Why Wealth Data is Really Hard Even With AI 

why wealth data is really hard even in the age of AI, and it's when it's becoming even more accessible. So, appreciate your attendance. And with that, we will get going. So, again, welcome. This is our topic for today, really jumping into why wealth data is really hard to understand and map on households in the United States and Windfall's approach to that. But before we do jump in, I do wanna make sure we cover a few housekeeping items. Firstly, we want to encourage you to submit questions, using the q and a tray that you'll find at the bottom of Zoom. We don't have the chat open, but the q and a will really allow us to ensure we're addressing questions as we go. We'll either address them within, our presentation today, or we'll answer them within that q and a functionality. And we also have some time at the end where we'll go through and make sure those questions are answered. We will be sharing the recording with folks after the webinar, and we'll be sending that out via email. In addition, we'll be launching a few polls throughout our time together, and so we'd really love and appreciate your participation in those just to give us more information on who's in the room with us and and making sure we can cater this conversation to that. To start with some introductions, my name is Meg. I'm a customer success manager here at Windfall, and I have the privilege of working directly with our customers, helping them leverage their data. Prior to my time at Windfall, I spent over seven years working in the nonprofit world, both in fundraising and programmatic roles. So I truly enjoy working with our many partners and hearing the ways they experience wins with their data driven workflows. And with that, I'll have my colleague here, Zach, introduce himself. Hi, everyone. I'm excited to be here today. I'm an account executive here at Windfall and work with numerous organizations in terms of how to use wealth data and prioritize the right donors, and I'm excited to dive into today's topic with you all. Great. So today, we are gonna launch our first poll, and you'll see the question on the screen. Why did you join this webinar? And I'm gonna launch that right now, so you should see the pop up. And you should see some options on your screen, whether you've joined today to learn how to improve your data foundation, ideas for improving your prospecting or segmentation strategy, or if you're looking for more information on how to scale with AI, strategies to make your data actionable, or learning more about Windfall itself. We've got some answers coming in. Appreciate those who are participating. We'll leave it open for about five to ten more seconds here. Alright. I'm gonna end our poll, but it looks like we have almost a majority of folks joining today, for two reasons, the most popular being ideas for improving prospecting and segmentation and getting strategies to make data actionable. So I'm gonna go ahead and end our poll, and I'll share the results for a moment so you can see. But some with this multiselect option, I see there's some some choices across the board. So that's great. We are gonna be talking through about some of the data points that Windfall has, and making sure you understand, you know, how you can best be leveraging the data that you have accessible to you and also giving some context for the current, fundraising and philanthropic landscape. So hoping that you can take some actionable strategies from that today. So to set some expectations here, we have a quick view of what you will and won't learn today. Firstly, we're gonna be talking briefly about our company and solutions overview in the first section of our our time today. And we're gonna go into a little bit about the landscape of fundraising, specifically thinking through the great wealth transfer as well as the rise in DAF giving. And we're also gonna be covering the main topic, why wealth data is hard and some of the key issues facing data and AI solutions today. Some things we'll specifically not be covering is details around Windfall's dataset and how we approach wealth. And we'll also not be discussing how our data can impact your specific organization or the pricing of it. And so those are discussions we are happy to have after the webinar, and you can reach out to either myself or Zach to continue those discussions if you'd like. For a quick agenda and how we'll be covering and going through our content today, we're gonna provide that overview, as I mentioned, jump into the great wealth transfer in that's happening here in the United States, which provides really a foundation for the complexity of understanding wealth data as it moves throughout different generations. Tracking those assets is really hard to do. And then we'll talk a lot about why wealth data is really hard and some of the nuances and consumer data in the current market, before we wrap up with our q and a. And we're quickly back into another poll here. So we would like to just have some additional context again on, what you're coming into with the experience that you're coming into with wealth screening. So I'm gonna go ahead and launch our second poll. You'll see the options on the screen here or on the pop up window around have you used Well Screening vendor before? Have you been using it regularly? Maybe screening yearly. Maybe you've used it at a different organization, but your current organization doesn't have it available. So just appreciate taking a few seconds here to submit whatever choice is most representative of your current situation. Looks like we've got a lot of responses already, so I won't leave it open too much longer. But just from the trend I'm seeing, a lot of folks are saying, We're currently using a loss screening provider regularly, which is great. Got a lot of folks really invested in loss screening, and our topic today will really cover around, like, why that's so important, especially in the landscape we're seeing. So I'm gonna end the poll, share the results so that you can see, and we'll continue. So to jump into our first portion here, just thank you, Juan, for sharing your responses and your experience with well screening, but that really does lead us into our next section walking through the quick overview of Windfall. And so Windfall's core focus here and our mission is to change the way that organizations understand and, engage their donors. And we help data driven nonprofit organizations of all sizes, as you'll see in a minute here. And we really do wanna help development teams to find their highest value donors and prospects so you can prioritize workflows and make the most of limited resources. And we do that by providing best in class accurate wealth data back to our