Why Private Equity Needs Its Own Data Lakehouse | PE Analytics & AI Readiness

[MUSIC] Greg, we're back in the hot seat. You and me talking about PE level data lakes, data warehouses. Glad to be here with you. >> Yeah, absolutely. Good to be here with you, Kate. >> So we've been chatting about this topic for a couple of weeks. What are we going to cover today? >> I think really where we want to start and really what we want to cover today is just what are the benefits for private equity firm in having their own firm level data lake house. We know there's a lot of different solutions out there that private equity firms are looking at. A lot of these are driven around the interest in AI, which is extremely valid. But there's a real range in terms of off-the-shelf solutions, platforms that you can adopt to build your own model or own platform, and really everything in between. Here at Blue Margin, developing data lake houses and developing power BI reporting and analytic solutions, we're typically doing that at the company level, but do partner with private equity firms. I think we're noticing an increased interest in private equity in their own data platforms. I think today really talking about that and what we've heard and seen private equity firms do and the value we've seen them drive with those kind of platforms. >> Yeah, the last time we spoke, we definitely dug into how private equity firms are using traditional data lake housing technology to get data from their portfolio companies up to the PE firm level. We talked about any number of technology solutions to do that. But at the end of the day, a private equity firm is a business just like any of their portfolio companies. What I've seen is that when we're able to stand up a data lake at the private equity level, we can still pull in CRM type data from solutions like Deal Cloud to help the business development team really track deals the same way that any sales team would add a portfolio company level to track performance of their assets throughout the portfolio with things like eye level, which we've talked about previously. And then even just looking at general financials. So really the concepts that you could pull data from all of those systems into one location in the data lake, building power BI reports on top of that applies with the PE level just the same way it would for their portfolio companies. >> Yeah, I think it's a useful way. And of course, it being analogous kind of makes sense here, the different functions of any given company. You see that of course at the private equity level too. And you touched on it there with business development. We talked last time about portfolio monitoring. You have the Stratop's team, you have the investment relations team. And each of those teams might have different use cases for data and different ways that they could leverage data lake. All while it's driving a certain amount of automated reporting and invisibility. But then there's those other use cases that I think don't always get talked about, which are pretty interesting for us because it's really a hub and spoke model where you say, okay, we are going to have federated data, ratified data within our data lake house. We're going to be ingesting that sometimes on a routine basis, let's say from portfolio companies. There might also be one off ingestions of data like when we're looking at companies and diligence and we want to bring in their data and quickly compare it against our portfolio companies for benchmarking. And then other use cases based on the different functional areas of the private equity firm, one area that can sometimes be forgotten or at least the demands of data on this group can be forgotten is the FPN 18. Because as we've talked with private equity firms, I think democratizing data, which I don't love that term, but really just increasing access to data and the ability to use data analytics tools can in a lot of ways lighten the load on the FPN 18. And that's also very analogous to what we see at other businesses where you have a certain collection of folks. It usually is the FPN 18 that has the best control and visibility and ability to pull different data sets. But then unfortunately become inundated with requests from other parts of the business because there are not good self-service solutions and data access solutions for the rest of the business. So you effectively create a bottleneck. And I think that's the kind of bottleneck we've seen at certain private equity firms just as we've seen with different portfolio companies. I love that concept of democratizing data. I've talked to a number of firms recently who are very focused on alleviating some of the manual stress around actually preparing these reports that go out to different audiences. So you think about valuation, for example, that's often a very manual process. Something that you're doing a lot in Excel, maybe doesn't even make sense to have that in a transactional system. But at the end of the day, you've got a team working on all of that data potentially reviewing with the team to make sure that that data is correct, the ratification, like you were referring to. And then of course, you know, if Excel is one-favored tool, PowerPoint is another. The challenge with PowerPoint is that you just cannot answer those questions live on the fly. You know, inevitably there's the, well, why is this performing this way? What are the key drivers of that valuation? If you're looking at a static PowerPoint, the answers stop there. What we've really seen is companies that are much more focused on actually getting more data to broader groups at the firm level so that everybody just has a better sense of performance across the portfolio. You know, unsurprisingly, we are back in uncertain times. It seems like we've been there since 2020. And really having more people who have a pulse on company performance, firm level performance just really helps the entire team move together so that they're bringing in the business that they need and driving towards those multiples in return. Yeah, you know, I love the concept too. And that's why I use the term even though I don't love it because of the image that it evokes, which is that everyone can access and work with data. And we kind of level the playing field. But you think practically about that, that is of course not going to look