The Butterfly Effect: Non-linearity, Marketing Mysteries, and the Human-Machine Partnership
The butterfly effect, the whimsical term for non-linearity, rouses frustration among executives who expect marketing to be logical. For complex systems such as weather and markets, non-linearity means actions and outcomes are disproportional. Non-linearity makes marketing mysterious. There’s no escaping the butterfly effect, but a partnership between humans and machine intelligence creates an effective way to surf the enigma.
What is non-linearity?
In 1961, MIT professor Edward Lorenz made a startling discovery. Lorenz was applying emerging computer power to the age-old challenge of climate prediction. His study attempted to discover how factors including temperature, atmospheric pressure, wind, and topography caused weather change. No longer would pilots and picnickers be flummoxed by weather, scientists were confident. Computing machines would ensure that future flights and excursions could be planned to perfection!
Following eight frustrating years of little progress, Lorenz repeated yet another simulation. But on that day, he took a short-cut. Using results from yesterday’s printout Lorenz started the same simulation mid-way through. Computers were slow in ‘61, so he went for coffee, expecting to return to a duplication of yesterday’s results. However, when Lorenz got back, he was shocked to find that results from the two simulations diverged wildly. After weeks of investigation, Lorenz tracked down the reason. Lorenz had discovered non-linearity.
Non-linearity describes conditions where there is no straight-line or direct relationship between one variable and others. Machines perform linearly. Changes to inputs result in similar changes in output. Lightly tapping a car’s accelerator increases speed a little. Pressing pedal to the metal increases speed substantially.
Lorenz’s accident revealed that natural systems don’t work like machines. Data used in his two simulations were not identical, as he had intended. They were just a teeny bit different. The computer’s memory stored value out to six decimal points. However, the reference printout used by Lorenz recorded only three. If weather worked like a machine, as scientists had earlier assumed it would, the outcome difference caused by this rounding error would have been inconsequential. Instead, Lorenz’s simulations careened in irrationally different directions.
Weather is an example of what scientists call a complex system. Non-linearity, in varying degrees, is an attribute of all complex systems. Changes made in inputs are disproportional to outputs. Small changes made early in a complex system process can result in huge changes later. Non-linearity can cause other lopsided outcomes. A huge initial change could result in a small impact later or no perception of change at all. People nick-named Lorenz’s discovery the butterfly effect after the poetic example Lorenz used as the title of his 1972 talk, Does the flap of a butterfly’s wings in Brazil set off a Tornado in Texas?
Non-linearity makes marketing mysterious.
Markets are also complex systems, more like weather than machines. Every natural and social system is a complex system, including ocean eco-systems, beehives, cities, military theaters, and traffic. A market’s complexity results from the interconnected and adaptive interactions of many independent agents - customers, companies, influencers, social networks, partners, and regulatory agencies. These interactions produce feedback loops. Situations change rapidly. There are many unknowns.
While markets have always been complex, digital exploded complexity. The number of interactions increased. Market participants became more entangled, and feedback accelerated. Digital dissolved the time and distance that had buffered humans from some of non-linearity’s effects. Markets became more volatile.
Because of non-linearity, anyone who expects marketing to be logical will be disappointed. Predictions and correlations can be made, but only within bands of probability. Non-linearity means marketing spending and outcomes will never consistently pair. A customer journey functions more like a child’s scribble than an orderly revenue funnel. Although much of a market’s behavior can be relatively stable in the near term, situations inevitably fluctuate. Change will be dramatic on occasion and in rare circumstances, shocks will disrupt violently.
Read: Because VUCA: How Complexity Causes Persistent Marketing Problems
Humans and machines – perfect partners for marketing’s non-linearity
We now accept the uncertainty of weather. As Edward Lorenz discovered, analytic systems make excellent partners to help humans with important tasks necessary to thrive in a complex world. A quick look at the weather app on your phone predicts the possibility of rain two weeks out. During volatile seasons, we check the app more frequently to keep on top of change. We have more insight into weather now, but we still understand that as we move out in time, the picture gets murkier.
Executives must adopt a similar mindset for markets. Accept marketing will never be a machine. All the data in the world will not tame non-linearity. Gain competence in working with probabilities. Think like an investor. Evaluate returns on marketing programs within the context of the ever-changing market. Use metrics as directional tools to answer the question “are we headed towards where we want to go?” Hold objectives loosely and be ready to pivot.
A human-machine intelligence partnership is one of the most effective tools for surfing the marketing enigma. In many ways, humans are well suited for working in an ambiguous world. Humans evolved within nature’s complex theater. We can sense the meaning of a slight pause or change of inflection in a customer’s answer. We can make value judgments about whether to trade a free service today for potential revenue next quarter. We can improvise and innovate in the face of new situations. Yet the same cognitive tools that help humans thrive under nature’s complex conditions have limitations. Just like our brains alone can’t figure out the weather, humans are unequipped to sort through the mountains of data required to find the patterns inherent in complex market systems. Human thinking is also subject to many cognitive biases.
“Augmented intelligence blends human and machine intelligence in a redesigned business process to achieve better outcomes than either humans or machines could achieve alone.” Judith Hurwitz, Henry Morris et al, Augmented Intelligence: The Business Power of Human-Machine Collaboration
Enterprises that will succeed in the complex digital era will be those that best harness the uniquely human potential for performing well in ambiguous and nuanced situations. One important enablement tool is machine intelligence. Machines can overcome limitations in human information processing. For example, analytic systems can accommodate the time lag challenge of the long B2B sales cycles by incorporating time-series analysis and historical data. The objectivity of machine intelligence helps counter cognitive bias. Analytics reveal patterns of reality even when these patterns don’t make sense to humans. One marketing executive reported that analytics discovered a strong correlation between deal conversion and buyers accessing service-related web content. This content was nowhere on marketing’s radar.
Machine intelligence complements human strengths and limitations. Machines perform best on well-defined tasks and function poorly when situations are ambiguous and nuanced. Artificial intelligence can assess the risks of holding an outdoor event in late October but can’t tell marketers whether that risk is worth it or how to mitigate that risk. In his book, Superintelligence: Paths, Dangers, and Strategies, Nicholas Bostrum provides an example of an autonomous vehicle given the task of finding the fastest route to the airport recommends crashing through roadblocks and smacking pedestrians.
For marketing to get the most value from the human-machine partnership, the right analytics style must be used. Analytics tools and methods that attempt to establish firm, cause-and-effect relationships between outcomes and tactics lead to a false sense of confidence. Although it may not be necessary to employ the compute-power needed for weather prediction, more sophisticated analytics are needed for useful results in complex markets. Special care must be taken with getting the data right. The interconnectedness of various marketing channels and programs creates challenges. However, with sufficient data and the appropriate advanced analytics, it is possible to find useful directional relationships between sales outcomes and the highest impact marketing tactics. Artificial intelligence will continue to become more sophisticated and make these insights even more useful.
Machine intelligence alone won’t be sufficient to optimize human potential. Effectiveness in the complex, non-linear world also requires modified organizations. Leading enterprises are forming semi-independent, multi-discipline teams accountable to a customer-centric mission. They support these teams with a collaborative network delivering guidance and service. Agile methods contribute a more complexity-appropriate operation. Leadership paradigms and roles foster conditions, culture, and purpose that bring out workers’ best selves.
The butterfly effect is alive and kicking. If we can’t control the mysterious currents of weather and markets, at least we can build better surfboards.