Enabling Buy & Build Strategies with Machine Learning

Tldr;

• In today’s market, where capital is expensive and competition for large platform opportunities is fierce, private equity firms are increasingly interested in buy and build strategies that allow them to spread their funds across more assets

• To be successful firms need to move quickly when they identify a platform opportunity, ensure that the opportunity aligns with their investment strategy, and move beyond personal networks to find suitable bolt-on opportunities

• Machine learning (ML) systems can be built to 100x the capacity of deal origination teams and help researchers by identifying soft signals, assessing the alignment of a target with the firm’s investment strategies, and uncovering bolt-on opportunities that many firms will overlook

• A firm’s deal origination strategy is proprietary and to ensure that competitive advantage is retained, you should be critical when you evaluate what parts of an ML system you should build and what components you can buy

Interested in learning more about how ML can support your buy & build strategy? Get in touch

Buy and build strategies are a tried-and-tested tactic for private equity companies to generate substantial returns to their limited partners (LPs).

A buy and build strategy involves acquiring a platform company and then acquiring complementary or related companies, known as bolt-on acquisitions. The goal is to create a larger and more valuable entity that can generate more revenue and / or achieve cost savings. These strategies consist of three steps. The first step is finding and acquiring a platform company that has the potential for future growth and expansion. The second step is identifying and acquiring complementary or related companies, known as bolt-on acquisitions, which can provide synergies and enhance the value of the platform company. The third and final step involves integrating the platform and bolt-on companies into a cohesive unit, where the combined strengths of the entities can be leveraged to create a more valuable company.

Deal origination for buy and build strategies is complex and private equity firms must navigate an array of challenges just to find quality opportunities. There are some common obstacles when sourcing platform and bolt-on deals but each investment type also comes with its own nuances. In this paper, we highlight these challenges and how Machine Learning (“ML”) can be applied to overcome them. It is by no means an exhaustive list, but we hope that the lessons and examples can be used as a guide to better understand the role that ML can play in buy and build origination and how critical system architecture is to solve the problems properly.

Platform company origination – Finding the diamond in the rough

Competition for platform investments is fierce, especially in today’s markets where capital is expensive and investment firms are increasingly selective with their unallocated dry powder: year-over-year deal volume was down 39% in Q1 2023 (see figure 1), but at the same time, global private equity dry powder hit a decade-long high $3.7 trillion (source: Bain & Company) at the end of 2022.

Figure 1 – US private equity deal activity by quarter (source: Pitchbook)

These facts suggest that firms are likely facing the most competitive market that they have seen in the last decade and it has become imperative to accelerate deal origination processes, especially when searching for a solid platform investment. For private equity firms, that starts with finding creative ways to identify opportunities before competitors do.

One way that firms try to get ahead is by allocating investment teams to scan markets for soft signals.

A soft signal is an early, usually qualitative, indicator that highlights a potential investment opportunity. They can indicate a change in a company's leadership team, entry into a new market, a key partnership or collaboration announcement, or a change in strategic direction, among other things.

If identified correctly and quickly, soft signals help firms act on opportunities ahead of the market. However, monitoring soft signals is a labor-intensive task when done manually. Analysts have capacity to monitor about 10-15 companies thoroughly at any given time but small to medium sized firms simply do not have the horsepower to monitor their entire target list in this way.

To address this constraint, firms are exploring artificial intelligence-powered tools to help them identify soft signals from unstructured data. In doing so, they enable monitoring of a much larger company universe and allow their investment teams to do what they do best: apply rigor and judgement to a smaller number of opportunities.

Case Study

We worked with a mid-sized firm who wanted to implement their own ML solution to monitor markets for 15 – 20 different signals across 5,000 companies. One of the signals that was important to them was “expansion into new geographies”. Waiting for a press release that says “Company X announces entry into Europe”, was a reactive approach and did not allow the firm to act quickly enough. We helped them build a monitoring tool that ingested text data from multiple sources (e.g. Dow Jones, CapIQ, press releases, website, etc.) and scanned all the companies in their supply chain for partnership / collaboration announcements. We could then cross-reference those announcements with the geographical presence of the new partner to estimate the likelihood that the company was trying to expand operations into a new region. Investment teams could then take these signals and perform more targeted analysis on the platform investment opportunity.

