Why Wealth Data Is Really Hard and Its Impact on AI

Thanks so much for joining us today. We're really excited to dig into this topic. I know it's a a pretty hot button, and given the fact that we're a data company, I think the the implications of AI is is one of the things that we're most focused on with a lot of our clients. So let's jump in. To give you some quick introductions, we're gonna, you know, dive into our webinar today, which is why wealth data is really hard and its impact on AI. To give you some quick intros, my name is Cooper, and I oversee the commercial retail sales at Windfall. So I work very closely with all of our clients and partners as they're kinda stringing up their new data driven strategies and looking for solutions that their marketing and go to market teams can all use. And then Sabrina is joining me, who is a part of our customer success team at Windfall. She works really closely with all of our customers to help them get the most out of our engagements and just does a phenomenal job of supporting them. So I'll go into a little more detail about our company soon, but I wanted to give you a quick overview if you're not familiar. So Windfall is a people intelligence and AI company based out of San Francisco, which is where HQ is. We have over fifteen hundred different customers across the country right now, and that ranges from luxury retail, hospitality and travel, financial services organizations, and even a long tail of of nonprofits that we work with today. So this is the agenda. Pretty straightforward. After I walk through some pieces, you know, setting us setting the stage on windfall, we'll look at some macro trends. We'll talk about the particular difficulties of wealth data, and then what this looks like in your workflows. And at the end, we'll have time for a q and a as well. Cool. Now we've talked about some of these wealth data topics, obviously, in a few past webinars, but this part is is pretty new even if you've joined us before. So we've been taking a closer look at how AI is starting to show up in the luxury retail and the travel space, and and tools like ChatGPT, Gemini, Claude, Perplexity. Right? These are all becoming a lot more common. Instead of going to multiple sources, like people look at, you know, look up tools like a Spokio or LinkedIn, you might just think like, hey. Let's use an AI tool, and let's put everything into one place, which is the right thing to do in this day and age. Right? But in a lot of cases, these tools, you know, these tools are able to save you time. They can offer you insights that you might be able to think of on your own, but they also come with a disclaimer. The main, you know, the main thing I wanna warn you about is that the results may vary. Right? So, you know, without a a single source of truth, so to speak, you need to verify your own information, which leads to an important question of who can I trust? Right? So if I run the same search on Gemini and ChatGPT, do I get the same answer? Do they use the same sources? You know, what happens if I even enter the same question twice? Am I gonna get the same result or a completely different answer? The inconsistencies really matter, especially when you're dealing with wealth data and accuracy and trust or pretty much everything in these customer interactions. Because the truth is that AI is only as good as the data gets fed. So just like any predictive model. Right? If you feed it some bad data, you get bad results. This is the old garbage in, garbage out mantra. So the AI tools don't have access to proprietary sources like Windfall's data. So if you were to ask them, hey. What does Windfall think about this particular client or, you know, what have you, right, they can't tell you that. That's not public information. So let's let's put this into practice. We asked ChatGPT about our CEO at Windfall, which is Arup Banerjee. And, you know, we have a a lot of nonprofit customers. So in this case, we asked it, hey. Is Arup Energy a good prospect for nonprofit giving? Now Arup Energy is a fairly unique name, but surprisingly, if you look in the United States, there's over fifty people with that same name. And what ChatGPT gave us was really interesting. It gave three different results. But when we look closely at these results, none of them matched the correct spelling of his name, and one of those results wasn't even for someone in the United States, raises a really big question. Like, how confident can you be that the information these AI tools provide is is really about the right person? So even if you're skilled at, like, prompt engineering, which means you're able to refine and rephrase the different queries that you're building, you might be able to get closer. So that's exactly what we did. We spent about five minutes working with ChatGPT, trying to prime it with more details and and really guide the responses here. So we also tried it with our other cofounder and our CEO, Dan Stevens. And we refine this more and asking about his gift capacity, which you could really think about in a commercial context is, like, his spend capacity. We also asking about things like your net worth and how I should be thinking about him as a prospect. And, obviously, you know, from the response here, you could see that ChatGPT doesn't have any public data on Dan's net worth. It gave us a range, but it also noted to us transparently that it was a hypothetical range and just meant for illustrative purposes. So it's an important point. Right? Even ChatGPT encourages people to get better data enrichment so that you can get these kind of questions accurately and use their product the way it's meant to be. So you need to plug in your CRM, or you can use a platform like Windfall combined with this AI that has access to real underlying data to make these tools more useful. So whenever I hear people say, yeah. We just do it on ChatGPT, we say, like, hey. It's it's it's really not that simple. Right? Without access to the proprietary data, these AI agents get get pretty limited on what they can provide, and that's