. But if you know you're consistently getting it, is that actually a workflow you can set up within the CRM? Reports, things of that nature? Segmenting on giving history, screening for wealth, those two are probably leading to number one. Well, let's pull a list on everybody who's given a lot who has a high wealth score. Again, as we think about those as workflows, just, assume where you're at on this data maturity curve as well. So we think about this as being three different phases. Kathleen showed it at the very beginning of her presentation today, but just think about these three buckets of, like, where you're at, and we're gonna ask you in a couple of slides where you follow on it. So remember, there's business value that's derived from maturity. Maturity does not necessarily mean how long you've been in business. In fact, you can go up the scale as quickly as you want would like. But to start with it, it's gonna be number one, which is establishment. So think about this foundational level. If you had no donors, no constituents, you have no data available, that's where you're really starting at the bottom. Generally speaking, we are then gonna segment on transactions. So when we say segment on giving history and you're doing that today or you're doing wealth screening, you're still in this establishment zone, right, of being able to, think about the next phase, which is really driving results. So when we think about driving results, this is really thinking about segmentation on other attributes, not just giving, and then that's transactions, but thinking about portfolio allocation. Are you doing direct mail? Are there other analytics? Like, where are you winning? Where are you not? Have you started to build propensity models or predictive AI to help you address this problem? Then we go into the third category, which is orchestration. And, again, I'll go through this again, and probably a little bit more slowly to speak about examples on this. But this is where you go to orchestration, which is now the beautiful part of being able to optimize even further and really start to hone in on the more sophisticated questions that you have. Now we will mention that we do have a, academy course, what's called predictive AI boot camp. And we have this so that all organizations not only can understand how these models are built, but they're able to communicate them back to their peers. What's really important around all of this, especially in today's world, is that it feels like a black box, but it's really not. A lot of it has to deal with what is the limitations of data, and then what data are you feeding into the equation in order to get the output. So let's go slowly, on each one of these phases for the next three slides, And then, therefore, you're gonna be able to really start thinking a little bit more about where you sit on this data maturity curve. So my hope is that most folks here are either on the upper end of this establishment curve or at least thinking about how to move over to the drive curve in particular. So, again, if you have no data or bad data and people thought about this a little bit earlier. So, as an example, do you know that Arup Banerjee is married to Elizabeth Banerjee? Well, to a certain degree, that's household information versus individual information. But that's still important here today of saying, well, can I look that up and really understand, are they the right types of people? Now this doesn't necessarily mean looking somebody up outside of your database, but somebody within the database itself. So as I just mentioned with Roop and Elizabeth, how do we unify that? If I know they have the same address, but it looks like two different people, am I looking at address consolidation, email consolidation, LinkedIn URLs? How do I think about my join key of unifying these in together? Do you want individuals or households as part of your outreach? Because on digital, if you decide to send emails, you might wanna have that as a one to one relationship, but direct mail being a one to many relationship, meaning that you're gonna mail everybody at that household instead. Okay? Well, for segmentation, a lot of y'all, eighty one percent said you segment off past givings. Well, if we're looking at live buns or side buns, how do you do this really quickly without having to make a request or validate the information or determine, you know, whether or not they are in a different category as well. Finally, as we thought about third party data, most folks here are doing wealth screenings on a set cadence. But have you really done it on a set cadence? Like, what is the likelihood that somebody changes their wealth in the next three months? What's the likelihood of even inbound request or somebody who has a prospect who you're about to go meet with, but you knew wealth screening? Well, how do you know to put that in a portfolio? How do you know is that the right person holistically for your organization? So all of that is to say, great. You are here if you can look at somebody, but you're not necessarily gaining all of the driving forces from that, which then goes into our next phase here in particular. So it's saying, do we actually look at this data on a consistent basis and start to operate with more data in our disposal? And so we think about portfolio allocation. This is interesting. Right? Why does it come with a portfolio gift officer? Is that plan