like everyone is going to become a sequel master and have these skills. That's not practical. On the other side, you would say, well, there are certain different types of data that might be valuable for different groups. And sometimes that might be structured data like we see in a lot of transactional systems and columns and rows of data. But it also could be unstructured data that is still valuable and worth analyzing, but doesn't come in that spreadsheet format or in column to rows. And that gets the heart of why the lake house architecture really matters because you have a storage that will accept all types of data, everything from structured to semi structured to unstructured data. And it provides you that truly centralized location where you can say, hey, if it's data and it exists, we can put it there. The other piece in terms of why it's impractical for everyone to learn SQL to learn some of the ways that data engineers work with data today. But very practical for them to ask questions of data. And this gets into a really hot topic in an area that I think is sometimes misunderstood, which is generative AI for analytics, sometimes how we refer to it. But really the idea of using natural language and simple questions to then run through an AI agent and have it translate and take care of all the technical coding required to actually query that data set and return an answer. That is practical. You see a lot of marketing pitches and tools that are promising the moon when it comes to this. But what we've seen and what we know works is a process of centralizing data, modeling data and building a semantic layer so that any AI agent has a dictionary in a way to navigate all of the various rows and columns and data sets in there to actually return an answer that the person querying and others in the firm in this case would have trusted. And so I think that's an important part of data democratization as well is, you know, practically we're not going to teach everyone technical tools. We'd like to get there an AI agent seem like they could get us there. But I think a lot of firms and this is where we certainly help to provide that perspective are understanding that it is a process of building a semantic layer. And I think the good news is that that can actually work, but it's not going to be a solution you just buy off the shelf, plug into your data. And then suddenly everyone's going to trust it. That trust is built slowly, but I think is a real use case and something that even over the next eight to 12 months as we see this technology get even better and better is a very real value that private equity firms can drive off of that lake house. Again, thinking about someone in maybe investment relations that needs a quick answer for a limited partner, being able to just simply ask that instead of sending an email, wait for a response, maybe sending another one because you didn't get exactly what you wanted. That process can stop and you can start democratizing data and then people get solutions and answers to their stakeholders a lot sooner. So very real definitely requires setup, but underscores a value of that lake house and why you want that as a foundation to drive these different use cases. One of what I love about bringing generative AI into the mix, natural language, querying, or NLQ is that that may still be future state. Still, we've got our own beta going on internally in London and Mark talked about that in a recent podcast. Very close, we're getting really great results, but still lots of work to do to make sure that the results that are coming up to folks who are asking those questions are accurate every single time. But all of that said, you raise a good point because anytime that you're investing in a semantic layer today to serve traditional power BI reports that folks can engage with, you can control access to those reports. So not everybody needs to see all of the same things, which is just a fantastic application of reliable security. All of that work that you're putting in for the foundation is exactly what's going to help you accelerate to NLQ in the future when the technology really starts to become much more robust. That's such a key point. And I'm so glad that you highlight that, Kate, because I think there's a tendency for all of us to want a fast forward and immediately start grabbing the latest and greatest technology and use it. But then the question is, will we use it? Are we going to trust it? Is it actually going to be helpful for what we're doing? I mean, those are real questions. And I think these capabilities are built up a little slower, probably than most people expect. But you raised such an important point that progress can be made in rationalizing and building that semantic layer and building reporting and automated dashboards that do great things and increase data visibility overall. At the same time, that's the exact same work you're doing to enable better generative AI, better natural language querying in the future. So the two can go hand in hand at our not separate work streams. I think that's such a critical point. So as I think about the work we've done at the Pee from level in the past, you're just to share what use cases might look like there. I've already touched on using Power BI and the DataLick House architecture and technology to support deal tracking, deal management, really helping your business development team stay on top of progress in a really constrained deal making environment. That can be incredibly valuable. Of course, we've seen it with the FPNA and the finance side of the house, as you've already mentioned, just getting teams time back in their hands so they can work on higher order strategic thinking, incredibly important. And then you touched on the investor relations piece of things. That team spends a lot of time preparing materials, power plants, really eliminating the need for that copy, paste, deck preparation can get them a ton of time back. So they're focused more on the relationship side than the hand jamming and PowerPoint creation side of the house. Where else have you seen Pee firms leveraging data? Yeah, I think you look at the operations team, the Stradops team, sometimes operating partners, but deal teams, the folks that are working closely with companies on operational improvements and the different areas of business that