Wealthtech vendors generate as much data (or more) as they consume
- The Seismic Shift from Batch Processing to APIs
- How Do You Define “Householding”?
- Integration by Design
- Managing Data Sources with Different Timing
- Data Lakes Aren’t Nirvana
Craig: Welcome to our recap of our recent webinar which was called “The Tower of Babel: Consolidating Wealth Management Data From Multiple Sources”. I’m your host, Craig Iskowitz, and on this recap we have my partner in crime here, Jeff Marsden, Head of Product at Xtiva Financial. Hey Jeff.
Jeff: Hey Craig, glad we’re doing this.
Craig: I’m glad we’re doing this, this is fun. This is like football right, after the game the announcers do a recap of the game, talk about the things they liked and didn’t like. We’re going to do the same thing with this webinar. And this webinar was part of a series we started doing last year. We started doing webinars about the engines driving wealth management, client experience.So this was cool, talking about consolidating data, we had a great panel. Rich Romano from FidX, Mike Stern from Riskalyze, and Todd Winship from Temenos. All these guys really know their stuff, it was hard to get a word in edgewise.
Jeff: It was a great start to our second season.
Craig: Exactly, Season 2. We’ll drop all the episodes so you can binge watch.
Jeff: Exactly. I liked the football reference. I feel like we needed to take a camera tailgating to see who has the best snacks.
Craig: We’ll do that in post production, we’ll add that. So before we start, we both picked some of the questions and topics that were covered in the webinar. Before we do, I want to throw out a statistic, because I love statistics, a study by IBM found poor data quality costs the US economy over $3 trillion a year due to lower productivity, outages, higher maintenance costs, to name a few possible bad outcomes. Consolidating data from multiple sources, one of the biggest issues is cleaning your data. Getting it clean, getting it accurate, reconning, it’s something we talk about constantly, we both work with our clients on.
Jeff: It’s pretty important for us and our clients because we’re making compensation and performance management decisions with them or on their behalf. That’s hard to do well with dirty data.
Wealthtech vendors generate as much data (or more) as they consume
Craig: Let’s switch to the questions here and talk about some of the highlights from this event. So I’m going to go first, one of the things we’ve talked about, I think Todd or Rich was saying that basically all vendors in our space and probably in many spaces that are similar, are both givers and takers of data. They both generate data and they consume data. Now the givers and takers, that’s different than the psychological use of those terms where takers are self focused and givers are others focused, as a psychologist would say, but it’s more of I’m generating a lot of data that other firms can use or my clients gonna use, I’m pulling data from many sources, it can be dozens of sources that my clients use.
Jeff: So true. No, you think back maybe 15 years, and I think the data path was much more linear. There was a smaller number of sources or a single source and the data sort of flow through that natural back end, front end of the experience, and ultimately out somewhere. And today that’s just not the case, it’s a much more complicated and interconnected landscape, which puts a lot of pressure on the need to figure out how to get the data that’s valuable in a complete and orderly fashion from one from one place to another, and recognize that that isn’t a linear flow. It’s a bit more of a web. It’s a web of data movement today, because something might be created that’s valuable and Riskalyze or valuable in Xtiva might benefit something else in that ecosystem that is adjacent to or even back up the food chain or back down the food chain.
Craig: That’s true, and we’re seeing a lot of a lot more data analytics being used and you look at a lot of firms now are hiring or signing senior executives who just handle data analytics, like Todd Winship over at Temenos, that’s his that’s his title, Head of Data, and Analytics. And in order to do analytics, you need data. So we’re working with a number of firms that are building out analytics tools into their systems that didn’t have them before. Whether they’re wealth platforms or or other parts of the ecosystem, realizing that they’re generating to their clients to help them manage their businesses better, help them optimize their processes. Are you seeing something similar?
Jeff: 100% and we see roles created at our clients that’s related to data, or more commonly, thinking about how to get more value out of the data beyond simple transactional operational processes, but enriching the customer experience getting more leverage out of staff in the front office or the middle office, using that information to advance and think through the product and service roadmap, so they’re constantly in a state of having data in a high frequency availability to optimize decisions around how they go to market or how they optimize even things as simple as pricing of products or other attributes of how a product or service is offered.