giving? Is that annual giving? Is it major gift officer? Well, can they be split? Why would that be the case? Remember I said that many of these folks probably could fit into different functions. Well, therefore, if you have a cadence outreach, which one is it? And how are you making sure that it's hitting and that it has the right context for those folks? We just mentioned households versus person because, again, if my wife gets a different message than I do, is that what you really want, or do you want them you know, your constituents to get the same message no matter what? By the way, when we have that once every three year screening, well, your competitors, other nonprofits might be doing that more frequently. So how do you make sure to automate this and get those alerts from either new constituents or existing ones who had wealth creation events, wealth destruction events, anything that you might wanna think from an outreach perspective. Well, everything I just mentioned was potentially manual workflows. Now to a certain degree, manual workflows can be automated with a machine. That does not necessarily mean it's AI, but it does mention that you are automating and simplifying your life. Well, the final component is propensity models. Propensity models is an AI score. What is the likelihood that that person would do x for your organization? That x is defined by you in particular. We'll get into that in a moment. But generally speaking, AI in here will help you route people or be your air traffic controller around anywhere where you wanna route these folks. So to this point, we're doing it consistently. We're now starting to set up their create or create the right workflows to orchestrate what we're doing, which is reporting. Hey. The elements that we're doing for outreaching is very helpful every single time. We are now engaging with those folks when we know that there's a lapse or that they've raised their hands appropriately. We're now understanding, does that have any throughput to gifts or donations? And can we report on those as well based off of different outreach categories? Well, your major gift officer now is unified because we have that review on additional data points. It's not because they had just a great portfolio versus somebody that didn't. We can actually evaluate that in real time. And then finally, what else can you do with additional modeling? An annual fund, again, plan giving, direct mail outreach. There's a lot of different things out here, and how do you orient and augment it for these respective orchestration layers? The key element here is that AI helps set up the workflows, but you as human being are there for judgment. AI will not create whatever I just said out loud. In fact, the amount of steps I just said today, if you spent probably twenty or thirty hours with Gemini, with ChatGPT, with Claude, you still wouldn't be able to do all of the elements here because humans still need to be in the loop. So to answer that question, there's still a lot of garbage out there. Right? There's still a lot of things that you have to do in order to make the AI agents or these AI workflows incredibly valuable. But if you do it here, it's because that we're identify identifying data that is changing rapidly, not because that you're noticing that somebody's yelling at you about it, which is what AI is really good for. You're being more proactive in this workflow than reactive, which is the best part. So I mentioned that I was gonna go through that slowly and make you think about it for a moment because where are you on today's, maturity curve? And I will pull it back up in a second, just so that we're all looking at the same, materials. But, really, I wanna make sure that we understand that so that I can tell a little bit more about areas where, our team has helped, folks move up the maturity curve. So let me go ahead, and I'll pull that up just so that folks can see where they're at. Establish, drive, orchestrate. And if you sit in multiple phases, like, if I was with my team here, I would probably say, hey. That's a cop out answer. You're probably in one or the other to a certain degree even though there are throughput. So pick the bottom one of what you're looking at in particular. If you're doing some prescriptive marketing, but you still haven't done any portfolio allocation, you're probably still in the establishment or drive consideration, or you're less on orchestration. Cool. Okay. So we got, some responses here. Let's make sure everybody can, answer. Cool. As I go through that, I'm gonna animate again, so everybody can see me go through that really quickly. So I'm gonna go through the five steps to advance your maturity. Now these things, again, do not have to be really time intensive. What's really important around the data maturity curve is that it can be done quickly, and an organization that's within one year of existence might have a higher likelihood of going up the maturity curve than somebody who's been in existence for fifty years where change management is actually a thing. Okay. Cool. And then the last thing that I'll mention as we go through this is I'll do a quick demo of our application, just to showcase, you know, some of these things that I'm telling you today can be do done even faster with automation. So we go through this in terms of a data driven program. We wanna have foundational, trustworthy data, better intelligence, better