Machine Learning and Big Data in Private Equity: Is Networking Still Needed?
The private equity industry has historically relied on networking to find investment opportunities. However, the perks of using big data and machine learning for deal sourcing have turned the heads of managers at private equity firms.
Many household name companies have received funding from private equity at one point in time. FedEx, Intel and Cisco Systems are all examples that you may recognize. Without the funding, these companies would not be as well-known as they are today.
The relationships between investors, private equity firms and their limited partners have proven to be beneficial for all parties; however, technological innovation changes these relationships and the way firms operate. The use of machine learning disrupts traditional deal structures and improves firms’ deal-sourcing and client interaction abilities.
The Traditional Way
Traditional methods of investment decision-making for private equity firms are based upon human interaction and meeting potential clients face-to-face. Sourcing deals depends on these connections and the firm’s ability to network effectively.
Following the capital-raising stages, firms use proprietary deal flow, a method of applying connections with the lawyers, accountants and executives of the industry to find investment or buyout opportunities before other competing firms. Joint deals, called syndicates, are also considered if the partnering private equity firm does not have the funding to enter a lucrative deal by themselves. The larger the network of a firm, the more connections it can use to find another company in search of funding or buyout. Following the connection, the firm may analyze the client’s financial and performance data to further understand if an investment would be feasible or not. These concepts are the foundation of private equity and venture capital deal structures.
Traditional methods have been effective for the decades that private equity has been around. They have made name to some of the biggest private equity and venture capital firms in the world, such as Goldman Sachs, Accel and Sequoia Capital. But other than basic financials of a company, how else may a firm use big data to support their investment decisions?
Too Much Data — Is it Good or Bad?
Thousands of businesses enter the market every day. As capital continuously pours into the industry, the decision-making process for private equity firms becomes increasingly difficult.
The abundance of data and crowded markets have created the need for more complex analyses. With the flood of tech startups today, many businesses appear to be almost identical at first glance. The opportunities for investments have increased, but they have come with the challenge of finding the viable ones among the potential failures.
This is where software may come in handy. Firm’s need to be able to:
- Analyze companies by the masses
- Process large quantities of data to identify trends, creating clearer graphs that allow for easier comparisons between the startup and the rest of the industry.
- Use the results to determine the startups with the most promising growth potential
The following survey conducted by Blue Future Partners and PEVC Tech shows the reasons why firms are inclined to use machine learning in their operations. Out of the 137 responses, the majority agreed that they wanted to use software to improve their ability to find and execute deals.
But the use of machine learning software and artificial intelligence in venture capital is not just an opportunity for firms. Today, it is becoming a trend. When one firm adopts the technological approach of finding startups, other firms are forced to do the same to remain competitive. The 137 private equity firms were also asked about how they expect to approach the new trend of using big data and machine learning software. The surveys produced the following results:
Traditional methods are still common, but there is a clear movement towards the use of data analytics — especially in deal-sourcing — as firms dedicate more funds towards software development. Although the types of deals may not change, firms are now expanding their investment teams to digitize their structures of deal sourcing.
The Digitized Way
Digitization is showing no signs of stopping as it incorporates itself into the operations of many firms. As time passes and more companies prosper and fail, there is more data for private equity firms to use. As a result, this has led to the increased usage of data analytic driven investments. The private equity industry is a world of risk and payoff, but digitized firms using machine learning technology greatly reduce their risk and reap larger returns from startups growing into large companies.
A survey conducted by KPMG asked several private equity firms about their takes on big data and machine learning. 79% were aware of the technology, 9% were considering implementing it, and 12% were already using a form of related software. Based on historical data, Critical Future projected that AI and machine learning industry revenues are expected to grow by more than six-fold by 2025.
The following venture capital firm is an example of the success that following the trend of digitization and big data can bring.
Motherbrain — EQT Ventures
EQT Ventures, a Swedish private equity firm, is an example of the success that confidence in digitized methods can bring. They have expertly used a newly developed software called Motherbrain to lead the investment decisions in their venture capital wing. In 2017, they used the software to make 20 investments, most of which focused on the industrial, consumer goods, technology and health care industries. The firm has closed more than 30% of its investment deals based on the software’s results and has raised an impressive $50 billion of capital across 27 funds. Unlike most other firms around the world, their portfolio was raised upon the connections found by machine learning technology.