customers. We have the privilege of partnering with over fifteen hundred nonprofit organizations across the country today, and you can see a few of them here. They range from all different avenues of the nonprofit ecosystem, whether that's higher ed, health care, large national or international nonprofits, arts organizations. There's ways for all types of organizations no matter the size of their database and no matter their level of sophistication to be more efficient and strategic in their fundraising. And here at Winfell, we're a data company, and we own in our own data, which allows us to do a lot of really unique things, like calculate or model a net worth figure that you can, be confident in and act upon. And so here, our customers primary primarily utilize Windfall, for these three use cases here that I'm gonna outline for us. The first is utilizing that precise net worth figure that we provide back to stack rank your constituent base, your whole database to really understand the highest value prospects and donors in terms of net worth, and how you can ensure that you're putting the right folks in portfolios or paying attention to those people who can potentially make a transformational gift to your organization. The second main use case here that our customers find with the data is that it can help you segment and specifically find what we call hidden gems. And this is where a lot of our customers have really exciting early wins when they start working with us. By layering in your first party data along with windfalls data, you can uncover insights into opportunities that you may not have had on your radar. So look at net worth in conjunction with how existing donors are behaving. Identify opportunities to uplevel folks. Maybe you find donors that you've never had in a designated portfolio or even had a conversation with, Or you find someone that's been giving fifty dollars annually with a net worth of ten million dollars. They're already donating to your organization. It's just changing up the conversation and hopefully hopefully seeing some nice wins in the short term. And the third way that customers are really seeing, high use use case for the data is through engagement. We've highlighted net worth already, but we provide back upwards of fifty other data points in your constituent and households that can be useful in terms of helping to craft messaging that's really gonna resonate based on what we know these households are interested in, based on recent events that they've experienced, based on if they're showing signs of liquidity even. So all of these data points put together can really help you segment, send targeted messaging, and take a very curated approach to outreach to your constituents and donors, right, having that personalization factor. So talking a little bit more about the actual offerings in our platform, it really is built for data driven development, and the cornerstone of everything is wealth screening. So that's comparing your data against Windfall's proprietary people graph and providing back confident matches enriched with constituent wealth data directly to you. And we can also enhance this with career intelligence data. That's another twenty five, twenty six data points that talk about someone's interests in firmographic relationships and career trajectory here. On top of that, we're able to leverage our data science team and machine learning to build custom propensity models that can really help our customers go even beyond who's just affluent or maybe broadly interested in similar causes to yours and really hone in on who's gonna be most likely to make a very specific type of gift to your organization within your set time frame. And so that's where we create models blended on your first party data as well as Windfall's rich data from our people graph. And the last offering here is data link. So it allows us to link records across multiple databases or data silos to help customers understand what is the three sixty view of how a constituent is engaging with them and making sure that as you're speaking to your customers, your donors, you're really speaking through the lens of showing them that you know them. You know all of the ways they're engaging with your organization, not just through one singular lens. And I'll talk about this more on the next slide, but pricing, if you know all of our subscriptions, is unlimited. It's not based on records or credits. It really is an all you can eat type of model, so to speak. And while we provide a lot of wealth signals and wealth data points, this is our premier package data points, which contains career intelligence information that you can see here. And we've also enhanced this recently with two really exciting new triggers that our customers have been asking for. So now we actually have a trigger that showcases if a household is affiliated with a donor advised fund, and we'll, again, jump more into daft giving trends later on. But this has become highly useful and highly sought after from our customers, and we're gonna be talking a little bit more about it later today. And so as you can imagine, there's a lot of major gift implications for understanding who is even sophisticated enough to already be giving through a DAF as well as that interest in a in crypto, which is our secondary new newest trigger that we released. And so this can really just flag together who may be a little bit more financially savvy, potentially younger with that crypto interest, and having, a slightly different risk tolerance there. And we're gonna go we're not gonna go too deep into everything you can do with these triggers, but, again, you can see that something might spark your interest. And if you wanna continue that conversation further, you can let us know, and we'll be happy to follow-up. And so getting that full view of a constituent is really, really helpful because you're probably and you're definitely gonna have your first party data, right, where you know their names, where they're, at the school, the last gift amount, how they've engaged with your organization. And then you can layer in those wealth insights with net worth and if they're philanthropic to other organizations. This is great data to have, but you can really round out that constituent profile by having insights into where they work and what they're doing in their career. Are they an executive with a lot of influence and affluence at a top company? Does this chick check out from, like, a net worth perspective? Or maybe they're an owner of a small business, which is a great opportunity for event sponsorship. There are so many other use cases and things you can layer in when you have access to both wealth insights and career insights on top of your first party data. And