In a data-led world, intuition still matters

by Christopher Frank, Paul Magnone, and Oded Netzer (Wiley, 2022)

For as long as there have been decisions, people have used facts (lately known as data) to make them, along with reason and intuition. Nowadays, though, data often seems to be making decisions for us. Just as ‘analytics’ has changed the face of sports, business data prompts computers to order products, cut prices, or take other actions that once required human thought and intervention.

But data needs people just as much as people need data. Business leaders are still the ones who have to decide whether to launch a new product line, expand into a new continent, buy or sell a business, or change a venerable logo. So how can managers use the torrent of data available to them to make the best decisions? That is the timely subject of Decisions Over Decimals, a concise guide to decision-making in the age of analytics written by Christopher Frank, Paul Magnone, and Oded Netzer, a trio of business veterans associated with Columbia University (as well as American Express, Google, and Amazon, respectively).

“The challenge in today’s world is not the lack of information but the judgment to use it,” they write. Data-based decision-making is unavoidable, but it comes with risks: “Data and numbers tend to provide the comfortable feeling of accuracy and certainty, but they rarely tell us the full story. Numbers alone can never provide a perfect solution or answer, and they will never immunize decision-makers from faltering.”

What business leaders need, the authors argue, are techniques for combining good data with their own good judgment. Over the years, they’ve developed an approach that they call Quantitative Intuition (QI) to help executives make informed choices. And they are convinced that QI, oxymoronic as it may sound, can be learned. “Combining quantitative information with intuition—human judgment developed through experience and close observation—is indispensable.”

In its broadest outlines, QI doesn’t look terribly different from what any sensible person would do to make a decision. But as the authors walk us through their process chapter by chapter, a variety of useful insights, large and small, emerges.

For example, they cite the late organizational theorist Russell Ackoff’s observation that “we fail more often because we solve the wrong problem than because we get the wrong solution to the right problem.” Their point is that you don’t know what data you really need until you properly frame what it is you’re trying to decide. The data can only take you so far on this essential question because you are the one who has a view of the broader business landscape, including, of course, the complexity of your own business. “Don’t expect the data to provide both the questions and the answers,” the authors warn. “It is your responsibility to home in on the essential question and then combine data with intuition to attempt to identify the answers.”

Because the authors consider clarifying the problem to be so important, they believe it’s worth senior management investing time to do so up front, which will save time—and error—for everyone later. “Why?” questions, asked of subordinates in near childlike fashion, can help in problem-defining. Why do you want to conduct such a study? Oh, to know customers better. Great, but why do we want to know them better just now? And so forth, leading eventually to the real issue at stake.

Defining the problem first and then working backward toward the data can put you in some good company. The authors cite the example of Amazon. There, when people have an idea for a new product or service, they have to write up a press release and FAQs to help them and everyone else understand what it is, how it will work, and how various contingencies would be handled. That process helps all parties gain insight into what they really need to know to determine if the scheme is a good idea.

This sort of thing will help you focus on the right data. But you also need to make sure the data is right. Here, again, the authors have good advice: in both defining the problem and confronting the data, they emphasize the importance of asking powerful, probing questions. In particular, they recommend developing what they call “IWIK”—I Wish I Knew—questions designed to elicit data actually relevant to making a decision. All data, however obtained or elicited, must be rigorously interrogated. Is it accurate? Do means and medians mask explosive outliers? Is the period covered by the data the relevant one? Were the numbers you saw the result of posing the right questions? What biases might have influenced how data was harvested and presented?

In working with data, managers are exhorted to cultivate intuition by using back-of-the-envelope calculations, which will give a ready feel for magnitude and plausibility. Data must be accurate, but it needn’t be too precise. And all parties must learn to synthesize rather than merely parroting or summarizing. What matters is meaning.

In presentations, for example, make the bottom line the top line or, as we say in the news business, don’t bury the lede. The authors: “We should encourage our teams to not stop at the stage of what the analyses say and inspire them to also think about what this means for them or the organization.” In fact, people have to go further, from “so what?” to “now what?” In other words, go beyond “What does it mean?” to address “What are we going to do about it?”

Sooner or later—and better sooner—a decision has to be made. The authors are writing for corporate executives and advocate a fairly structured process, including, for instance, “Begin with agreement from stakeholders on the objective, timeline, and who must participate in a decision” and “Consider whether the decision is reversible or not, and if it is not reversible, can it become reversible?” But they also speak with the voice of experience to offer such good general advice as, “Focus on being vaguely right rather than precisely wrong. Get the best data you can, even if it is not perfect. With or without perfect data, a decision (and a good one) needs to be made.”

“Successful decision-makers,” they go on to tell us, “balance data, experience, and intuition. They quickly sort through information, apply judgment, and are fierce interrogators of data to cultivate sharp insights. They know there is more to decision-making than just the data. They resist being intoxicated by information.”

You should too.

"Big data”—the explosion of quantifiable information, much of it generated by people’s behavior on the Internet and social media—has captured the imagination of companies, academics, and the business press. Executives are rightfully intrigued by the idea of drawing conclusions about their customers’ buying propensities from details of their activity: who they’re connected to, what they like. This promise is born partly of the fact that the data for Internet-based analytics is already being gathered on computers, there to be sorted, filtered, and modeled.

Undoubtedly, big data will be the next game changer for marketers, but some will gain more advantage from it than others. Because of the automated tools that are available for mining data today, many executives assume that they will be able to easily uncover trends that weren’t previously visible. But analytics involves more than just knowing the facts. It requires the right analysts asking the right questions to make the right decisions. Any analysis of data that stops after asking “what,” which is already a big undertaking, isn’t analytics. You have to ask “why?” and “what next?”

