If you want to become a linguistic maven, or the life of a party, read Made in America – An Informal History of the English Language in the United States, by Bill Bryson. The book is saturated with interesting stories about the origins of everyday words like dashboard, refrigeration, airplane, radio, aquatics, megabyte, jackhammer, jazzercise, microbrew, and groovy. The author reminds us that these words were once as novel as IoT and Gesture Tech are today.
After reading Made in America, I take few words for granted. Lately, I thought about optimize, a word I encounter with amazing frequency. Its current meaning, “to make the most of,” had its first recorded use in 1857. It languished for about 100 years, before becoming more prominent in the 1950’s. Today, optimize and its derivatives have soared into ubiquity, rivaling trendy expressions like paradigm shift, synergistic, and transformational change.
I didn’t need to look hard to find examples. In biz-dev alone, I found “Optimize customer experience.” “Optimize customer loyalty.” “Optimize your website.” “Optimize your push-notification strategy.” “Optimize visual storytelling.” “Optimize customer retention programs.” “Optimize your call center conversion rates” . . . The list stretches over multiple pages.
I first encountered optimize in 1978, in an operations research course titled Quantitative Analysis for Management. Our textbook carried the same name, and its graphics were limited to plain statistical charts. Other than the cover, which sported silver lettering on a gray background, the book was devoid of color. Clearly, the authors intended to suck all excitement from their topic. In that sense, this book was impeccably honest. When done correctly, operations research suppresses the flow of adrenaline.
We began by manually grinding out matrix manipulations for simultaneous algebraic equations. As our professor explained it, these expressed simulations of real-world problems, and the mathematical approach for optimizing variables was called the Simplex Method. This must have been a marketing moniker, because I didn’t find the Simplex Method very simple. But the professor, a logistician for the US Department of Defense in his pre-academic life, believed that manually slogging through matrix math would be helpful for understanding what computers do when performing optimization calculations. An approach that proved correct.
“In mathematics, optimization problems involve a quantity that we are trying to make as big (or as small) as possible, subject to some constraints, or boundary conditions. The constraints might be time or available resources,” Eugenia Cheng wrote in her column, Everyday Math (Wall Street Journal, March 18, 2017). In B-school, I expected a hefty dose of revenue, profit, scrap, or cost among the variables to be maximized or minimized. Quantitative Analysis for Management didn’t disappoint.
When [Stuff] happens, Operations Research gives you guidance about how to proceed. Here’s where things got fun. After presenting a problem, the professor changed constraints such as raw material cost, lead-time, cube, and weight. “Refer to the original problem above. The factory in Arkansas no longer produces the titanium alloy lock nut. The VP of Operations has asked you to consider using a less-expensive Korean substitute, but the part takes three times as long to receive, and the sample shipment was 28% defective. Should Company A source the part from Korea?” The “right” answer was not always evident, and it wasn’t unusual to find my classmates promoting widely-varied recommendations.
Often, key information was missing from the problems. What was the cost of the Korean part? How many lock nuts would have to be procured to compensate for the expected defective parts? We were forced to figure out what else we needed to know, and to ferret out the right information. And when we couldn’t, we formulated assumptions – which we always had to justify.
A bumpy path. But in this way, we discovered that new conditions spawned new, downstream optimization challenges, requiring more number crunching and more analysis. In fact, this happened for every problem. Even small changes in constraints or variables had potential to upend a carefully prepared optimization analysis. Mercifully, the professor soon had us abandon longhand calculations, and we moved on to use Linpro, an optimization software subroutine. Linpro made it possible to complete the course in one semester, and had the added benefit of preserving our sanity.
Importantly, our mathematical results were incapable of conjuring certainty. And they never decided anything. If anything, our results fostered additional questions. “Following the supply chain interruption, what will be the consequence for delivery lead times and customer service? Should Company A continue its strategy to be a premium-quality producer? Explain your recommendations.” I quickly discovered that optimization demanded grappling with thorny questions that were massively more time consuming than the manual number crunching.
I reveal these details about my background with optimize to highlight a complaint: many people perfunctorily proclaim “optimize!” without understanding or acknowledging the large ancillary issues that inevitably accompany the effort:
1. Tradeoffs. Optimizing anything requires sub-optimizing something else. Want to optimize customer experience? First, determine what you’re willing to sacrifice. Shortening service queue time will probably increase labor or IT costs. Optimizing order fill rates usually means boosting inventory investment as well as carrying costs. And optimizing customer retention can mean giving up a portion of profit margin.
2. Interpretations and recommendations. Optimization is a decision, not something given as an admonishment, or presented as a foregone conclusion. Five Reasons You Absolutely Must Optimize Your Website for Mobile. Sound advice for some companies, when given as a recommendation based on context. But optimizing a mobile website doesn’t apply to every business.
3. Ongoing calculation and analysis. Because conditions change, optimization problems are never fully solved.
4. Project reprioritization. Does optimizing customer loyalty make sense if meeting that goal means tanking your profit? That sounds like a rhetorical question, but it’s not. Many executives maintain slavish devotion to “optimizing,” and when conditions change, they fail to question the efficacy.
5. Strategic realignment. When corporate strategy is to be low-cost producer, do protectionist politics add risks that are too costly to absorb? Optimization analysis sometimes reveals the need to change strategy.
6. Inflection points. More optimization isn’t necessarily better. Optimization rarely continues ad infinitum. In a key piece of insight, Eugenia Cheng wrote, “If the manuscript I’m revising improves very quickly at the beginning of my work before I hit diminishing returns, then the outcome-to-effort ratio will increase to a maximum and start decreasing. That is the point where I should stop.”
When it comes to optimizing, there are few guidelines for when it makes sense to reevaluate, curtail, or discontinue. That’s where I get most bothered about optimization hype. When hotly pursuing an optimization goal, many executives blow past the point of diminishing return, which can sometimes be at the outset – or earlier. Compounding the problem is the steady, incessant drumbeat of messages from vendors and consultants that says, in effect, “the thing I sell you optimizes [fill in name of result de jour]. So, don’t stop! Keep optimizing!”
That’s folly. Making the most of something sounds great, but whether it’s worthwhile or valuable depends on context. Optimization is a decision tool that considers and reflects forces that exert pressure and risk on a company’s strategy. And it helps managers adjust and respond when conditions change.