gonna set the stage for why wealth data remains to be so difficult to uncover and why some more specialized data tools matter in your go to market strategy. So getting into sort of the windfall overview here, you know, one of the reasons why we exist as a business is because there's all these legacy data vendors out there, and there's a huge portion of them that are just giving inaccurate insights. So this is coming from a study done by Deloitte that showed that most of these data providers that have existed, some of them for over a hundred years, were only ever about fifty percent accurate. So why is that? Well, here's how you can start to think about it when it comes to choosing the right data data layer. So on the left, you have your CRM. This is where all your customer activity lives. And the question is, what kind of data are you layering on top to drive decisions? So option one is more of your traditional data provider, might be very low cost, but the data is very sparse and tends to be incomplete, not a ton of context. You get a few attributes, but not enough to drive the meaningful segmentation or action. And then option two would be something more high fidelity, like a windfall. The focus for these providers would be more around depth, accuracy, and a lot of rich context. So we're talking about having very complete profiles, faster updates, really strong match rates, and ultimately, wanna translate that into insights that you can use. So the trade off isn't just cost. It's thinking about whether your data actually enables your team to prioritize the right people, to personalize outreach, and ultimately drive revenue. Right? So that's the ultimate goal. Context that your team needs to execute effectively. So the quality of your data directly impacts the outcome, and that's what we're gonna talk about. So, again, on the left here is gonna be your lower context data provider. We're missing a ton of key information. The net worth is understated. The job level is unknown, and that that really creates a lot of uncertainty right from the start. And that uncertainty flows throughout the entire workflow. Right? So you get a maybe qualification that leads to a very generic drip email campaign. The outreach is incorrect, and then you're improperly segmenting these folks. Right? So the result is we have these longer sales cycles that don't give us any context. There's just a weaker client experience. Then on the right, we're talking about data provider b, which is the high fidelity. You get a much more clear picture of the individual. We have a precise net worth. We know they're philanthropic. We have a definition of their job level. So this just streamlines the whole activity. Right? We have better qualification. We personalize our outreach. We put them to the right team right from the start, and we get a faster close time. So this is just an example, but the takeaway really is better context, across every step of the funnel, and improves your outcomes. Awesome. So I'll just tell you a little bit of background on Windfall. We were founded in twenty sixteen, and our whole vision from the beginning has been to democratize access workflows and insights on people data. So PeopleData brings two things together for us. Right? That's looking at someone's wealth profile. That's also looking at their career profile. Some vendors give you one or the other. Sometimes they give you big ranges. We combine them both and keep them current. We actually refresh our data every single week. And the reason why we did this is because we entered this market where there's a ton of existing problems, and there's still big problems today. So the first is that there's really no good source of net worth truth. There might be large ranges. They might cap out at a certain point. Like, we're not gonna give you any data on anybody over five million dollars, right, which leaves out huge segments of multimillionaires and billionaires that we don't have any precision. So good wealth data has always been really hard. And folks are looking at things that they use as proxies. Right? And and then that's what we do. If we can't find a good source, then we're gonna make a comparison. So household income is one of them that we see a lot. Home value is one that we see a lot. But our team is in San Francisco. So when they look at ZIP codes or home values, it's like everybody seems like a millionaire. Even though I'm in an apartment that I'm renting next to a multimillion dollar mansion, it's just not a good way to organize. And finally, even with AI, organizations still have to manually filter through good versus bad data, and it's it's really tough to do. So how do you know what data points to use, and how can you trust in your overall workflows? So Windfall has built a data asset. We track over a hundred trillion dollars in wealth across a hundred million households in the US. About twenty million of those households have a net worth of a million dollars and above, which is what we call affluent. And we work with many commercial go to market, marketing, clienteling teams across at at all the different industries. So luxury retail, we work with a lot of travel and hospitality, wealth management firms. We also have a nonprofit arm that works with nation's leading nonprofits, universities, and health care systems. So how do we work together? How do we support our clients? At a very high level, we empower you with accurate people data, and the foundation of that is always identifying and prioritizing who teams should be focusing on as their ideal customer profile. So how does this data jump into your c r m sorry, your CRM and help you understand who you should be focusing on? The second piece is sort of deeply understanding your database. So this could be us building additional segments, cohorts, personas. How do we give you insights through this robust analysis and segmentation? That helps us help you determine a data strategy based on where you're actually winning. And lastly, this data isn't really useful, right, until it's being fed into your workflows. So