getting full insights into your constituency is important. Before Windfall arrived on the scene, the way that wealth screening used to work is that usually you were confined by credits or only doing screenings annually or even every three years. And so it's while most of the folks in this room are very used to, screening regularly, you still, in the past, may have had to be very judicious about which records you were going to screen. And so sometimes maybe you would screen only your existing major donors or people who had given, for example, above a certain threshold, maybe a thousand dollars or more, or maybe only new donors, and or maybe only people your gift officers thought were wealthy based on other signals they were seeing, like the car they drive or the kind of handbag they carry. And whatever way you're picking and choosing, you're gonna miss out on some folks that are affluent. You might get some you know, make some progress with the information that you do have, but you'll definitely be missing out on folks that you opted not to screen or not to prioritize that are affluent and could make significant contributions to your organization. And so we encourage our customers to screen utilizing our unlimited model. You don't have to burn time picking and choosing who to screen or figuring it out. You can just screen everyone in your database, everyone on the file, get all of the data insights back like we talked about. Stack rank your database and start to understand who has the ability to make the largest, most significant contributions and start there as you prioritize. Okay. So that was a quick overview of Windfall. And what we wanna shift into now is the great transfer of wealth that's happening in the United States and how your organization can start to think about that. We have some data here to help illustrate this trend and this transfer. On the left, what we're seeing is the total US wealth held by a generation in terms of absolute dollars. So you can see here that the silent, especially the baby boomer generations, are holding over a hundred trillion dollars in wealth currently in the United States. If you look on the right here, this is the total US wealth held by a generation. This is just a percent of the total amount of wealth in the United States. So not absolute dollars, but just a percent of the total. And as you can see here, that baby boomer generation and the silent generation are holding around seventy six, seventy seven percent of the wealth in the United States, which is a ton of money. And because of where we're at in terms of timing, these things are about to change. What we're seeing right now in our data is that there's a really historic transfer of wealth underway, and this image here really, depicts that flow. By twenty forty eight or twenty fifty, we're gonna see a lot of wealth transfer, especially from the baby boomer generation down through the Gen X generation and even down into the millennials. So you're seeing that thirty nine trillion dollars, but of that total transfer, what we actually see here is the eighteen point four trillion dollars of wealth that is expected to go to charitable causes and philanthropy. This presents a really historic opportunity for nonprofits in the United States to really engage those next generation donors as this massive monetary shift occurs between generations. And to illustrate this transfer even further, you can see that eighteen trillion is going to be going to charity. Focusing on planned giving becomes a foundational strategy. But with the wealthiest households driving half of the wealth transfer, your major donor strategy also becomes essential for capturing some of this philanthropic giving. Nearly forty trillion of that will be left to spouses, widowed women is what we're typically seeing, and next generation donors are gonna be playing a larger role in growth. The one thing that we cannot stress enough here with this great transfer of wealth to younger generations is that philanthropic tendencies we are seeing are not the same as that of a wealth management firm. Wealth management firms are considering deeply how to keep the relationship with this household as they transfer wealth and make sure that they continue to work with the family throughout the generations. We've seen that philanthropy doesn't follow that same trajectory. Younger generations don't necessarily share the same interest as the older generation's parents and grandparents do. So the onus really becomes, on the organizations to really focus on these younger populations, making sure that you can continue the legacy of giving across that household throughout the generations. Additionally, we wanna provide context on broader macroeconomic trends, particularly among affluent households. Over the past several years, wealth has become increasingly concentrated at the top end of the market. Today, roughly eighty percent of total US wealth is held by households with a net worth of one million dollars or more, as we see on the right. For those who have been following our webinars, this is a continuation of a consistent trend, and we've seen this share steadily increase over time, just reinforcing the growing importance of understanding and engaging high net worth audiences. And with this context and the importance of engaging and understanding your high net worth donors, I'm now gonna hand it off to my colleague, Zach, to walk you through why wealth data is really hard and windfall solution to ensuring you are equipped the most accurate and available data available when it comes to really focusing on these affluent households. Thanks, Bank. So we're going to jump into our next poll here, and this is about how much time you spend double checking your current, you know, wealth screening data. From some of our prior polls, it seems like many folks here do have a wealth screening provider or use wealth data in some way, shape, or form. There are many who, may not have those resources today or choose not to invest in that, but just please take a few seconds and submit whatever choice best represents your current situation. Of course, there's no right answer here. You know, historically, it's been kinda tricky, because if you aren't spending time validating this data that you get from a legacy data provider, there's a chance that you could be misprioritizing folks. And on the other hand, if you're spending time double checking the data and going through all of it, it's expensive and labor intensive. It takes a pull on your team and keeps you from working on maybe more important fundraising efforts. So we're gonna end the poll here. It looks like most folks have participated, And we'll share these results so you can see. It looks like, many organizations and