This article was originally published by Kevin Troyanos, SVP Analytics, Saatchi & Saatchi Wellness on the SSW Analytics blog. You can follow them on their blog here or on Twitter @saatchiwellness.
In its strictest sense, “optimization” is a mathematical principle — one that marketers often misapply.
Back in 2013, lovers of language everywhere underwent a brief existential crisis when esteemed dictionaries like Merriam-Webster and Cambridge officially adopted a second definition of the word “literally”: “in effect; virtually — used in an exaggerated way to emphasize a statement or description that is not literally true or possible.”
While, as Merriam-Webster points out, “the use of literally in a fashion that is hyperbolic or metaphoric…dates back to 1769,” there’s an argument to be made that formalizing this largely colloquial — and self-contradictory — usage can only result in unnecessary confusion.
In the marketing world, there’s another term whose meaning has slowly been eroded by broad-based misusage: “optimization.”
Marketing’s Terminological Co-optimization
Strictly speaking, optimization is “an act, process, or methodology of making something (such as a design, system, or decision) as fully perfect, functional, or effective as possible.” It’s a term most properly used in mathematics, wherein practitioners solve “optimization problems” as a means of selecting the maximal (or minimal) solution from the set of all possible solutions. Given an objective function subject to a specified set of constraints, a mathematician can be said to have optimized the function once they have solved for the minimum and/or maximum value possible within the constraints.
And while marketing can — and, in the wake of the digital revolution, should — be considered a science, it is a decidedly “soft” (or “social”) science — more akin to psychology or sociology than chemistry or physics. As is the case with all social sciences, marketing contends with human behaviors, a set of phenomena which are profoundly unpredictable.
Whereas once a mathematics problem has been optimized, the solution will remain unchanged, ceteris paribus, once a marketing problem has been optimized (inasmuch as it can be), the solution is liable to change in response to even the slightest systemic fluctuation. Humans tend to tire of repetition, meaning an ad or a marketing channel that worked today might not work tomorrow, even if nothing but the date on the calendar has changed.
Ultimately, optimization requires a solvable problem, and a solvable problem requires a situation with a clearly defined set of consistent parameters. Such conditions are attainable in largely theoretical disciplines like mathematics, but they’re little more than a fantasy in messier, more “lived in” fields like marketing.
The Limited Applications of Optimization in Marketing
That said, there are circumstances in which marketers are able to engage in bonafide optimization — especially when prescriptive analytics are involved.
Generally speaking, marketers use prescriptive analytics to simulate a large number of contingencies and determine which course of action will deliver the best ROI for a campaign. These algorithms are not a crystal ball — neither man nor machine can ever foresee the future with clarity — but they provide marketers with an evidenced-based idea of which approach is most likely to deliver results given the current circumstances and the historical record.
Prescriptive analytics cannot optimize an open-ended marketing problem, but given a defined set of contingencies, they can find the best possible course of action for a marketer to take. In this sense, it’s not inaccurate to claim that a (theoretical) marketing problem has been optimized.
Of course, in reality, marketers are always working with a constantly shifting set of variables, meaning real time optimization is effectively impossible. To optimize a problem, a marketer must set their conditions in stone, but due to the unpredictability of human behavior, these conditions are wont to change while the marketer is solving the problem. In other words, a marketer can optimize a “snapshot” of their market landscape, but this optimal solution will only be accurate for a brief interval.
A More Accurate Alternative
Real time optimization in marketing would require solving multiple optimization problems simultaneously in perpetuity, a task which demands processing power far exceeding that of the human mind or a standard computer. With the rise of machine learning algorithms in conjunction with simulation techniques and evolutionary algorithms, this calculus may be on the brink of a radical reconfiguration, but for the time being, marketers must accept that little of what we do is literal optimization.
Rather than further dilute the meaning of “optimization,” we should consider incorporating an alternative term into our lexicon: “continuous improvement.” Unlike mathematics, marketing does not always present a solvable problem. Indeed, our jobs are so interesting precisely because we must always be on our toes, ready to respond to and analyze the marketplace’s unpredictable sea changes.
Every time a new variable enters the equation, we tweak our messaging, we respond to market forces, we continuously improve our overall approach. Doing so effectively requires studying trends, understanding news cycles, and utilizing history just enough to identify things that resonate over time. This is a science unto itself, but at the end of the day, it doesn’t amount to “making something as fully perfect, functional, or effective as possible.”
WRITTEN BY KEVIN TROYANOS
I lead the Analytics & Data Science practice at Saatchi & Saatchi Wellness. I have focused my career within the healthcare marketing analytics space, empowering healthcare marketers with data-driven strategic guidance while developing innovative solutions to healthcare marketing problems through the power of data science. I’ve worked to measure, predict, and optimize marketing and business outcomes across personal, non-personal, digital, and social channels. I’ve led engagements with brands that span all stages of the product lifecycle, with a particular focus on established brands. My role is to guide the departmental vision and lead innovation initiatives, effectively positioning marketing analytics as a competitive differentiator and organic growth driver for the agency at large.