Platform company origination – Automation without an identity crisis

Firms are leaning heavily on their investment theses to drive differentiation and are using them to help focus their investment teams’ efforts. At the same time, they need to be careful about how they allocate capital given rising interest rates and economic headwinds. Tactically, this means firms may try to diversify portfolios with smaller investments in a larger number of companies. The problem particularly for small to medium-sized firms, is that increasing deal volume forces them to consider a difficult trade-off. They can either scale up the size of their teams by hiring more analysts to scan for soft signals or they can use out-of-the-box tools to automate company screening. The former is expensive, and the latter forces the firm to sacrifice the nuances of their investment strategy in favour of overly simplified, rules-based filters that SaaS platforms provide.

ML gives firms a third option that automates soft signal monitoring without having to abandon the nuances of their strategy. The solution usually requires multiple classification models that can ingest and process unstructured (e.g. news articles, emails, transcripts) and structured (e.g. financial data, sales volumes) data to automatically identify signals that are specific to the firm’s strategy.

Through our experiences implementing these models we have learned that their value is compounded when we architect a system that combines multiple signals, investment criteria, and rules in a way that matches a firm’s strategy.

Case Study

We worked with a North American firm who wanted to completely encode their investment strategies into an ML system. One of these strategies was to focus solely on tech-enabled companies, who needed less than $25MM in funding, with a strong leadership team, and growing brand recognition. Automating this requires a system that is much more complex than a single model that classifies a news article into one topic or another. It required a combination of logic rules (i.e. if / else logic) and various types of ML models (e.g. regression, binary classification, multi-class classification) to produce decision trees; one for each of their “investment hypotheses”. Once each tree was defined, company data could be fed through each rule and ML model and the system produced a list of opportunities that were most aligned to the firm’s investment strategy.

Bolt-on company origination – not just a matter of patience and timing

When performed manually by an investment team, bolt-on origination looks very different from platform company origination. A short list of obvious bolt-on targets is provided by industry subject matter experts, that short list is expanded by analysts who search their company universes to find targets that have similar qualities, and that expanded list is monitored until the timing is right to make an approach and kick-off a deal process. To make matters worse, competition intensifies because a firm is no longer only competing with other private equity and investment firms. They are also up against strategics who are interested in acquiring the target for integration into their own operations. In small to medium sized firms team capacity quickly becomes a constraint and the reality is that buy and build success relies heavily on the team’s ability to move quickly, before their short list is depleted.

Applying ML changes the arithmetic of this problem and the shape of the opportunity funnel. Systems powered by rules and ML can be scaled to tens of thousands of opportunities that cover a platform company’s entire ecosystem (e.g. suppliers, competitors, customers, similar companies in adjacent verticals, etc.). If a firm has the capability to sort through bolt-ons at this scale, their teams can focus on evaluating opportunities, narrowing down the list, and picking the most valuable ones while they leave the work of finding and monitoring to an intelligent system. To provide value in the origination process the system must have the capability to:

1.     Constantly monitor known targets for soft signals. Potentially specifically for signals that indicate the timing is right to approach the target about a buy-out (e.g. founder or CEO approaching retirement age).