many folks here spend a sizable amount of time double checking, whether that's from spot checking or, you know, that they spend a day or two to go through the records that matter most, maybe that's the most affluent folks, or there's other, you know, criteria that bump those individuals up in importance. So as we move forward, we just wanna start talking through because sometimes it's kind of this catch twenty two. And the last thing that we want for you all is for the time that you spend validating or checking data to keep you from achieving those fundraising goals that you have. So let's talk about what has made wealth data really hard historically and the issues with existing legacy data vendors. So one of the reasons that our company was founded in twenty sixteen is that we took a look around at the existing wealth data providers and found that there wasn't really a great solution in the market, and we recognized a few different opportunity areas. The first issue, which is still true today, is that there is really no good detailed net worth data available from these legacy vendors. So we're gonna unpack what that means as we go on. But, basically, what's available in the market today is often aggregated estimates based on surveys or ZIP codes, and this approach is not necessarily based on the household itself, and it's not detailed. So you will likely receive a range, and sometimes it can be a very big range. The difference between four million dollars and fifteen million dollars is rather large. So that makes it really hard to target prospects with precision and understand who might be the best constituents to focus on. So second issue here is that many of these legacy data providers also use proxies for net worth, things like home value or income. The problem with using something like home value as a proxy for net worth is that it is actually not that reliable, and it can lead you to overvalue or undervalue or even miss really good prospects. Just because you own a million dollar house in California, it doesn't mean that your net worth is a million dollars. And a million dollar home in California is very different than a million dollar home in, say, New Mexico. So this unreliability creates, validation and a verification hurdle for whoever wants to use the data. So for fundraising purposes, it's really up to you and your teams to decipher the good data from the bad, which takes a lot of time like we just reflected on in our prior poll, and it also makes it hard to trust and rely on this data to surface great prospects. One of the most interesting studies that we've seen as of late was done by Deloitte, where they went through and compared all the different data points from these legacy data providers. What was listed by legacy data vendors versus what the actual reality was. And they found that legacy vendors were less than fifty percent accuracy accurate. So when you use this legacy data, it's essentially a coin flip as to whether this data is reliable or not. So I wanna put this in a real life example of how legacy vendors missed the mark. Here, we have this sample constituent, James Smith. So in the left column, we see what the legacy data vendor lists about Jane. And on the right, we're able to compare that to the reality for Jane. The legacy vendor is getting some of these data points correct, such as gender and home value. However, we're only getting a net worth range back between five hundred thousand and a million dollars. And granted, that's a pretty wide range. A million dollars is obviously twice five hundred k. And when we look at the column on the right, we see that Jane Smith's network is actually eight point two million dollars. That means that the legacy vendors band is way off by a factor of about eight, which is a huge difference. So if you were relying on the legacy data, in the left hand column to qualify whether Jane Smith should move into someone's major gift portfolio, you might decide no. But in reality, we see that she has the net worth to potentially make a large gift to your organization. And there are a few reasons that legacy vendors miss the mark. First is that we talked about it just a little bit. Legacy datasets are usually built on survey and census data. This data is aimed at larger populations. It's self reported or even estimated rather than calculated at the household level. Second, legacy data is refreshed infrequently, sometimes quarterly, but more often than not, it's only once a year. And with how fast the world changes, it's easy to see how quickly the data can become, you know, outdated or stale after an annual census is reviewed. Wealth creation and its destruction events happen every single week. So Deloitte in a study concluded that this leads to billions of dollars and hours wasted due to inaccurate sources and validation. This is really hard. Wealth data is a tough problem to tackle. There are, you know, millions and millions of data points that come from this need to understand someone's true net worth. And so, you know, what about AI? Is AI the magic bullet that the world seems to think that it is right now? Is it really that helpful in this case? So even harnessing AI to crawl websites and other sources of information for a specific individual can get really overwhelming very quickly. So, you know, in this example, we used James Smith, and Smith is the most common last name in the US. You can see that the on the right hand side, there's the top ten surnames here as well. And just think if you have someone in your database with this name and then try to match them to third party data, we can see if they're located in California, you might have over thirteen thousand different James Smiths to choose from. So the takeaway is that coming through all these data sources to find the exact right person is going to be very difficult. Now we've talked about some of these considerations in past webinars, but this part is new even if you've joined us before. Lately, we've been taking a closer look at how AI is starting to show up in the prospect research space. So there are lots of tools like ChatGPT, Gemini, Clobb, perplexity. All those are becoming more pop more common and popular. But instead of going to multiple sources like Socio or LinkedIn, you might think, can I just ask an AI tool, and it'll synthesize everything in one place? And, yes, in some cases, these tools can save time and offer insights that you might not have thought of on your own. But as many of you know, these AI tools also come with a disclaimer, and it's, you know, posted everywhere now as they've added these that results may vary. And you have to verify the information that it spits out, which leads to the important question of what can you actually trust. If you run this