Craig: We’re also seeing data being generated in other areas like operational areas, whether it’s research or practice management. They’re all generating data that can be analyzed and leveraged and providing value, because if you can’t measure something, it’s hard to manage it, it’s impossible to optimize it if you can’t measure it. So be able to capture all the data is something that a lot of firms are are building out tools to do that where they didn’t have them before. And they’re also prioritizing the data for their clients based on expected business value. Don’t just show them everything and let the clients try to figure it out, help them figure out what data is most important. And one of the things we recommend for some of our larger clients where they’ve got 10,000 of advisors or even 100,000 advisors on their platform is show us what other people are doing, be the Google of our space and you can see what other advisors are looking at, what other firms, broker dealers or large RIAs, what they are interacting with, which reports they like the most and bubble those up to us will make us wade through all this mounds and mounds of data to figure out what we need.
Jeff: Crowdsource what’s valuable in many respects. We’ve noted and we’ve enhanced our capabilities, but we’ve noted a desire for context, whether you’re a senior executive or a financial advisor, when you’re given access to a dashboard, or you’re directed to some analytics or report that’s asking you to make some conclusions, there’s an evident change in how individuals now look at that information and say how do I get more context to understand this better? And I think it’s an interesting push pull dynamic because the amount of information that now can be added to that analytics to provide context around it is fueling a desire for more context.
Craig: Is that a negative feedback loop? Or is it a positive feedback loop?
Jeff: Let’s call it a positive feedback loop.
Craig: It makes more sense positive. I think it’s negative word. It makes it worse the more it goes, but either way.
Jeff: I think you’re probably right in a technical sense. I just prefer the positive word.
The Seismic Shift from Batch Processing to APIs
Craig: I agree. Alright, so next topic. We’ve seen a shift in the industry over time. I’ve been in financial services over 30 years and wealth management for 17 and we’ve seen the industry slowly shift from 100% batch processing, overnight files, shipping critical data via FTP now to more API connections and streaming data, what are your thoughts on that?
Jeff: Well, one, I think that that’s 100% true. It’s certainly changed a lot. We had some discussion with the panel on on this topic and I think there was some compelling description, some compelling challenges that are that are result from a desire for higher frequency information and some challenges and getting access to a complete set of high frequency information.
Jeff: There’s no doubt, there’s no question that the market wants more information at a higher frequency. I’m not always convinced that you need to have it at the frequency that the desire exists for. I’m not sure there’s enough change sometimes between that completeness date, but there’s no doubt that people want information at a very high frequency. And the technical landscape is evolving to support that as well, so the ability to do that exists, but it makes the debt that exists in old systems, that have daily or weakly run processes, it makes that debt a bigger burden on the business when you’re able to move data or create valuable data at a high frequency. Real time or multiple times during the day and you want to drive decisions or business processes off that and you’re encumbered by information that’s necessary to drive the process that we’re still runs on a daily batch or worse yet a weekly or semi weekly completeness process.
Craig: This question generated a lot of discussion on the panel. This went on for about when 15 minutes, just the panel amongst themselves, I didn’t have to say anything, because there’s so much to talk about. And there were two new words generated that came up in this question was one you mentioned, “trustibility” of data. That’s a new word for the wealth management industry. And Rich, asked the question, when you’re looking at your data, you want to know “what’s commoditizable”.
Jeff: We should make sure we let the folks at Oxford know that we’ve got a couple of words.
Craig: Another part of the shifts we’re seeing shifting from static processing to more dynamic and streaming processing, was the question on you still need to have a good data dictionary, you still need to standardize your data. Rich was asking what’s an account number? Just asking that question can give you a multiple answers.
Jeff: Well, Craig, put very specifically in a wealth management context, and we in our conversation, sort of cut across banking, wealth and insurance, but very specifically, if I just think about it from a wealth management perspective, we have 95 financial service relationships at Xtiva and we have some customers where there could be six or seven sources that have unique rep codes for the same human in them.
Jeff: So Rich’s point about account numbers is bang on an account number across multiple sources. Trying to match that up to a client then to a household. Well, we still wrestle with a fundamental issue related to wraps, which is there’s different ways of identifying wraps for financial advisors in different source systems and that’s a that’s a challenge for how you supervise them. It’s a challenge for how you manage them, it’s a challenge for how you compensate them, it’s a challenge for how you connect them together into teams of advisors, it’s a solvable problem. We have solved for Xtiva but it’s evident the challenges that are exacerbated by more and more sources at play.
How Do You Define “Householding”?
Craig: I’m going to throw out another stat here, according to a Deloitte Survey, 86% of respondents have increased their spending in data and analytics over the past three years. That’s a lot and I think everyone sees the value here. So it’s there’s no secret that you need to be able to understand your data, be able to manage your data if you’re expected to remain competitive. Alright, so next question. This is yours, you picked this one out, we asked the panel about householding, what does household mean to your firm? Why did you pick that question?