we take all these insights, and we leverage them to help you engage with your customers. We're talking about things like personalized messaging, marketing audiences that are modeled off your best customer, and plenty of other use cases. So the last slide, this cool animated slide I wanted to show, is just talking about the different ways that we work with your organization. Right? So you might have a multitude of different databases or datasets. You got information living in a CRM, a data warehouse. You got event attendees in a spreadsheet, you got transactions. There's all these disparate sources that when you put through in a windfall, we could put together these these really smooth outcomes that are driven from that. So whether that's segmentation or analysis, how you can think about additional upsell and cross sell for your customers, or, like we said, getting in front of new people, which is building those predictive models for your acquisition methodologies as well. Cool. So it's just a little bonus that I wanted to throw in here, which is just some macro insights that we've been seeing. So, you know, setting the stage right now, you know, what is the k shaped economy, and what does it have to do with me? Well, this is a dynamic that isn't just reflected in wealth concentration. Right? This is showing up in a lot of different segments. We're looking at how people are saving and spending. So since the pandemic, the top ten percent of earners in the US accumulated a lot more excess savings than the remaining ninety percent. So as a result, that group now accounts for roughly half of all consumer spending, which is what you see on the left. And we see this same pattern play out in Windfall's client database as well, where the top earners and high net worth individuals tend to convert, purchase, donate at a significantly higher rate, and they drive a disproportionately large share of total spend and overall revenue. So as we look ahead to twenty twenty six, these trends are even more important because that divide is getting even bigger. So today, roughly half of all U. S. Consumer spending is driven by top earners. And in many categories, about seventy percent of growth comes from the affluent households over a million dollars in net worth. So they're not only driving the majority of spend, but they're also growing at a significantly faster rate than the average U. S. Household from a wealth perspective. Now, from a strategy perspective, this makes prioritization really critical. So when allocating your time and your budget, focusing on these high income, high net worth segments is where we consistently see the highest returns. And this isn't isolated to a single industry. We see the same pattern across, you know, retail brands, nonprofits, travel companies, even financial services, alternative investments, wealth management. Management. I mean, you name it. This is, like, one of the number one things that we talk about on a daily basis with some of the largest companies in the world. So lastly, I I I wanted to provide context on broader macroeconomic trends, specifically looking at affluent households, right, this resilient group. So over the past several years, wealth has become increasingly concentrated at the top end of the market, and right now, eighty percent of total U. S. Wealth is held by households with a net worth of a million dollars or more. So for those of you who have likely been to, you know, several of our webinars, this is a continuation of a trend that we keep talking about. Right? We see this share steadily increase, and it just reinforces the growing importance of understanding and engaging these high net worth audiences. So with that, I'm going to hand it over to Sabrina to walk through, you know, more specifically why wealth data is really hard. Thanks, Coop. So, yeah, in terms of, you know, everything that we reviewed so far, you know, how is implementing this in terms of workflows and practices, practices, you know, really difficult and, you know, how we end up, working to solve for that. So if you'd like to move to the next slide, Coop. Awesome. So let's get into a little bit more about why Dell Wealth Data is so hard and kind of wrap our arms around all of that. So, historically, there's been a plethora of data available. There is more data available today, and then there's gonna be more data tomorrow, data that we could look at, that we could pull in, that we could access, but it is extremely challenging for a myriad of reasons. And we look at these challenges as the four v's of data. So the first is just sheer volume of the data that's available. There is a ton of historical data out there. There are even more places that you can pull it from. In order to aggregate everything from all those sources, you need to have a pretty good scale of infrastructure in place to even take that data in. The next is variety. This data can come in all different shapes and forms. We have structured versus unstructured highlighted here. So think about an Excel sheet of very structured transactional data that's coming in. You know what each column means. Everything's filled out, and it's pretty clean. So this is an example of structured data. Whereas unstructured data could be forms that people have filled out by hand, so a lead form, anything where there's, you know, a box that you have to fill in and kind of make sense of that. This means that when it comes to processing all of that, in terms of wealth data, we have to understand that it's going to come in and look wildly different from, you know, source to source. The third is velocity, and this is how quickly the data is moving. Think about the stock market and property transactions. You know, these things are moving on a weekly basis. And you have to keep up with this data with how quickly it's being created, how quickly it changes, and how, you know, we need to actually process that at scale so we can action on it. The last fee is veracity, and that is how clean the data itself is, so how much you can actually trust it right out of the box. Not all