same search on a name in Gemini versus ChatGPT, will they give you the same answer? Will they even cite from the same sources? What happens when you enter the same prompt twice? Do you get the same results or something different the second time? All of these inconsistencies matter, especially when you're dealing with wealth insights where the accuracy and trust are really important. Because the the truth about AI is that it's only as good as the data that it's fed. It's just like any other predictive model. And if you feed in bad data, you're going to get bad results. Garbage in, garbage out. Right? And we all know that you can trust everything that's on the Internet these days. So as a key point, AI tools like ChatGPT or Gemini don't have access to Windfall's proprietary data. So if you were to ask them, you know, what does Windfall know about this particular person or donor? They can't tell you it's not public or accessible data. And, you may be aware that here at Windfall, we launched our own AI copilot, and we've designed this specifically for fundraising and research teams to leverage AI, but with this correct foundation. Right? Real, verified, and up to date data. That way, we can train the tool to reflect the kinds of questions that researchers and development teams are actually asking. And then just as importantly, avoid many of the risks and pitfalls that come with these generic AI tools, set more guardrails. So even outside of that, we wanna put into perspective the uses of AI. So as an example, we asked ChatGPT about our CEO here at Windfall, Arup Banerjee. Specifically, we asked, is Arup Banerjee a good prospect for nonprofit gaming? Now Arup Energy is a fairly unique name, but, surprisingly, there are over fifty people named Arup Energy in the United States. And what ChatGPT returned was really interesting. It gave us three results based on that simple prompt. And when we looked closely, none of the results matched the exact spelling of Root Energy's name, and one of those results wasn't even for someone in the United States. So that raises a really big question of how confident can you be that the information the AI tools provide for you is actually about the right person. And if you're skilled at prompt engineering, which means, you know, refining and rephrasing your queries, you might be able to get closer to the right information. But we tried that with another search. Right? Like, let's continue down this hypothetical. Let's get better at our query. We'll refine the prompt and ask more about gift capacity and net worth and how to think about Dan Stevens as a prospect. And what we found is that ChatGPT doesn't have public data on Dan Stevens' network. It gave us a range, but it quickly noted that it was just hypothetical, and it was meant for illustrative purposes to continue the conversation. That's honestly the most important point of this is that even ChatGPT encourages users to get data enrichment to answer these kinds of questions accurately. You need to plug in your CRM data or use a platform like Windfall combined with an AI copilot AI copilot that has access to real and underlying data to make these tools really useful for prospect research. So whenever I hear people say, we we just do that research on CheckGPT or on Gemini or Perplexity, I I have to admit it's really not that simple as without access to proprietary data. These AI agents are limited in what they can actually provide, and that sets the stage as for, you know, why wealth data continues to be so difficult to uncover and why these specialized tools matter so much when it comes to fundraising. So as we continue, I'd love to launch our final poll for the day. This one's more about how frequently you're screening your data. Let me go ahead and launch this poll. I know that, I feel this way, and I I bet many of you do, but things are moving faster today than ever before. Right? And so I'm just kinda curious how often you're trying to understand your constituents and the insights that you can gain from some of the well screening tools that you're using. And as these answers are trickling in here, it's all across the board, actually. We we have, you know, weekly, monthly, quarterly, annually, every three years, and we don't wealth screen. And we've got a pretty good spread of results. The leader right this second is annually as the refresh rate for the constituents. I think we are just about there. So I'm going to end this poll and then showcase the results here so you can see. It looks like we have a a good spread. Right? Many teams, it looks like the the leading result is that teams are refreshing this annually. Some teams are refreshing this weekly. Like, here at Windfall, we refresh all of our net worth and other attributes on a weekly cadence so that you have those updates when they happen, the liquidity events and whatnot. But let's jump in to the next section here and, you know, break down why wealth data is so hard to get our arms around it. Right? And so, historically, there is a plethora of data available. There's more data today, and there's going to be even more data tomorrow data that we could look at and pull in that we could access. But it's really challenging for a couple of reasons. And so we look at these challenges as the four v's of data. So the first one is the sheer volume of data that is available. There's a ton of historical data out there. There are even more places that you can pull it from. And in order to aggregate everything from all of these sources, you need to have a pretty good scale of infrastructure in place even to just take that data in. The next one is variety. So this data can come in all sorts of shapes and forms, and so we have structured data versus unstructured data highlighted here. Think about an Excel sheet of a very structured transactional data that's coming in. Right? You know what each column means. Everything is filled out. It's pretty clean. That's a good example of structured data. Often, you know, donations could be, whereas unstructured data could be forms that people filled out by hand, where anything can be in a box. You have to make sense out of the story that they tell you in the box. You know, this means that when it comes to processing all those different types of wealth data and other data points that we might wanna look at, we have to understand what's going to come in and how different it is going to look as we compare these different sources. So another reason it's not that easy is the velocity of things. This is how quickly the data is moving. Think about the stock market, all the property transactions happening every single week. These things are moving on a weekly basis, and you have to keep up with this data. And with how quickly the data is