4 MINS READ
M&A transactions have always been an integral part of corporate strategy.
And that’s how it’s been since the birth of the modern corporate merger in 1708, when the East India Company joined forces with an upstart English competitor to preserve its global trading monopoly. Used to eliminate competitors, grow capabilities, achieve economies of scale, and reduce costs, the right deal can mean the difference between market dominance and market irrelevance.
However, not all M&A deals work out. Famously, when AOL merged with Time Warner in 2000, it led to the destruction of more than a hundred billion dollars of shareholder equity. Likewise, in 2005, when Sprint acquired Nextel for close to $38 billion dollars, it created the world’s third-largest telecommunications provider. Yet only three years later, Sprint Nextel stock would be reduced to junk status. M&A dealmakers face constant challenges ranging from geopolitical complexity to regulatory hurdles to due diligence minefields to integrational challenges. What’s worse, countless potentially successful deals fail to be realized for the simple reason that no one ever thinks to suggest them. Clearly, the C-suite needs guidance, and the good news: help is at hand.
M&A dealmakers can utilize AI algorithms that excel at making accurate predictions.
Generative machine learning models—models that can automatically discover and learn the regularities or patterns in input data and produce new data consistent with those patterns—have been having a moment recently. Whether defeating grandmasters at chess, returning stunning photos in response to text prompts, or powering chatbots that can pass the Turing test, AI has demonstrated a remarkable talent for making highly accurate predictions about what should come next in a given sequence to achieve a given objective. All they require to accomplish this feat is past examples of input-output pairs—many past examples—and from this information, they can work out which inputs have predictive power and which actions will result in desired outcomes. To illustrate this process in action, here is a simplified example of how an AI “thinks” about the next move in a chess game:
The days of most deals originating solely from investment bankers mixing and matching buyers, sellers, and acquisition targets will soon be over, and ultimately, AI will provide deeper insights and probabilities of deal success.
At its heart, M&A deal origination is also about making scenario-based predictions.
Like which companies when combined will form a more profitable entity. Or which companies possess individual elements that would be worth more if sold off separately? Although corporate structures and capabilities are immensely more complicated than the positions of pieces on a chessboard, the underlying principles remain the same. Given initial conditions, certain actions will be more or less likely to produce certain outcomes. The question, then, becomes when, not if, generative AI will drive M&A origination due to their unrivaled talent for making accurate predictions. In another simplified example, here is how an AI might successfully evaluate a potential merger or acquisition.
The only thing separating the present from this AI-enlightened future is data. Specifically, the lack of usable training data. When the inputs needed to build a machine learning model are as straightforward as the location of pieces on a chessboard, it’s relatively easy to input all the necessary information. However, corporate organizations defy such simple descriptions. While chessboards have thirty-two pieces placed on any of sixty-four squares, corporations comprise hundreds, thousands, or even hundreds of thousands of employees, plus physical and financial assets that span the globe. They possess all sorts of varying capabilities (manufacturing advantages, IT structures, distribution channels, sales partnerships, products, brands, leadership, and culture) and even within these capabilities, nuances abound. Kodak and Blockbuster, for example, were brand leaders within their field—until they weren’t.
Before AI algorithms can work their magic to tease out which corporate capabilities are predictive of M&A success, they need a way to describe those specific capabilities in the first place. In other words, they need to identify the attributes and strengths of the companies in question. Although the data required to measure corporate capabilities will be difficult to collect, store, and process, it is definitely possible.
“It’s only a matter of time,” says Alexis Christofides, UK Regional Head for TCS M&A Services. “Increasingly, companies are discovering ways to codify their capabilities, even for traditionally unquantifiable things like corporate culture. Sooner or later, all that data on ‘e-mail sentiment analysis’ and ‘product time-to-market’ will be thrown into a machine learning algorithm.”
In practice, the algorithms need not understand that they are measuring “company culture” when they input, for example, a hundred different data points on “median employee tenured stay” and “duration of new employee onboarding.” Once corporate capabilities have been roughly approximated, it becomes easy for machines to connect inputs (company attributes) with outputs (the results of an M&A deal) and discover a formula to accurately predict which organizations will combine most successfully.
Banking on AI for M&A success.
“Humans will still have a role to play, of course—no model is foolproof, no large dataset is without errors,” notes Christofides. “But the days of most deals originating solely from investment bankers mixing and matching buyers, sellers, and acquisition targets will soon be over, and ultimately, AI will provide deeper insights and probabilities of deal success.”
Not long from now, it will be algorithms reading the corporate tea leaves, guiding the origination of front-page, multi-billion-dollar M&A transactions, and no CEO anywhere will think to consecrate a deal without a machine validating—or suggesting—it first.
Don’t let the future catch you unprepared.