Clayton Christensen was exceptional, both personally and professionally. He impacted the world on many levels, and today I’d like to focus on how a narrow aspect of his work has proven to be amongst the most profound: profit-seeking.
At its most basic, Christensen restated what we already know – businesses want to make money.
(Although that’s sort of like saying Newton merely restated what we already know – things fall.)
By deepening our understanding of profit-seeking, Christensen moved the world ahead in its ability to understand markets and predict business success (or failure). He discovered order laying beneath chaos. In doing so, he also exposed shortcomings across vast swaths of business decision making.
Hidden in Plain Sight
Everyone knows businesses want to make money. Yes, nonprofits may be different, at least in principle. Yes, some businesses (ex. Amazon) are happy to sacrifice near-term profits in exchange for longer-term adoption and market dominance. Yes, businesses can also have social missions. That said, I hope we can at least agree that, for the most part, businesses live or die by their abilities to eventually make money. There may be exceptions, but this is the rule.
Christensen revealed how profit-seeking gives us clues about what businesses do, and don’t, want to do. It helps us predict what investments they’ll be more, or less, willing to make. What changes they will or won’t be likely to adopt. What strategies they will or won’t be likely to pursue. What competitive battles they will or won’t be likely to win.
For example, he found big incumbent companies are likely to defeat new competitors that threaten them with higher performance products. Big businesses aggressively defend high-end profits. However, those same big incumbents are less likely to fight new competitors who offer lower-performance products because they don’t pose as obvious of a threat. It’s hard to overstate how far these discoveries opened the door to better modeling of firm behavior and more accurate predictions of business success or failure.
A Better Conversation
These discoveries by Christensen laid the tracks for the discipline of quantitative probability to enter (and challenge) the highest levels of strategic decision making that were historically the domain of human intuition. By showing how profit-seeking behavior causes big incumbents to prioritize certain battles, thus lowering likelihood of success for new entrants, it begged the question “how likely is it to lower a new entrant’s success?” 5% likely? 50%? 99%?
Christensen’s work ushers the conversation towards one of probabilities. “Under the circumstances, how often does X happen, instead of Y”? This is the language of statistics. Actuarial science. We’re being asked to measure the likelihood of one outcome versus another – not based on how we feel, but on evidence.
For example, to figure out how much a startup’s odds of success are lowered by going to market with a higher performance product, you’d need to start logging variables about businesses, and their strategies, to be compared with statistically meaningful samples of other businesses with and without similar variables. Definitions need to be set. Results need to be measured. Control groups need to be organized.
Uh… wait a minute…
This is where some people get a little uncomfortable. In principle it seems okay for strategy decisions to be influenced by probabilities. After all, it’s how insurance actuaries and pension plans often make decisions. Yet if you’re like most people you’re also a little worried, with good reason. For example, would you be comfortable investing in a startup just because a statistical model told you to? What if a model can’t really quantify what’s important? What if a statistical model is wrong? What if it only works in the past, and can’t account for unknowable futures? What if the market is, in reality, unpredictable and governed by chaos?
First, I’d like to point out a double-standard.
Nobody asks these questions when a startup is funded because investors get a good feeling about the founders. Nobody asks what the hard criteria for a “good” and “bad” founder is, how it’s measured, whether those variables were compared with statistically meaningful samples of other startups with and without similar variables. There’s no mention of control groups. If it worries you that an actuarial approach could lead to errors, you should be just as worried that an intuition-based approach could lead to the same errors. Your worries are legitimate in either case.
It’s also worth pointing out how algorithms tend to be more explicit, easily audited and improved than intuitive feelings. Intuition is, after all, just a person’s mental algorithm coupled with a body’s psycho-physiological reaction. Measurement and improvement are the lifeblood and day-to-day work of actuaries and data scientists, whereas I’ve never heard anyone ask an executive to prove how reliable his or her gut intuition is, backed by meaningful sample sizes with satisfactory levels of statistical confidence.
Second, the future may be impossible to predict with perfect clarity, but that doesn’t mean probabilities can’t help.
Probabilities help in countless facets of human life – you’re a believer even if you don’t realize it. When you get in a car, you’re relying on the convergence of staggering probabilistic analyses to deliver you safely to your destination. The engineers calculated the probabilities for every piece of your car under a wide variety of conditions in order to build it with a low probability of failure.
In this car example, you may not know what the exact odds are, but you’re willing to bet your life on them because of your own knowledge and experience with the vehicle. In other words, when you get in a car you’re making a forward-looking prediction yourself, based on your personal sense of the odds you’ll survive. If you knew a specific car model blew up 80% of the time, you probably wouldn’t get in. Models can be wrong and incomplete – those are bad models. Models can also be accurate and very complete – those are good models. I’m only advocating the good models, and that businesspeople make the effort to tell them apart.
Third, while in some ways it’s easy to feel as if the market is governed by chaos, nobody I know actually behaves as if this were true.
If it was pure chaos, there would be no need for business schools, strategy or planning of any kind. You’d just flip a coin every time there was a decision to be made. In practical terms, human decisions are inherent predictions about likely cause and effect, even if it’s just your best guess.
A Universal Constant
This isn’t to say business and markets can’t be very, very complicated. They certainly can be. Business can involve daunting spectrums of customers, competitors, stakeholders, ecosystems, technologies and transactions. Modeling and even predicting anything in such complicated systems is hard, at best.
This is where Christensen’s articulation of profit-seeking behavior is again useful and enduring. In the massive swarm of complexity that surrounds most businesses, the pursuit of profits is as close to a universal constant as you’re likely to find. In other words, markets and business might be chaotic – if not for profit-seeking.
In business we may not know much, but we know our customers, competitors and everyone else in the ecosystem is doing their best to make money. We may not know exactly what path they’ll take, or if they’ll survive the journey, but at least we know where they’re trying to go. Knowing the destination makes it easier to anticipate which routes others will or won’t take – bringing a degree of order to the system.
Said differently, the assumption of profit-seeking behavior brings some order and stability to market systems, and can be used as the basis of better market models, more accurate predictions and better decisions.
Christensen’s elegant work on profit-seeking has been the foundation of multiple breakthroughs in our efforts to model markets and predict business success using big data, machine learning and advanced computing. From my perspective, he provided a framework and set of game theory rules that have continued to shift market mysteries from unknowable to known, or at least knowable.
It’s also worth noting that improvement, not perfection, is the appropriate standard when measuring the usefulness of this work. I bring this up in the context of Christensen’s contributions because there are exceptions to the rules. Anomalies. Some businesses succeeded, despite what Christensen’s work would have predicted. Others failed, despite what would have been predicted. For what it’s worth, we’ve found the combined gross accuracy of Christensen’s core methodology to be vastly more accurate than the historical batting averages of corporate innovators, startups and venture capitalists alike.
Rather than being ignored or dismissed, Christensen himself was quick to embrace anomalies, seeing each one as an opportunity to improve, adapt or refine his work. Every time a theory evolved to account for more anomalies, it was made more robust and nuanced. That’s the essence of the Scientific Method and intellectual honesty. As a result, his work continues to evolve and endure, laying the groundwork for generations of future market models, statistical analyses and deeper understandings of the world we live in.