being created and how quickly those changes happen, we need to be able to process it at scale so you get a result that is actually timely. The last v is veracity. So how clean is this data? Can we trust it right out of the box? You know, not all data sources are created equally, and that's where this veracity comes in. There's some data that you might want to do additional cleaning or additional transformations on, and there's some data that you may want to take right at face value. Understanding which data source is which can add another layer of complexity and makes this really challenging. So I've talked about some of these issues at at a high level as we talk about these four b's of the challenge, but we're gonna dig into a few of the challenges with specific data sources and walk through how we navigate this just a little bit. So we're gonna talk about real estate data, plain ownership, and SEC data to illustrate some of these problems. So in terms of real estate, as we're thinking about the data that's available in most search platforms or many of the the competitors, they're oftentimes using off the shelf AVMs. So these provide insight into the value of a home, and this is helpful in understanding the equity or the list price of that property. If we were thinking about, you know, how much a home is worth and how much equity an individual has stored up in that property, you really need to understand the present value of that asset. And so many platforms like Zillow have become really famous for their publicly available estimates of what a property is worth. And when we look at them in aggregate, these models are pretty accurate. So we can see that the median error rate is one point eight three percent, but this is only for those homes that are on market today, and this represents a very small percentage of homes in the US, particularly as we think about, some of these slower sellers markets where fewer folks are listing their properties. These public AVMs really struggle to keep up with the characteristics of those homes and aggregate detail from the county records, which means that as a result, they can be really far off in valuing someone's asset. So we can see on the right, you know, for the homes that are currently on market, AVM, Zestimate, is really on par with the value of the home, and this oftentimes reflects changes in the sales price. As someone is selling their home, we'll update their estimate based on that information. But as we look at the off market homes, you can see that the model are fairly off in terms of the value of that asset, which poses a lot of challenges when you're relying on this, you know, Zestimate or legacy data provider that may be using this type of ABN model as well as an AI search that is accessing this information to come up with a a reasonable or an accurate conclusion. So when we look at this at the state level, you know, this is more aggregated information. This is based on data available on Zillow's own website and their analysis that they've done on their own model that they share back with customers to help them better understand the accuracy or limitations of this ABM model that they built. We can see that the model is really off here across the US. It's not just beholden to a single state or an isolated, you know, incident where the model is performing better or worse. We can see that there's a very large percentage of homes where the Zestimate model is over twenty percent off. So as we think about, you know, homeownership being one component of someone's wealth, this can have a huge impact on understanding their overall net worth and their capacity to make a larger gift. So by their own reporting, over sixteen point eight percent of homes nationally are over twenty percent off in terms of understanding the accurate value or the present day value of that property. And the way they do this analysis is looking at the sales price for homes after they get listed and then back testing their model, right, to kinda check back. And we see that across these top states where they're most off, only one of them is a nondisclosure state, indicating that the data is less readily available for them to construct this model. So we'll get into that a little bit more in detail, but this really just shows that the Zestimate or, you know, other AVM models out there, available online is just not that accurate. So when it comes to individual properties, there can be limitations as well with publicly available datasets where they're not exactly tracking every single piece of land or parcel. This is especially so when it comes to luxury real estate. So it can be really difficult to find benchmarks in some of these markets to understand how much that home is worth. In some cases, this is due to the home not changing hands that often. There could have been many updates to the property since the last time it was sold as well. Right? So this example, this could have been a vacant lot, and now there is an absolute mansion on this parcel. A lot has changed from when this lot was purchased and to when the seller is trying to sell. And if we look at this particular property, it's listed for forty two million dollars, but Zestimate doesn't really have a benchmark available for us to understand what this home is worth. If we look at public tax history down in the left hand or the right hand corner, excuse me, we can see that this is getting assessed at a much lower value than the forty two million dollars, which begs the question, you know, how much is this property really worth based on all the data points that are available online? To make things even more complicated, regulations at the state and the city level can really impact the data that's available to train some of these models and that AI has access to. So nondisclosure states make it fairly challenging for public models to understand the value of real estate. These states do not provide as much data on the sale of homes, and as a result, it's difficult to see the sales comps. So you don't know what the home next door is selling for that's, you know, a very comparable property, and it's hard to understand the values over time. So the lack of public sales data makes it really difficult in these particular states to understand the true, you know, present day value of them. So if we look at the city level, there can also be additional considerations that we look at. You know, some of the most affluent areas, California and New York, are two examples where additional limitations and publicly available data can really make it challenging to understand the value of a property. So first, in California, prop thirteen is in effect, which basically limits the amount that an assessor can increase a property value to two percent