AI’s role in the private markets data analytics revolution

As private assets have risen in prominence across a wider range of institutional investors, this has created new logistical and operational challenges for both investors and asset managers. Data management, in particular, is a function that is experiencing transformation due to the increase in the amount of available data to analyze around private assets. Fortunately, this coincides with new advances in technology – particularly breakthroughs in artificial intelligence – which could present new ways of not only staying on top of these growing data demands, but also helping deliver unprecedented value for investors.

Chris Sparenberg, head of iLEVEL, S&P Global Market Intelligence’s private markets operating system, recently sat down with the Buyouts publication to share his thoughts on the subject, including the role of AI within our private markets solutions.

Q: How are expectations regarding data and data management changing in private markets? What challenges does this present for both asset managers and allocators?

We are at an interesting point right now in terms of how the industry is evolving. I am referring to the explosive growth we have seen across private assets, with increased allocations to private markets funds, in general, commanding larger percentages of the average institutional portfolio. That has in turn led to scaling, and in some cases hyper scaling, of asset managers in terms of the growth and differentiation of new strategies, vehicles and asset types.

Added together, this creates a tremendous additional need for improved reporting, transparency, and more granular data. This impacts asset allocators acutely as they now have more positions. These assets are now such a large part of their portfolio that they need to better understand the underlying investments, drivers of performance, risk areas and so forth. For a very long time, investors have not had enough data to properly judge portfolio performance and have therefore had to rely on PDF reports and spreadsheets spread across their inboxes and file folders. As their concentration on private assets intensifies, their demand for the same kinds of transparency and risk metrics they are accustomed to for publicly-traded assets has now expanded to these newer holdings. They need more data from managers, which in turn creates an additional reporting burden for them. Managers not only have to satisfy these requests in an efficient way, formatting data that is digestible for clients, but they are also facing their own challenges in terms of how they utilize this data.

Therefore, as the market continues to evolve, this demand for data – and the need to collect, clean, and analyze all of this – has been considerably dialled up as well. Specifically from an S&P Global perspective, this becomes very interesting as there are opportunities for us to support our clients and add value by managing and enriching data for them. This allows us to really enhance that middle office experience, in particular.

Q: Where are you seeing AI being deployed to streamline and enhance data management? At what point does this happen in the typical workflow?

A lot of managers have spoken about this and the rhetoric around AI generally tends to be all-encompassing, with people excited about how it can be used in so many ways. At a realistic level, the application of AI-powered automation is about accelerating processes and improving the quality and accuracy of analysis conducted. It is about automating human processes first and foremost and being able to do these kinds of processes at scale over a larger data set. This not only drives efficiencies within the specific systems that are being automated, but clearly frees up people from manual tasks so they can offer greater value elsewhere – essentially allowing their time to be used much better.

Going further, the deeper applications of AI involve finding new insights that humans would struggle with themselves – or at least within the same timeframe. The applications really are vast, from how the data is collected to new insights at the highest level. Traditionally, a lot of analysis time is taken up sifting through data manually. Once this is automated, and reduced to a fraction of that time, the whole analysis process is taken to another level.

Right now, we are already seeing some of this potential come through. For instance, I head up iLEVEL, our operating system for asset allocators and managers, which encompasses portfolio analytics, valuations, peer comparables, commitment pacing, and reporting. We are the main dashboard that shows our clients the health and performance of their private market holdings. It becomes key if you are going to feed data into something as mission critical as this that both the quality and quantity of data must be as high as possible. To that end, we offer managed data services to our clients today where we manage data collection on their behalf. Our managed services centralize private investments data from clients’ portfolios using AI and subject matter expert review.  We also use AI to power our recently-released Automated Data Ingestion capability, which revolutionizes the monitoring the private investments by significantly reducing the time required to gather documents, extract key data points, and upload the data into iLEVEL.

We have seen first-hand the importance of automating these processes to maximize efficiency and accuracy. In addition, we have also seen an increase in the amounts and types of data being reported and we have expanded our models to reflect that.

Q: AI relies upon data quality. Is this sufficient within the private markets world?

Ironically, AI can support this over time – helping identify weaknesses and potential errors in data sets as it learns more.

More than quality though, in the allocator and manager world there is a transparency and repeatability component that has to come first. Any concerns around data quality have more to do with what happens in the transmission of data when it comes from a PDF or some other file and then is extracted by whatever means. That is specifically where there are more risks of something dropping off or being misidentified. Generally speaking, with data quality we have reasons to believe this will continue to remain high and reflects the capabilities of data collection agents and data management systems. The added value comes from placing an expert human in the loop to view results and train the model so that it can perform increasingly more complex data collection over time.

Q: Where do you see untapped potential to optimize data management practices using AI? Are there use cases which remain aspirational for many firms?

Automating and scaling up human processes is undoubtedly the first priority. Working to shorten the time to value for data collection and reinforce this quality are the biggest opportunities after this.

When it comes to opportunities on the horizon, there is still no singular solution where you can push a button, and a system immediately identifies the perfect investment. That just doesn’t exist yet.  Each manager and investor’s process is unique. However, with so much low-hanging fruit in terms of process automation that can free up resources, these firms will have plenty to focus on in the meantime.

Q: With so much excitement about AI in the market, how are firms deciding which use cases to prioritize?

Simply put, the tasks currently creating the highest volume of manual work will be prioritized for AI automation. Anything done at scale - be it on a daily, weekly or quarterly basis - will likely be the best opportunities for immediate automation. With these processes, firms can effectively take a set of predetermined rules and then train a model to tackle them. The ability to do this successfully means an increasing amount of work can be run through the same models.

Personally, I tend to say this a lot with regards to data collection and how financial statements can be processed en masse to get the best extractions: no two companies or funds will report the same way, so there is always a level of mapping and mastery that goes into the data collection process. There is a core set of data points we can consume from the average financial statement, which can help inform how this data collection is automated – making it a great candidate for automation.

Building on this, how much deeper can AI take us in terms of how we look at a financial statement? From there it is up to other parts of the workflow, and every firm will be different in how they tackle this, but any manual or repeatable tasks will be logical for AI automation.

Q: AI is dominating the headlines. Are there are other important tech innovations that are exciting you?

It is hard to say because so many new tech innovations have an AI component in them. However, some of the other applications that are exciting me now are around visualization of data and how this is made more tangible for users. This could have significant benefits for investors.

Elsewhere, the confluence of public and private markets data is a really interesting dynamic. We are at the very beginning of an analytical revolution in private markets, but this very necessarily relates to the relationship between public and private markets. A lot of very interesting things spin off from this. 

This article was originally published in Buyouts' AI and Advanced Technology report, March 2025.