How it Works
The machine learning software works by identifying a trend within startups and using it to label them as either high or low growth potential. The software analyzes several time series — a set of data points indexed over time — of the startup’s financial performances and attempts to match them with the times series of successful companies. The more similar the data, the higher chance of success and more inclination to invest. Likewise, trends matching those of companies that experienced failure or had less success indicates lower growth potential.
As EQT feeds Motherbrain with data from its own and external investments, the software’s algorithms will continue developing a better understanding of the trends that led to the success of the surveyed companies. The longer and more often Motherbrain is used, the more capable its algorithms become in differentiating between high and low growth potential startups.
Big Data Makes Big Results
EQT initially trained the software using data from their own successful investments found through traditional methods. Since then, the focus has been compiling financial data, including past funding, web ranking, app ranking, social network activity, and much more to add to Motherbrain’s data base for its ever-improving algorithms.
Motherbrain does not only find new companies for EQT. Its capabilities extend to providing EQT’s clients with data on the external analysis of companies. Some examples include:
- The effects that the behavior of competitors have on the sales/feasibility of a project
- The correlation between location and sales of a store location
- The effect that competitor sales in the market may have on the performance of the company.
The following mobile gaming startup is an excellent example of how Motherbrain had supported an investment deal.
Small Giant Games, a mobile gaming startup based in Finland, was identified by Motherbrain in 2017 after EQT performed an analysis on the mobile gaming industry in Europe. Just ten months after EQT agreed to provide the startup with $5.7 million in Series A funding, the company’s revenue grew by an astonishing $33 million. Due to the promising results, EQT has returned to Small Giant’s owners and are discussing a second investment deal of $41 million.
Provided the data, Motherbrain can guide a company all the way from its startup stages to becoming a major corporation. It may sound like the software can hand EQT’s investment deals to them on a silver platter, but the truth is far from it. As with any technology, it has flaws.
Gut Feeling vs. Data Analytics
Motherbrain can process huge amounts of data. The more the program runs, the more accurate its predictions become. But not all data sets are accurate, complete or available for the software to use; thus, there will be gaps in the calculations of the software, creating skewed results. For example, Motherbrain has a more difficult time analyzing emerging tech industries. Its method of comparing time series to other successful companies is less effective as there is not much data yet on established companies in the growing industry.
There are also types of variable data that cannot be easily inputted into the software. Key parts of a startup such as strong team dynamics, body language and the motivations of the owners are all determining factors in the success of the company but cannot easily be quantified into an acceptable form of data for Motherbrain to use. Whether firms are digitized or not, traditional networking and learning the motivations of executives within the industry remain of paramount importance to the success of the firm. EQT’s deal with Small Giant Games highlights the value of these areas.
After Motherbrain identified Small Giant Games, EQT analyst Lars Jörnow met with the CEO of the gaming startup to discuss the direction of their new releases. After considering the respective customer segments, Small Giant Games decided to change their genre from casual to midcore games. Although the decision of marketing towards a new audience seemed risky, Lars wrote in his Medium article that he was “intrigued by the team’s ambition to mix this evergreen core game mechanic with a deeper metagame”. Their strong team dynamics and motivation indicated that this company, which already had the performance to be labelled as high growth potential, would be a worthwhile investment and connection in the industry. After the meeting, the two kept in touch and the Series A investment deal was eventually finalized.
A fully digitized firm’s practices would not be perfect. Due to the limitations of data input, networking and learning about the management team of a company provides valuable insight that a software’s programs cannot.
So, Which One? Traditional or Digitized?
Both the digitized and traditional firms can operate feasibly, but question remains. Which method can bring a firm higher return, lower risk and more success overall?
The truth is, depending on the size of the firm’s network, any method could be effective. Cutting-edge software like Motherbrain has proven to enhance deal sourcing and client interaction, but a strong network of connections is also important to the process.
The limitations of both methods are clear. Machine learning, although powerful, cannot cover the qualitative aspects of the company. However, solely using networking to source deals limits the amount of companies that a firm can analyze. Considering these factors and the modern capabilities of machine learning technology, a mixture of both digitized and traditional methods of investment appears to be the best way to approach an investment deal. Hendrick Landgren, a VP at Spotify before joining EQT Ventures, puts it best:
“To build models where you can find great companies is great, but it’s not as if you just press a button and then you have the investment. “It requires much, much more work.” Much of that work is in the traditional investor realm of building relationships. “It’s about knowing which relationships to build and when. “That’s what we use Motherbrain for.”