Jeff: I just find householding to be really interesting case study and some of the challenges of trustability to bring that word back to the table, but it’s a very interesting, specific piece of data. If you ask somebody what a household is, there’ll be lots of different perspective on how you would define it, and it would depend on who you’re speaking to and what their role is. But our customers, your customers, might have 3, 4, 5 different householdings for different purposes. There might be one for statement consolidation, there might be one for customer relationship management, there might be one for fee billing, there might be one for financial planning. There could be a variety of different householdings.
Jeff: And that seems that seems like a pretty simple piece of data. Well, a household name, a household key, some unique identifier of a household, but now you’ve got four that I’ve just enumerated, and those households will change over time. Households are created and destroyed. How do you do reliable longitudinal analysis? This is a single example of the complexity of bringing full data perspective to the table that allows firms to really understand in a reliable data driven way, what their business looks like, what the topography of the business is, and how is it changing? And the solve for that isn’t easy.
Craig: Some of these things can be solved with grouping, simple grouping like we’re going to group these accounts, these people if we’re going by Social Security number, we’re going to group them for billing. We’re going to group them for reporting.
Craig: But it gets more complicated when you’re doing things like models. So if you say, Well, I want to assign this model to a household. Well, then a household may have five individuals, and they may have 30 accounts. In my family we have five individuals and we have close to 30 accounts and we’re far from wealthy. It’s just you have multiple jobs, you wind up with a bunch of rollover IRAs, you’ve got things for the kids and 529 plans and joint accounts and single accounts and taxable non taxable so it winds up being very complicated. So in order to do a household model, if you’re applying one model across a household, the software has to be able to figure out on a rules based engine what the location of certain assets are, where the best place to put them is, how to reduce the tax consequences, and how to manage drift in the model all at the same time.
Jeff: Absolutely. How do you balance the needs of the enterprise to understand households for the purposes of reasonable operational processes or compliance processes with the needs of the advisor and the advisor team to manage the customer or the household, the collection of clients in the household in the way that they have built their service model? There’s two very distinct needs themselves.
Integration by Design
Craig: That is true. All right. Moving on to the next I’m trying to keep everything to five minutes, five minutes per per recap question here. We’re on target. And so next one was a question that I picked, we asked the panel, what is integration by design, and how does it help an enterprise wealth management platform handle new types of data delivery mechanisms, such as streaming data or other types of delivery of data? And I thought one of the keys was we go back to the sources of truth. What’s your golden source? What’s your silver source? And in order to do that, you need to understand where your data’s coming from. But more importantly, what integration by design means from my point of view, I come from a computer science point of view, software building, that you need to be flexible in your design. Because you’re assuming that our company is going to buy another company. We’re going to be integrating, we’re going to be merging other systems. And if our software isn’t able to integrate with other tools easily and without even knowing what those tools might be or what that data might be, but keeping it flexible, then we’re gonna have problems down the road when we let’s say we’re an order management system and now we have to integrate a billing system. Well, how do we do that? We don’t even understand that data in our system is very closed.
Jeff: I think Todd was talking about an integration by design, a very thoughtful way of articulating having the data strategy with a means of accessing and availing data for the purposes of whether a back office middle office or front office purpose, whatever that is that the decision in the approach to doing it to make it deliberate to basically build it into the strategy of the product or the service, not to just be a not an adjacency or a must have, because you have to have it to FTP some data from point A to point B. They talk about it in a very deliberate way of how they focus on on that as part of their product design and customer strategy.
Jeff: Again at Xtiva we have a with similar philosophy is probably not quite as fully formed as Temenos would need to be about it but we call it Xtiva Connect. And it’s basically a philosophy that our integration team follows and relies on some technical capabilities to be as thoughtful and deliberate as possible about how we onboard data and how we make the utility of data available through the ecosystem as efficiently as possible. Your point from a few minutes ago about a common data dictionary a common understanding of what that data looks like is critical to to any of those things being successful.
Craig: It’s all about having a strategy, the way I look at it if you when you’re building software, what’s your strategy? Don’t just build, here’s the features we need. This is build it, but what’s the strategy? Where are you going with it? Do you expect it to scale do you expect it to integrate the expected to be robust, effective, be accurate? These are what are called non functional requirements that have to be considered when you’re building software.
Jeff: Maybe a shout out to season one, Craig because we had a lot of talk about data strategy in season one.