a year. So property values in places like San Francisco or other high affluent areas likely rise a lot more quickly than two percent in the real world, but the data that you're able to get from the assessor's office is only going to reflect a two percent increase year over year. This can cause the real estate tax records and assessment records in these areas to be wildly undervalued. Another example is New York, which presents kind of a different problem. So a lot of buildings in New York City may have vanity addresses. So there's, you know, a certain status in having a building address that's on Park Avenue or Madison Avenue, and we see a lot of buildings have these vanity addresses that may not correspond to the real address. So, yeah, another piece that's even more challenging is that across New York, a lot of records don't include the unit number. New York is full of multiunit dwellings, and so it becomes really difficult to understand if the person that we're looking at and researching owns just one unit or owns the entire building. And this distinction has a lot of implications for estimating the assets or the net worth for that person's soul. So keeping this in mind, Windfall has developed a multimodal approach. So we've invested significant r and d resources into these models to accurately forecast the value of a property. So when we have models specific to nondisclosure states, that gives us a better picture as well as, you know, where those sales comps and sales history are more readily available. And we can also see for Windfall's model is that we have a median error rate of zero point six eight percent. And this median error rate includes those nondisclosure states that we were talking about and the off market homes that can be so tricky. So this indicates that we have a much stronger model than what's publicly available with sites like Zillow or or Redfin or similar and can inform your outreach to prospects more readily as you think about this being an important piece of someone's overall net worth profile. But we also understand here at Windfall that home values aren't just the only thing that we should consider. Right? They're they're only one part of the puzzle. And they can represent a very significant holding for a household, but it gets trickier as we move into these more affluent routes. So for an average household that owns a second property, maybe a vacation home, it's likely that, you know, that owner may be listed on the deed of both of those properties. But if we look at the very high network space, it's apparent that things become quickly more challenging. Right? So folks are using trusts for estate planning and setting up different shell corporations or LLCs to hide their assets from the public domain. And as a result, it makes it really difficult to understand who owns the property and then, you know, composing the correct householding of all of their assets into one calculation. So maybe an extreme example, but a very illustrative example is Jeff Bezos here on the right, where he's holding assets in a variety of different trusts and LLCs across the US, multiple states, different kinds of financial vehicles, and getting a comprehensive profile on what his real estate holdings are actually becomes really difficult. Now in this case, if you saw Jeff Bezos in your database and you did some additional research, not like you don't know who he is, but you can take a look at his public Forbes profile and understand, great. We know he has significant significant capacity. Right? But as we think about, you know, all the other folks who are high net worth or ultra high net worth that have less public profiles, this becomes a really big challenge, especially especially being that, you know, these affluent individuals often like privacy. So another well signal that we wanna discuss is plane ownership. This is a great signal of affluence, and, you know, I think everyone kinda wants to understand who does own a plane. So there's data available from the FAA that can help us connect the dots, but it gets complex really quickly as well. So on the right side of your screen, we have an actual example pulled from the FAA registry. We can see that each aircraft has a serial number along with some details about the owners, their aircraft, and the type of registration. So based on the registration type, we can see that planes can be owned by individuals, corporations, LLCs, or partnerships. Essentially, we can see the co owned planes and fractional ownership. This quickly becomes complicated. So how do we make sense of all of this and ensure that the plane ownership is being associated with the correct individual? You know, what do what do we do with it? Do we just throw this data out because it gets messy so quickly? This becomes a little bit easier to untangle as we get into a specific example. So we can see that this aircraft had has been registered to an LLC. And that may lead you to think, can I do anything with this? Can we go track that down? It's not associate associated with a household. So how do we figure out who may own this aircraft? We'll dig deeper and look at the address of the registered owner. In this case, you can see that it's a residential address. So it may help link this plane ownership to an actual household, but in order to do that through the residential address, we need to go to another secondary data source. So you might use secretary of state data to figure out the connection between this LLC, the residential address, and the household itself. So what looked like a simple data source of John Smith owns a small plane, it really becomes a lot of triangulation and verification across multiple data sources in order to correctly attribute that plane's ownership. So moving on to this third data source that we we've been talking about, SEC data. This, again, goes back to some of the challenges of common names and fuzzy matching and the lack of information that may be available across the public domain that we mentioned earlier with James Smith. So we really need to be thoughtful in how we're combining these data sources together. In this example, Michael Smith here at Public Supermarkets, he's listed as the SVP. But if we take a look at how many Michael Smiths there are even within the state of Florida and Lakeland specifically, there are quite a few folks that we could potentially match to just by name alone. In this case, over four hundred people who share a name within that particular city. So the PO box could make it really difficult for us to match the right household as well as we need to verify that with other sources to actually match to the right person. So this is where combining different datasets together becomes really powerful in understanding the