2.     Use an existing list of bolt-on opportunities to find new prospects. More on this below.

If a firm is looking for a bolt-on to a platform company, they have likely already leveraged their network of industry experts to create a list of obvious bolt-on targets (these are the companies that a firm will be monitoring for soft signals regularly). Where they are likely struggling, is putting together a list of additional, less obvious, targets. These additional companies might not be a “perfect” match on paper but with some changes to strategy, structure, or operations they could be a strategic match for the platform company. The challenge is finding enough “similar companies” to those already on a target list so that we increase the likelihood of finding a good bolt-on. Like soft signal monitoring, finding similar companies is a time-consuming activity for an investment team that does not scale well to an enormous company universe. But it becomes a much more efficient one if you can cut through the noise and give teams a prioritized list of opportunities to look at. The task that ML can help with in this case is scanning an enormous company universe to find the opportunities that are “most similar” to the list of bolt-on targets that you already have.

Example

Imagine that you have a list of 25 bolt-on targets that you know would be a good fit for your platform company but do not appear ready for buy-out in the coming years. Your job now is to find other targets who are as similar as possible to the existing list. You start by articulating what you like about the existing 25 companies (e.g. product offering, geography, revenue growth, team size, etc.) and the system assigns those attributes to the 25 companies. Next you feed the system an enormous company universe that is much larger than your team could cover manually. In this case an unsupervised ML system is built to evaluate “how similar” each new company is to the original 25 and it will produce a list of the most similar ones. It can also score how similar the new company is to your existing list and highlight the attributes where it is similar or different.

Buy and build strategies are complex and executing them requires a fine balance of art, science, and timing. Resource prioritization is critical, especially for small and medium sized firms, and this is where we have seen custom ML systems add the most value. These systems are not meant to replace the role of the investment team or even change the team’s approach. But they certainly can help scale the team’s approach to larger universes of platform investment opportunities, encode nuanced investment strategies to produce quality short-lists, and uncover bolt-on opportunities from parts of the company universe that were not covered before. In upcoming perspectives, we hope to provide some additional insight into the data landscape for private companies as it is a critical component to build a high-performing ML system.

Running a buy & build strategy for your firm? Get in touch

Appendix

What is a buy & build strategy

A buy and build strategy in private equity refers to an approach in which a private equity firm acquires a company, then leverages that acquisition to make additional acquisitions, all with the goal of creating a larger, more valuable entity that can eventually be sold for a significant return on invested capital.

Step 1: Finding a platform investment

The first step in a buy and build strategy is identifying a suitable platform company to acquire. This is typically a company that is well-established in its market, has a strong brand, loyal customer base, and a successful track record. The private equity firm will then work closely with the management team of the platform company to identify potential targets for acquisition.

Step 2: Finding bolt-on targets

The next step is to identify and acquire complementary businesses that can be integrated into the platform company. These businesses may be competitors or suppliers, or they may operate in a related industry that can add value to the platform company's offerings. The goal is to create a larger, more diverse entity that can benefit from economies of scale, increased market share, and improved operational efficiency.

Step 3: Integrating the two companies

Once the acquisitions are complete, the private equity firm will work with the management team to integrate the new business into the platform company. This may involve streamlining operations, consolidating supply chains, and rebranding the newly acquired businesses to fit within the larger corporate structure. Integration execution and deciding what parts of the bolt-on company to retain as-is, integrate, or discontinue depends largely on the strategy for the combined entity, the dynamics of the industry the company competes in, and the capital market environment at that point in time.

Over time, these three steps may be repeated as the entity continues to grow via additional acquisitions. This may involve entering new markets or verticals, or acquiring businesses that can offer new products or services to the company's existing customer base.

There are several advantages to a buy and build strategy in private equity. For one, it allows private equity firms to quickly and efficiently build a large, diversified company that can compete more effectively in its market. It also allows for greater operational efficiency and economies of scale, which can lead to higher profit margins.

However, there are also risks associated with a buy and build strategy. One major risk is overpaying for acquisitions, which can lead to decreased profitability and decreased shareholder value. Additionally, integrating multiple businesses into a single corporate structure can be a complex and challenging process, requiring significant time and resources that distract from the core operations of the company.

Ultimately, value is returned to shareholders when the combined entity exits through an IPO, sale to a strategic buyer, or another M&A, where the company’s shareholders are bought out typically by a much larger entity.