sources are created equal, and that's where this comes in. So there are some data that you white might want, you know, additional cleaning of or additional transformations on, and there's some data that you may want to look at face value. Understanding which data source falls into each bucket again adds another layer of complexity that makes this really challenging. So I've talked about these issues at a high level, but now we're going to dig into a few of the challenges we face around these areas with specific data sources. So as examples, we're gonna talk through real estate data, plane ownership, and SEC data to illustrate some of these problems. So what we're seeing here in terms of the data on the, left starting with real estate data, most platforms and providers rely pretty highly on just the, you know, off the shelf AVMs to estimate home values. So these models are useful at a high level, and they give you a sense of property value, which can help approximate, equity or list price. And platforms like Zillow have, you know, really popularized this approach with publicly available estimates. In aggregate, these models do look fairly accurate. So, for example, median home error rates can be as low as one point eight percent, but that's primarily for homes that are actively on the market. And the challenge is that on market homes represent a pretty small share of total properties, especially in a slower supply constrained market. And the vast majority of off market homes is where these AVM calculations really struggle. They rely on limited or outdated public data, and they often can't fully account for property specific details. So as a result, valuations can be pretty meaningfully off, so specifically when you're trying to understand true asset value and homeowner equity. So you can see that dynamic on the right for on market homes. Estimates tend to track closely with actual sale prices. But for our, off market homes, we're seeing that the gap really widens. And this creates real challenges if you're relying on legacy data providers or even AI tools that are built on these AVM models. And then when we look at this at the state level using Zillow's own published analysis of their model, we see that this isn't actually an isolated issue and that the inaccuracies are consistent across the US and not even concentrated in just a few states. So a meaningful share of homes are significantly misvalued. Nationally, over sixteen point eight percent of homes have estimates that are more than twenty percent off from the actual sale price. Why does this matter? Because homeownership is a really large component of overall wealth and how we're calculating that. So if the underlying property value is off by that margin, it can pretty materially impact how we're accessing someone's net worth, their capacity, and understanding, you know, their ability to spend. Zillow arrives at these figures by backtesting their estimates against actual sale prices once homes are listed. And even in states where data availability is relatively strong, we still see pretty large error rates. Only one of the top states with the highest inaccuracies is a nondisclosure state. These are states for your information where limited public data would typically understand, you know, weaker performance, but a majority of them are not. So in terms of, you know, what this all means, takeaway, you know, even by Zillow's own reporting, these models really have meaningful, limitations in terms of how we can understand a household's wealth and then act on it. Taking a look at the data that we're seeing here at a individual property level, these limitations become even more pronounced. So public datasets don't consistently capture every parcel or reflect all changes to a property over time. So this is especially true in luxury real estate where it's often difficult to find reliable benchmarks. And in many cases, these homes don't transact frequently, so the property itself may have changed very significantly since the last sale, but that's not captured. And then what was once perhaps a vacant lot could now be a fully developed estate, but the underlying data hasn't caught up to reflect that transformation. So looking at this example, this property is listed at four, forty two million dollars, but there's no reliable estimate available to benchmark its value. And then if we turn to public tax assessments, those valuations come in significantly lower, which is raising a fundamental question. What is this property, you know, actually worth? And this really highlights the challenge of relying on incomplete or outdated data when trying to accurately assess home value estimates. So taking a look at, nondisclosure states, which makes things even more complex, These are states where regulations at that state or local level can significantly impact the data available to train these models. So in these states, home sale prices are not publicly reported, and that makes it much harder for, you know, models to establish, accurate comparable sales to understand home values. And without visibility into what similar homes nearby are selling for, it becomes even more difficult to benchmark that value. And without, you know, strong comparables, the accuracy of these estimates consistently degrades, and it degrades pretty quickly. So in markets where public sales data is limited, these models face a pretty fundamental constraint and making it even more difficult to rely on, property value. And then taking a look at the city level, so there are additional factors that can further limit data accuracy, particularly in some of the most affluent markets. So here, we're gonna be taking a look, at California and New York City. So starting with in California, proposition thirteen caps annual increases in assessed home property values at two percent. In reality, home values, especially in markets like San Francisco, Los Angeles, often grow much faster than that. But the top public assessment data only reflects that capped increase, which can lead to properties being significantly undervalued