insights, then start to use AI. Finally, we have to activate this data, and then you iterate. So this entire process of a data maturity curve does not stop with that line being in orchestration. It really means that you have to iterate over and over again. So what do we start with? I'm gonna start with something that I like to do. But, like, again, as everybody here, I think eighty one percent of the folks in this webinar said that they looked at giving data. Okay? You look at giving data. Well, where are they on the wealth curve, and how does this actually materialize? Well, I'm showing you on the left hand side is a box and whisker plot. What that mean is that we're not showing the bottom, but we're showing you the median, and then we're showing you an order of magnitude. So if you remember statistics, right, one standard deviation away and trying to determine well, for somebody that's worth a million and two and half million dollars, the top twenty five percent give two hundred k two hundred dollars, excuse me, or more, but then the median amount is probably closer to the fifty dollar level. Let's look at two and a half million dollars. It's actually slightly less even though these folks are wealthier. This nonprofit happens to do a lot better as they go up in wealth. But even at twenty million dollars plus, let's look at the median value. It's probably closer to, like, five hundred dollars. There's opportunity as well because as we look at penetration across these different segments, we have high penetration on both of these on the left hand side, but can you drive this higher? Wait. For the twenty million dollar plus pool, I only have forty percent. So this is only capturing a fragment of where I potentially could have that opportunity. Now remember, these are constituents in your database. I think about this a little bit differently of saying, hey. Well, how do we know that they're gonna give or not? To a certain degree, they're already in your database, and they've raised their hand of saying that they like your affinity, they like your cause, they want your institution, they have some relationship with you one way or another. So either way, having this foundation really helps us understand the next slide, which is really, well, how do you look at wealth and donations? A lot of y'all talked about the segmentation. Well, we hear about this every day. Folks are looking at this as a two by two where they say, well, great. There's low wealth, low donation history. Let's tend to ignore them. Low wealth, high donation history. Okay. Well, I'm not gonna get any more out of them. Well, high wealth, high donation history, this is where most people sit today. But, like, as Kathleen mentioned, with the great transfer of wealth, with the k shape economy, what we're really looking for are high wealth, low donation history donors. And so, ultimately, if you think about this as being hidden gems on that prior slide where we take a look at wealth and donation, how can you start to translate that up to, hey. They've been donating a hundred dollars for the last three years. We know that they're wealthy. They're a hidden gem. Maybe I can get ten thousand dollars from them. That's a conversation to start. It might not mean you're doing it tomorrow, but it's at least a conversation to start. Well, if you have a large database, how do you do that at scale? Oftentimes, we know that wealth is not the only qualification. So on that maturity curve, wealth screening was on the establishment, but predictive modeling was on the drive. If we look at that AI, predictive AI, can be used to say, well, I have a lot of constituents, and how do I actually want to think about this more appropriately for those folks that might wealthy but have no intent to give versus those that are wealthy. And we think that based off of a multitude of factors, they have the ability to actually contribute to my organization. So let's take a look at two people in particular on the right hand side, somebody who's worth twelve million versus somebody who's worth five. They both donate to nonprofits. The one on the right is a boat owner, one on the left is not. Again, we just mentioned that they're both philanthropic. Hey. There's multiproperty owners on the left hand side. They're both qualified in terms of our wealth screening. The problem is and it's not necessarily a problem, but we're gonna score the one on the right hand side as a higher likelihood to give versus somebody who is on the left hand side. Now this might be a ten thousand dollar gift this year, fifty thousand dollar gift, but it's based off of the factors that we see within your organization specifically. So that's where we think about predictive models. Predictive models for us are really around classification models. If you heard about random forest, gradient boosted trees, neural nets, like, lot of the work that we're doing and by the way, we have our predictive academy class on this as well. It's saying, we're trying to figure out for your organization what is the likelihood to give. So this means that the dollar threshold, the giving threshold, your organization is specifically customized. The output is a score of zero to ninety nine so that you can determine where you wanna sit and send these folks back. So we're updating these every single week the way that you would see your wealth information update every single week as well. So