Craig: Yeah, throw a quick plug if you want to. You can go back and listen or rather well watch the video recordings of our previous webinars season one as we’re calling it, of the data as an asset webinar series on our website EzraGroupllc.com, if you look for the webinars, do a search for webinars you’ll find it and you can register for them. You have to register one at a time, but then you get a link, a private link that you can then watch them at your leisure. And we also did blog posts. So if you go to our blog, and you search for the webinars data as an asset, you’ll find some quick summaries of some of the key talking points. So if you’d like to listen to you like to read about it, you have we support you both ways.
Jeff: A little nighttime reading and a little car driving.
Craig: Yeah, I drive a lot and I listen to podcasts all the time.
Jeff: Me too.
Craig: You get you get a lot done. Although the one thing I find with podcasts, not to get off topic, is it’s hard to take notes. Like I want to remember what I listened to. And what by the time I got there I’ve forgotten half. We need some sort of some like an Alexa for or Siri for podcasts and it’s going to stop it and say oh, remember this. Remember the last bit just clip it and pull it out a podcast and put into a note for me.
Jeff: Apple added this thing recently where you can report if there’s an accident or report there’s a speed trap or whatever. And it’s pretty easy, two taps on the screen and it’s done. I want the same thing for podcasts. I don’t even I don’t even need to make the note of what it was. I just want to know that these six spots in the 45 minute podcast, I want to go back and listen to it. Easily find it trigger the memory.
Managing Data Sources with Different Timing
Craig: Another. good idea, we’re generating a lot of good ideas right so let’s get on. We can do our separate innovation podcast later. So we just did we just did integration by design. We’re still on target, next one is pain points. We kind of talked about this right when the pain points were talking about was how do you know where to get the data from? That’s back to the golden source, we don’t need to cover that again. But let’s talk about data expiration, knowing the timing of the data. And that was an issue. When you’ve got 25 data sources, it’s not enough to just be able to handle them, incorporate them and store them. You got to know how do I build a view of my client? I’ve got 25 data sources, and they’re all coming in at different times of the day, week or month.
Jeff: Absolutely and how long can I rely on a piece of information and can I rely on that piece of information for that duration, and in any context or only in certain contexts? It’s no wonder that our that our customers are adding folks with responsibility for getting value out of data and managing the quality of it. In a truly front office data as an asset kind of way because those are our critical aspects to doing that successfully is being able to have confidence in when data is going to be passed it’s best before date.
Craig: This is part of our data assessment that we do at Ezra Group is building a data dictionary and part of the data dictionary is not only what each field means because a lot of times the people who made the system don’t work there anymore., if it’s an in house system, and we don’t know why we don’t no one remembers why we created a certain field. Or they’ve reused a field which we call a Cluj. They reused a field that was just available for something else and the data you see in the field doesn’t mean what you think it means. Or as you mentioned, how long is that data good for? That should be another field, this data has to be refreshed every week, every month every year. Whatever that data is that whatever that frequency is, that’s got to be captured.
Jeff: You probably would encounter when you’re doing that work, the important distinction of you know, I use the gold, silver, bronze notion of source there’s other ways of describing that hierarchy. But the notion of being clear about you know, there’s an assumption that gold is better than silver or bronze. It’s generally the case but not always. Sometimes there’s an improvement downstream that’s valuable for a specific purpose, maybe not generally across all purposes, but you must encounter a lot of need to really crack into the nuance of gold, silver, bronze sourcing and and how it influences those trust factors, that best before date factor the full context of what that particular data element is.
Craig: Sometimes there are some heated discussions. And the reason why we do that is because in an emergency or in a crisis, you don’t want to have to go figure out well, what’s the backup for this piece of data? We’re not getting it anymore, the data feed that was providing it is down. Now what do we do? You don’t want to have to decide then, you want to do it when cooler heads will prevail and when you’ve got time to consider and that way you build it into your data dictionary. Here’s each source and here’s each data field, here’s the sources we’re getting from and here’s the primary, secondary, tertiary, golden, silver, bronze, whatever you want to call them, but define them in advance so it’s easy for the customer support team, the data operations team to know exactly what the backup is, where to get it from, how to provide it, how to deliver it so the clients don’t see anything. That’s the goal is you want to you want it seamless to the client, so they don’t know that the data has changed that something has gone down.
Jeff: You want to want to avoid as much as possible the law of unforeseen consequences raising and rearing its head and while I don’t think this is exactly a black and white rule, I do think that the closer you are to bronze, the more likely that the law of unforeseen consequences is going to pay you a visit.