in, you know, tax records as well as public records. And then taking a look at New York, which presents a different set of challenges, many properties use vanity addresses. So for example, a building may be listed on Park Avenue or Madison for Prestige even if that's not the true underlying address that we're seeing. So this can create inconsistencies when we're trying to match and validate property data. Additionally, many records in New York don't actually include unit level details. So in a market that is so dominated by unit buildings, apartments, it can be really difficult to determine whether an individual owns, you know, a single unit, an apartment, or if we're taking a look at an entire property. You know, it's something that has a pretty major impact on accurately estimating assets and then eventually net worth. And then taking all of that together, these, you know, local nuances really highlight how difficult it can be to rely on this public data in terms of real estate information to accurately value real estate assets and then apply that to an understanding of a household's net worth. For the average household, property ownership is relatively straightforward. So if someone owns a property, a primary residence, and a second home, they're typically listed directly on the deed. Then if we take a look at more high net worth individuals, ultra high net worth individual segments, and things become a little bit more than that. So these individuals typically use trusts, LLCs, and other entities for estate planning as well as asset management. As a result, ownership is typically pretty obscured, and it makes it difficult to determine who actually controls a given property and how to accurately roll these assets up to, you know, the correct household. So on the right is a pretty extreme example with Jeff Bezos whose real estate is held across a network of trusts and entities. And even with a highly visible individual building, you know, such a complete picture of holdings, you know, it does require significant effort to put that together. And while someone like Jeff Bezos can be easily identified through public sources like Forbes, the challenge is much greater when we're taking a look at those high net worth individuals who have more of a private, profile. So this is where traditional data approaches that we've discussed do tend to break down, making it difficult to fully understand asset ownership and total household wealth. So far in this presentation, we have focused on property value, but debt is actually just as important when we're trying to understand a household's true net worth. So at a glance, two households can actually look very similar based on, you know, the total property value, that we're seeing. But once we factor in outstanding debt and loan to to value, the picture changes quite quickly. So as an example, a household with two point five million, dollar value at their home and has no debt, has a very different capacity than someone with a similar property value but a high LTV and significant leverage. So this is where a lot of traditional approaches fall short. Valuations alone can be misleading without understanding amortization and debt structure, and they can overstate someone's actual financial capacity. And, importantly, owning property doesn't necessarily mean that that wealth is accessible. Equity is what matters, but liquidity is what is driving action. So the takeaway here is that to accurately assess capacity, you need both sides of the equation, both assets and liabilities, not just those surface level, you know, valuations. And then taking a look at another type of asset ownership, another strong signal of of wealth is plain ownership, something that a lot of organizations are interested in understanding. And there is available available from the FAA that help, you know, connect the dots, but it does become pretty complex pretty quickly. So on the right is an example from the FAA registry, and each aircraft is tied to a serial number along with details about ownership, registration type, you know, other attributes. And the challenge is in how ownership is structured. So aircrafts can be registered to individuals, corporations, LLCs, or partnerships. We also see co ownership and fractional ownership models of planes as well. So very quickly, it becomes quite difficult to determine who owns the asset and how much of it they own. And that raises a key question. How do you accurately associate this signal to the right individual and make it actionable, or do we just ignore it because it's a bit too complex? And that's where data quality and entity resolution really matter. So taking a look here at this example, it becomes easier to to understand this as we walk through. So in this case, as an example, the aircraft is registered to an LLC. And at first glance, that might seem like a dead end. You know, it's not directly tied to a household or an individual. However, if we go a little bit deeper, if we look at the registered address for that LLC, we may find that it maps to a residential address. And that app opens the door to linking an aircraft back to a household, but that doesn't happen just in one single step. You need to bring in additional data sources. So for example, you may use secretary of state filings to then connect to the LLC to the individual and then tie that back to a household. What initially looks like a pretty simple dataset actually requires triangulation across multiple sources to accurately, attribute plain ownership. And without that level of, resolution, it is quite difficult to turn, you know, a single, signal like plane ownership into something that's actionable. Another signal that we like to look at is boat ownership, but not all boats are created equal. On the surface, you know, oh, someone owns a boat. We're gonna make a inference about their their affluence. But then when we dig in, there's a wide range from smaller recreational boats to large luxury yachts that really differentiate in terms of their value. So we can see here that a ninety foot yacht