you can think about portfolio allocation a little bit differently. So it helps with the maturity curve of being able to do this and then start to assign folks appropriately. Now all of this has to be evaluated. You have to think about the strength of the model that you're actually pushing out. So on the right hand side here, we have something that's called the rock curve, and we're taking a look at how good the model is. This is a propensity model. If you did a coin flip, right, you'd be up and down this dotted line. Sometimes yes, sometimes no. What we're trying to do is maximize the likelihood that we can identify those major gift prospects, annual gift prospects, plan, giving prospects. All of those folks are in this rocker. So if you're not doing this internally, you should be asking your vendor or you should be asking us what is the strength of the model. Finally, as we look at predictive modeling, we have things like lifting or gain charts that support the validity of this AI actually happening. So a lot of folks said, hey. What is gonna happen? Well, we look at backwards looking tests. If you developed a model and your top twenty percent of your scores contribute to eighty percent of your donors, you've done a really nice job because, otherwise, it'd be twenty twenty. Right? You would have twenty percent for twenty percent of the donations. That's the coin flip as we talked about. Another way of looking at that gain chart is looking at the lift associated with it, which is basically an inversion of this graph. So you're saying after that top twenty percent, I'm still getting a four x lift, which, again, is eighty percent of what I wanna be able to get to. And I only have so many time, or so much time in the day, so this is really how you're optimizing your time, energy, and effort. So we'll put this here as kind of the way that we look at Windfall before I get into a quick demo. If you have any questions, please use the q and a function. But the way we look at this is you can submit a lot of different data points to Windfall. We're gonna be able to take a look at that, screen it for wealth, screen it for their career, help you with portfolio allocation or donor segmentation. We'll build our predictive AIs on this, and then we'll also determine other use cases, win back campaigns, generative AI. We can integrate into a lot of different places for y'all, but what's really important is being able to activate that data in those workflows that we're talking about. If you just have data and you're screening it and it's sitting on the side and you're pulling lists, my assumption is that you're actually not as data driven as you think you are. So that's not a that's not a cop out to you, but it's saying, how do we do that more quickly? So we'll be demoing this for about five minutes right now. As I mentioned, if you do have any questions, please feel free to put that into the chat. But this is a quick demo of our Windfall application. And the reason that I wanna showcase it to you is because we have a host of different ways to take a look at all the things I just mentioned a little bit earlier. And so the first thing that we like to showcase is that here, you're on the Windfall homepage where you have an overall match rate around forty five percent affluent. Now Kathleen mentioned at the very beginning, over a million dollars in net worth. In fact, we show you how many people or constituents are in there with your total wealth. So this is a larger database, three hundred and forty two billion dollars of wealth, and we'll show you that we match to these folks correctly or not. In fact, we'll also give you the different buckets of the affluence even if we're matching the people that are under a million dollars in net worth. We'll give you some additional data points here, but, realistically, what we want you to be able to do is come in to discovery side. The discovery element enables you to actually leverage AI to create different segments or cohorts. Imagine that Windfall can now screen all of your constituents, all of your fundraisings, and we can come in here, and we can start to generate with AI the different segments that you should be going after. Either year end donors, we have something coming up for Giving Tuesday in a couple weeks from now, that's there, promotions, companies that have changed, you actually can go in here and actually see the segmentation and add it to your account. In addition, right, we can start to create the segments on the fly ourselves. So creating the segments, you might say, hey. Instead of household targeting, that's for direct mail, I just wanna target all the people in my CRM instead. Well, I come in here, and then I say, great. I this would be your data, and you could come here and say, I'm looking at everybody who is excludes any of those appeals or, for example, includes anybody with no appeal category. So we could put in there. We could say, great. I wanna take a look at a net worth greater than or equal to two million dollars. I'm looking at folks, as an example, that live in in New York, But in fact, I know that I'm looking for multi property owners that are more vacation owners. So that includes Florida. It could include Colorado, Arizona, and now these are effectively snowbirds that we know are coming from New York to Florida, Colorado, Arizona. Greater than two million dollars in net worth that I haven't reached out to