To answer these questions, and to leverage big data’s full potential, companies need to go back to basics. Having been closely involved in the evolving marketing analytics landscape over the past 25 years, I’ve learned three pragmatic lessons that have always been at the core of a strong analytics program, and should guide your initiatives today: Rely on theory-based approaches, not blind data mining; develop a holistic view of your customers and markets; and learn by doing.

It Starts with a Theory

Without a theory about how consumers form preferences and act on them, an analyst will quickly be overwhelmed by the amount of data available, and all the processing power in the world won’t help. The starting point should be an explicit hypothesis about the needs of your customers and how you create value for them. It may be a new product in the lab that you think has the potential to be a runaway hit. Or it may be that there are customers in your market who aren’t really loyal to anyone—“undecided voters” whom you could capture with some slightly altered proposition. Once you’ve gathered the data required to test your hypothesis, the analysis will usually lead you to specific ideas for developing winning value propositions and taking them to market. Superior segmentation—clustering customers and prospects by similar behaviors or preferences—can lead to much more effective targeting strategies.

One pharmaceutical company trying to increase sales of a drug that was in decline, and for which the sales staff had been severely cut back, decided it might benefit by deploying its remaining salespeople more strategically. The company leaders had a hypothesis that their current sales plan was not effectively targeting physicians with the highest potential to prescribe their drug. To test this, the company gathered a large data set on all the physicians treating the medical condition for which this drug was prescribed—how many prescriptions the physicians wrote each year, whether the number of prescriptions they wrote was growing or declining, and to whose formulation (the company’s own or that of its main competitor) the physicians were loyal. This data allowed the company to identify a sweet spot in the market: physicians who wrote a lot of prescriptions to begin with; who were increasing the number of prescriptions they wrote each year; and who were not loyal to either manufacturer’s formulation, dispensing prescriptions for both about equally. The sales team made a beeline to pursue this opportunity, with results that exceeded their expectations.

Remember that this is not the first time a data revolution has changed the game for marketers, and previous transitions were not always smooth. In the mid-1980s, the introduction of barcode scanning enabled companies to gather information at the checkout register. Before that, data was limited; companies knew what they shipped, and they could ask people what they bought. But with the advent of scanners, they could actually see what was happening at the point of sale. In the early years of the technology, this new awareness led to a number of missteps. Executives became overly focused on the impact of price promotions on sales and lost sight of marketing fundamentals: brand equity and brand building. Over time, however, companies developed more sophisticated statistical models and refocused their energy, and scanners became the biggest boon to consumer marketing and retail in the last 30 years. (Today, this point-of-sale knowledge has been extended to include loyalty card data, which provides retailers with insight into what individual households put in their shopping baskets, and into online shopping behavior.)

Like the flawed marketing ROI models of the barcode’s early years, the newest big data analyses can be misleading. Many retailers say, “I know everything about what moves off my shelves. I know a lot about my customers who have loyalty cards. But when we put more things on the shelves that resemble the things they’re already buying, we don’t see the growth that we were expecting.”

What’s missing? It’s likely that by focusing on the newest source of data, the retailer has unintentionally developed a one-dimensional view of its customers. What it needs is the broadest possible view of a customer’s path to purchase. We sometimes call this perspective “a day in the life.” It means understanding more completely how your interaction with a customer fits in with all the other interactions he or she has with other retailers or (as the case may be) other businesses, shopping channels, or activities. Without that insight into what is prompting a customer to go somewhere other than to you, your growth initiatives can become a crapshoot.

Learning to Walk

The first steps you take to acquire, harmonize, and mine new data sources almost always lead to exciting new insights. As you gather these insights, it will be important to be open to new approaches and to challenge sacred cows. You may learn things about your customers that cause you to question certain products, services, or strategies. It can be a lot to take on. Rather than recommending that companies go whole hog with analytics all at once, I often advise them to undertake a few pilot projects. It will benefit them to learn to walk before trying to run: They can pick a product, a geography, and a problem that they want to focus on, and demonstrate to themselves that the return on effort and cost justifies the new undertaking.

For example, a global energy giant decided to tackle the question of quantifying and improving its return on marketing investment using more advanced analytics. Senior leaders selected two business units within three countries, spanning developed and developing markets, to conduct pilot projects. The conceptual framework and goals for each project were the same, but operating gas stations in Europe and selling motor oil in Asia required different data sets and analytics tools. This diversity enabled the corporation to gain experience with a wider set of possible approaches, and to determine which ones should be employed in which situations. In addition, they shared their success stories with other business units and countries to build enthusiasm for the initiative. The result was a sophisticated yet pragmatic program that was eventually rolled out, accepted, and used around the world.

Getting Back to Basics

Many executives are interested in using big data but have relatively little direct experience with the latest analytics tools and techniques. Right at the start, they typically ask me what it will cost. The response I am tempted to give is: “What’s the cost of making the wrong decision? What was the cost to Kodak of not reacting quickly enough to the advent of digital photography?” My less glib answer is that analytics can require a major investment, beginning with just assembling and harmonizing the data. On top of that, companies need specialists who are trained to do the more advanced work to find hidden patterns, interpret them, and turn them into insights the company can put to use.

But as these three fundamental lessons show, it can be a manageable process with the potential for significant rewards. In fact, it’s been my experience that once companies start investing in analytics, they almost never stop. The things they learn drive improvements in the business that more than pay for the effort. Analytics becomes a self-funding way for companies to improve their position in the market.

Reprint No. 00150