they're trying to work with. So it depends. Maybe you do or do not have companies data hooked up to your data lake house. If you do, then I think that gives a lot of autonomy to the private equity firm, just as we talk about with FPNA teams, not have to wait to the CFO send you a data extract or the company gets back to you with the data that you're asking with the Pulse system, but to really have that data and visibility into it so that you can catch and call things and point things out to the management team as you're working with them. I think that's a particular use case that's interesting. I touched on the idea of diligence and using a data lake house in that method as well. So the exact method of ingesting data that a firm will get during diligence, there's a couple of different ways to look at that, but the idea is that we have some data on our current portfolio, current performance, we want to benchmark and compare that against a diligence data set that we get. And we want the data lake house to provide some extra horsepower behind that so that we're not doing a lot of manual work in Excel that could stretch out and take longer. And maybe that means we speed up our analysis and we have that much more time to request additional data or ask extra questions because we finished it and we can now move on. So I think that's another valid use case as well. I think there's also use cases where you would look at the lake house being a foundation and ratified data that we trust that we can now push into different applications. So you talk about other generative AI platforms, ones that can build decks for you, that can deliver different outputs, but that usually are gonna require data. And you're gonna want that data and you're gonna want trust in that data before you feed it into any platform to generate something. And so I think that's another use case and kind of going back to the hub and spoke conception that the hub is of course the lake house and one of those spokes that we can plug in later might be a platform we purchase. We already know we're gonna get a ton more value out of that platform because we're feeding it high quality data and it's the same data set that we're looking at on dashboards and that the IR team is looking at and so on. And so I think that's another potential use case where firms can say, you know, we're not exactly sure what tools we want to purchase in the future. We do know they're gonna need quality data. So we can build that and it really opens up value later on with those kind of platforms. - I just love the idea of thinking about it. As a tool that's really extensible. So it is going to serve the needs that you have today. It's going to serve the needs that you have in five years that you can even think of right now. Spending this time cataloging your data, understanding the business logic that's going to drive the firm now. It's just an incredible investment that's going to pay off and have real ROI today, but it's also just going to position you so well for the future. - Yeah, it's like you said, Kate, it feels like since 2020 there's just been different levels and errors of uncertainty that we've gone through for different reasons. I don't personally say that changing are really ever changing in the world of business. And so I think where we've seen so much interest as private equity firms that have recognized that data is crucial to what we do, we just haven't invested as much as we probably should have over the past 10 years and the technologies to allow us to handle data and produce value from data. I think the last five years, the last couple of months have probably only underscored that even more. - One of the main things too, as I think about firms who want to get something like this in place and I worked with many separate private equity firms to do this is to really pick that first use case that's going to help you generate buy-in throughout the firm. Maybe it's something that gets your team hours back a week that they're not actually having to crunch on decks. And you can see that material return in terms of where else they can spend their time. And alternatively, maybe it's something that people have been clamoring for for ages, but you just have never actually been able to serve reports to them. So it's really that first space that you're going to attack from a reporting perspective can be so important. And definitely just advise any firm that we're working with, really any company that we're working with to put some effort into charting out that roadmap to really thinking through what's going to help me generate buy-in, get the rest of the organization excited about investing in this with me. So you're not just rowing upstream, trying to get a BI initiative off the ground. I think it's best practice for the portfolio companies that we work with and would definitely say it's a best practice at the PE firm. - Well, absolutely. I feel like it might even be more critical. But just like you said, to start understanding and featuring some outputs from this new foundation, this new data platform that the firm has, so that the rest of the firm can start to understand and maybe even generate some of their own ideas in terms of how this central hub of data could be leveraged in different ways. But the sooner you start featuring those outputs and folks are interacting or seeing data in a new way, that really does start to spark interest, particularly when it's also, hey, that's saving me time, that's saving me from having to wait for something. Everyone loves that and that also helps to underscore the value and why we're investing in data. So yeah, couldn't agree more with that. - Awesome. Well, any other thoughts Greg? - No, I think this is good for today. There's a lot more on this topic. And so we're certainly invite folks to reach out if they want to chat more about it. We're always happy to share our perspective. - Awesome. Well, always glad to be in the podcast booth. - Yeah. - Thanks Greg. - Yeah, same here. Thanks, Kate. - Not bad. - That was a pretty good. (laughing) - Yeah, I looked over at the time. - I was like, I know, I was like, "Whoa man, we're just rolling." (laughing) - Yeah, yeah, yeah. - I'm funny to say, we're ripping off each other. - Yeah, that's very conversation. - Hopefully a couple of spaces you can get out. - Yeah, but like, (phone ringing)