Data Lakes Aren’t Nirvana
Craig: That is true. And you really can’t avoid unforeseen consequences, but you can try to mitigate your risk. We have time for one more. So let’s talk about a favorite topic that comes up in a lot of our panels. Data lakes. It’s time for data lakes. It’s a lively conversation between our panelists, no matter who it is, no matter where they come from, when we mentioned data lakes everyone’s got an opinion. What’s yours?
Jeff: I think we’ve had some I think we’ve had our best sound bites when it comes to things related to data lakes. I think maybe we’re going to spend a little bit of time in one of our sessions later this season talking about that maybe in a little bit more detail. But absolutely, they’re an interesting debate. I think there’s a lot of difference of opinion about what they’re meant to do and what they can solve.
Jeff: One observation I would share you know from our conversations with customers, our webinar series, is there is an awakening to data lakes not being Nirvana. Todd made some articulate points today about approaching data lakes not as not as Nirvana but a piece of the puzzle in your data strategy. Not to avoid them, but to see them as a piece of the solution, not the solution.
Craig: Absolutely. That’s something almost every one of our panelists on any one of our events has talked about. You want to know why you’re doing it, why you’re putting this data lake in. What’s your reason? That’s similar to almost any piece of technology. You don’t want to throw it in, any piece of an enterprise technology. It’s different than just Well, I want to add an app on my phone. If you’re bringing something like a data lake, you need to know why you’re doing it. You want to understand the ramifications. You want to know what data it’s good to hold and what data it’s not good to hold. Why do use a data warehouse versus a data mart versus a data lake?
Jeff: I generally try to look at them from the perspective of being a pretty positive indicator of the fact that the market customer, our customers, the market, your customers, are becoming much more focused on data as an asset to their data strategy. Lakes seemed like a reasonable way to get your data collected up, to put it somewhere where you knew you were going to be able to do something with it in the future, even if you didn’t have the whole strategy sorted out.
Jeff: I’m generally you know, with our panelists that we should be careful not to drown in our lakes but there’s a positive there, which is it reflects the importance that this marketplace is putting on getting data into the hands of decision makers that allows them to make good, smart, contextual decisions that improve the velocity of the business.
Craig: Data lakes, can have errors, can be inconsistent, can not be structured properly. Or it can be unreliable, because they’re not built properly. It’s prevalent and I like to get more into the technical aspects of things. So, anybody listening, if you’re CIO, CTO, and you’re an Azure client, you’ll know this term, they have a term called a delta lake like alpha beta gamma delta, delta lake, which is an open source storage layer that sits on top of multiple data lakes to ensure the reliability and it works with other Microsoft technologies.
Craig: But what’s nice about it is we’ve been talking about different ways of delivering data, and these firms, whether it’s Amazon, Google, Microsoft, or any of these other cloud providers, realize that different types of data require different technologies to manage. So I know Azure has something called an event hub. That’s for streaming data to ingest streaming data. They have Azure data bricks, which is for batch data ingestion. Or maybe it’s called Azure data factory, and then then you’ve got your extracted transformed data can be loaded into the delta lake layer.
Craig: It gets really complicated but it’s the point I’m trying to make is a data lake isn’t the be all end all you need to realize that it needs to be reliable. You need to be able to use the tools that are available to handle different types of data delivery mechanisms, so that your data doesn’t corrupted and it can be managed and support whatever downstream systems you want.
Jeff: Every one of those cloud providers has their own version of those tools. As long as you’re on the cloud journey, you’re going to be able to access those tools on a managed service basis and make some terrific progress with your data strategy.
Craig: Yes, you will. And that’s the end of our recap, we’re out of time. This was a great discussion. I think that the webinar, which we recorded earlier today was a fantastic discussion. It cut across wealth management, banking, insurance and really tied everything together. And I was thinking with our season one, season two of all these, these webinar events that they’re like free classes you can take for a mini MBA in data analysis and management, specifically to to our industry.
Craig: Go back and listen to them, you’ll be able to go to our website EzraGroupllc.com and register for this. The private link isn’t available yet. It’ll probably take a couple days for it to get to get there but you can register still. And when the private link is posted, you’ll get an email with it as well.
Craig: So thanks for listening. We appreciate your time everyone, and remember to go to our website, EzraGroupllc.com and sign up for our newsletter on the homepage at the bottom, once a month you’ll get an email chock full of wealth management, goodness analysis, news, links, and other information. You will not regret it. So thanks for listening. Talk to you again next time.