is a very different signal than a twenty five to thirty foot, fishing or leisure boat. And the usage and cost and ownership profile typically vary pretty dramatically. There's also differences in how these vessels are used. So lake versus ocean, leisure versus commercial, which further, impacts how indicative they are of wealth. So pretty similar to plane ownership. The signal itself isn't enough. You need context. So the type, the size, the usage to determine whether this is something that, will actually impact your workflow in terms of assessing affluence. And without that, we really risk treating someone, you know, very similarly across people with very different types of asset ownership in this category. And then lastly, looking at SEC data, this highlights another challenge around fuzzy matching and limited information in public records. So, for example, we might see a filing for Michael Smith who's listed, as an SVP at Publix. On its own, that seems pretty useful. But then when we zoom out, there are hundreds of individuals with the same name even within a single city like Lakeland, Florida. And without strong identifiers, it becomes very difficult to confidently match that record to the correct individual. Details like a PO box don't help much either, so we're often left with ambiguity unless you use that data to validate against other sources. This is where simply relying on name based or fuzzy matching, which are pretty common amongst most providers, can lead to incorrect associations. Instead, you would need to look at multiple datasets to accurately resolve that and then tie back to the right household. And there's a pretty similar challenge with ownership context in SEC filings. So if we take someone, like a venture partner who is involved in financing an event, They may appear in filings both as an individual shareholder as well as a fiduciary representing a firm. And then from public data alone, it's not always clear whether these holdings are personal or tied to an organization, and that distinction is quite critical when we're trying to assess true individual wealth. So, again, the takeaway is that aggregation alone isn't enough. You need context and proper attribution to build an accurate picture of household level assets. So in terms of this, example here, just again on that vein of SEC data, we're seeing an individual who's registering Zendesk stock. And on the surface, that looks like a pretty clear signal of personal ownership. But it becomes more complex when you realize that this individual is also a partner in a fund, And that raises a pretty important question. Is this stock held personally, or is it owned by a fund that they're affiliated with? And that distinction is critical. If it's personal, it should be attributed to that individual's household. But if it's tied to a fund, then it shouldn't be considered as a part of someone's personal wealth. And this is another example of why context really matters. Without a proper understanding of ownership structure, it's easy to misattribute assets and really overstate or even understate, you know, the true wealth, of a a household or individual. So taking a look at wealth data within your workflows, you know, wanna understand, you know, how that can potentially be implemented. So as we look at this next slide, what you're seeing once the slide is built out is a maturity curve for how well data is able to support AI and action. So the x axis that we're seeing here is maturity, which means how sophisticated your use of data is, not how long, you know, you've been around. The y axis is business value because the more mature your data foundation is, the more AI can actually do for you. And we break this out into three stages. So first, establish, is where most organizations are going to start. So they're working out of their CRM, segmenting on transactions, maybe layering a third party, data enrichment provider. But, you know, that's kind of that basic segmentation, at that established level. Drive is where you start getting more value from the same data. So smarter segmentation, better reporting, tailored direct marketing, and predictive models. So, you know, we definitely want to optimize on that, but this is definitely an improve upon, the first stage in terms of targeting and prioritization. Orchestrate is the last where data tends to begin to trigger action automatically within workflows. So AI can generate content systems that can alert teams when something changes or prescriptive workflows. And in terms of embedding this into, you know, workflows and stages, this is kind of how we would consider that. But the key in terms of what we're seeing here isn't you know, it's not good or bad. It's just about identifying, you know, where you feel like your organization, really fits so we can start, you know, thinking about ways to, you know, work with data in more sophisticated ways. So we went through a lot today. Would like to, you know, get to some questions if we have any, but just to recap a few takeaways from our conversation. So the first, in today's environment, it's critical to identify and engage high net worth individuals. Ultimately, they're driving, you know, a very disproportionate share of wealth and spend, and it's important to capitalize, you know, based off of folks who do resemble that. Second, not all data is created equal. So in terms of legacy providers, a lot of them are missing the mark and relying on incomplete or extrapolated data or just really stale data that can lead to poor decision making that doesn't really have any correlation in, you know, reality with what these households actually look like in their capacity. Third, even with all of this data available, actually assembling it and making sense of it is incredibly complex, especially at a household level. And finally, to really capitalize on this opportunity, it's not just about having that data. It's about operationalizing it. Implementing data driven work workflows is ultimately what drives better targeting, engagement, and at the end of the day, you know, outcomes.