beforehand. Now you can duplicate this group. You can add to it. But on the next step here, we're actually gonna go through in suppressing your donors. Now these are no donors already. And so I don't have to suppress anybody at all. And then I can come in here and say, you know, greater than two million snowbirds in New York. And then I click generate segment. Now what this is gonna do is it's gonna actually say we're gonna start calculating this in particular, but let's take a look at some of the other elements that we had in particular as well. So we have something here called lapsed donors. So you can always click this button here saying they are philanthropic. They have a network greater than two point five million, and their states are in California. They're lapsed. So let's take a look at those folks in particular so we can view the insights here. And what you can see on this page is that we're gonna zoom in to their specific counties. Now this is where Windfall believes that these folks have their primary residence, which is another another cool feature even if you have them in a different location. Now you can scroll down here, see their net worth density, the average net worth, median, or you could see some of the other attributes. Now there's two things that we can do here. One is we can export this data. You have forty two hundred records. You wanna take a look at them for email marketing as an example instead. So we'll say, great. We're going after major gift outreach. We don't need all the data. We send this, or we could set a recurring schedule. If anybody comes in here every month, let's email them. Cool. Without doing that, that's now a workflow that we've set up. But if we just want this, we click next, and then we can send it to our SFTP. Now you might go, cool. I'm not doing that for email marketing. I actually just wanna see who's in this segment. Well, that's why we have this view profiles page. And specifically here on the profiles, you'll see that it now populated everybody who's in that specific segment. In fact, here's the segment name here in particular, and this is all the data that's based off of stuff that you have within your CRM. Meaning, here's all the total donations. Here's all the information that you also have, phones and emails. This is all dummy data, by the way, so it's not real information. But if you clicked into this, now I could say, great. I'm gonna take a look at somebody named Avery because they have a low donation count but high net worth. So let me go ahead and see their full profile. As I go through it, I can see, hey. Here's the windfall information. Here's their wealth. Here's some of the other attributes. In addition, we can see all their historical donations from your data. Again, this is coming from your database. And then the final component is we can actually generate a dossier. So this is dummy data. It's not necessarily gonna come up with the right consideration, but imagine that Avery was in your database. Now for your frontline gift officer or even for your executive team who might, be visiting folks or thinking about a gala or gala, you can start to put in here different ways. And in fact, as we're waiting, you can see that the segment for two million plus snowbirds in New York is available. So while you're doing a lot of the work here, you can start to assess where there might be some different ways of looking at generative AI versus looking at your predictive AI. So with Avery, seventy four million dollars. Remember, she only gave six hundred and fifty. So depending on the AI score, we're actually showing you that this is a low one. We might not wanna spend our time with Avery even if we're trying to summarize. Now that being said, we'll give you some recommendations or outreach. You can give us some feedback on this so that it can continuously iterate on this moving forward. So with all of this, it's very printer friendly. Again, if you're on the road, you can start to do this as well. But the goal behind this is saying, how do I become data driven while I'm segmenting on donations? I'm using my AI model. I'm using generative AI approaches, and I can start to put this into to progressive workflows. That's really the secret behind a lot of the things we talked about today. Now I didn't go into the rest of the settings here within Windfall or some of the applications, but for the purpose of showcasing you some of the strength of being able to automate that in a matter of moments, you can do that with the right tools. So with that, I'm gonna, go back to the presentation here, where we, simply want to, go through the final thing, which is around our checklist, around do you screen your database fully? Do you prioritize? Do you segment on triggers? Sounds like a lot of folks are doing that. But then the next step here is saying for machine learning, do you have the data where you understand the goal? You understand that your underlying data, even though the elements that you think might be incorrect, at least flagging those, and that you're gonna refresh that on an ongoing basis. If you do all of this, even without windfall, you're ready to really make this going up the data driven maturity curve. And, again, this can be done over the course of the next couple weeks. It does not necessarily take months, quarters, or years to do this. It just makes sure that you are organized and thoughtful about the process.