appropriate Michael Smith to associate this SEC transaction with versus, say, fuzzy matching that many of our competitors use to link this transaction to a general, you know, Michael Smith within Lakeland, Florida. So last example around SEC data. In this case, we have an individual who is registering Zendesk Zendesk stock. So this gets complicated because we're looking at someone who is also a partner in a fund, And this requires us to understand whether Dana is, you know, associated to this stock because he owns it personally or if this stock is owned by the fund with which he is associated. So there's another critical variance that we have to dig into in order to understand if this is something that we should attribute back to his household or if this is something that isn't actually held or in control by this individual. As you see, you know, pulling all these sources, come combing through them and making the right judgment calls, all of this nuance is something that gets complicated really quickly. So a a tool like ChatGPT or similar can't do it yet. And development teams also can't spend their entire week deconstructing SEC forms or playing ownership across partners. So I'm gonna throw just one more wrench into the mix here as we start to think about how do we understand wealth data, and this comes from donor advised funds. This is another potential blind spot for identifying the wealth and understanding, you know, who's giving and where. So in this next slide, we'll walk through what we've recently seen with donor advised funds in the last few years. You all have likely, you know, seen the the surge in donor advised funds. There's been a rapid growth of the use of them. And the issue with that, you know, rapid onset of donor advised funds is that it obscures the giving history data points that we historically had. You know? And that's, in times past, been really important for development teams, especially because legacy wealth screening tools will calculate the gift capacity rating or the score that an individual is given as a function of the historical giving associated with the household. And so when you give from a donor advised fund, that's oftentimes obscured in terms of the giving history. And within the last five years, from the top fifty largest philanthropists' donations, three billion or nineteen percent of the top fifty gift dollars was moved into donor advised funds with noncash donations growing increasingly common. So what we're seeing is that DAF sponsors are seeing double or triple digit growth in donor contributions, and we're also seeing that the floor for opening up a donor advised fund for a contribution perspective is lowering. So this means that more and more individuals have access to this complex financial vehicle, which shields the the data we have on specific giving history. So in this blue chart, you can see that the percentage change year over year into twenty twenty four is pretty massive, especially for organizations, the big names like Vanguard Charitable and Fidelity. And this growth has only, you know, accelerated in recent years. So, again, when someone opens up a donor advised fund and they're giving from a donor advised fund, they're essentially obscuring this giving history, making it even harder for you to understand the true capacity if you're thinking about it from a gift capacity perspective. And you're using some of these legacy tools rather than approaching the conversation from a network perspective. So, additionally, what we're seeing in these charts is that when the donor advised fund contributions increase, it limits prospect researchers' ability to look at those historical contributions. And we've seen about one point eight million open accounts as of the end of twenty twenty three, holding around twenty two hundred and fifty billion dollars, and the average fund size is around a hundred and fifty thousand dollars each. So the payout rates are similar. It's been consistently year over year about twenty four percent. And, you know, over time, this becomes hundreds of thousands and even billions of dollars every year holistically that is obscured from our understanding as to, you know, where those donations are are headed or keeping track of them. So this is something to keep a note of as we're trying to identify wealth or the top prospects in, you know, our sites is that tracking these donor advised funds is really tricky. But here at Windfall, we've found ways to match households to being affiliated with donor advised funds. And we have done a a webinar series that goes into more detail on how Windfall is now doing this with donor advised funds, and we're doing this deterministically, meaning it's known to be true. You know, we're doing this through nine ninety forms, our understanding of family foundations, as well as, like we've discussed earlier, being able to triangulate across these many different data sources with confidence. So if you'd like to check that out, please do if you're interested. The reason we're focused on linking households to donor advised funds is so that you can have a clear understanding of which folks are putting assets into these donor advised funds for eventual payouts while being able to understand the complete financial picture of this household from a net worth perspective. The increase of donor advised fund participation is just, you know, one more wrinkle to consider in this challenge that we're talking about today of building accurate wealth data and then being able to prioritize the right high net worth donors for your organization. So as we wrap up here, just some key takeaways from what we've discussed. First is that most data vendors today really do miss the mark, and trying to extrapolate from that data can be time consuming, tricky, or even set you up for your, you know, hard conversations or an uncomfortable moment with donors. Next, there are lots of tools out there, but amassing and understanding and making sense of this data and then doing so at scale and at speed to keep up with all the changes that are happening week over week is really complicated. And then finally, as we saw, which, or with each of the example sources of this data that we looked into, Every source has different variances and challenges that we have to account for as we're taking it in, assessing it, and deciding on whether, you know, it's a a valuable use for us in terms of our prioritization and our our actions that we're taking. So we only covered some of the issues. There's definitely more, I'm sure, that you all have encountered, but this was a high level overview of the challenges that we face when we look at wealth data out in the wild, and then we can kinda understand how development teams are using this wealth data today.