Case study after case study has confirmed the value proposition for analytics across a wide range of business functions, including pricing, demand prediction, targeted marketing, supply chain optimization, CRM and HR. In my view, analytics is something much more than a technology with an ROI; it’s a transformational phenomenon that will fundamentally change how business discourse will be conducted and decisions made. An analogy may help in understanding why.
If you drop a feather and a rock at the same time from the same height, which will hit the ground first? At one point in history, this was a question for philosophers to resolve. Aristotle opined that the rock, because it was heavier, would fall faster and hit the ground first. Aristotle’s armchair wisdom was not questioned until the 16th century, when Galileo, through cleverly designed experiments, proved him wrong and established an empirical basis for answering such questions about the physical world.
Much the same way that an empirically based scientific method became the basis of our understanding of the world around us, analytics will eventually bring empiricism into business discourse and dethrone many of today’s business practices.
Recently, I received a memo saying that all employees at my location would be required to keep their offices clean, subject to inspection every other Friday. I wanted an explanation, so I asked if there was any data to show that clean offices lead to higher productivity.
My question, of course, was side-stepped, and I was told that clean offices would make a better impression on clients. Undeterred, I asked if there was any data to show that clients walking through our offices buy more of our services or express their “better impression” in any other way. Not unexpectedly, I was asked by the powers that be if this really was a battle that I wanted to fight.
I chose this example to illustrate how average, mundane decisions are made in organizations daily based on well-intentioned, plausible yet armchair theories—those that, like Aristotle’s, lack any empirical evidence. While highly specialized functions such as pricing or customer segmentation may be based on sophisticated models and empirical data, my contention is that the long-term impact of analytics will be in instilling a culture of data-driven decision making at all levels of an enterprise.
Or, put more bluntly, business proposals and decisions—big or small—will have to provide satisfactory answers to this question: “Do we think this is true or do we know?” (This particular formulation is attributed to Gary Loveman, CEO of Harrah’s Entertainment.)
A sophisticated and analytically oriented enterprise of the future will behave and operate differently from today’s enterprise along five major dimensions.
High analytical literacy
Data is a double-edged sword. When properly used, it can lead to sound and well-informed decisions. When improperly used, the same data can lead not only to poor decisions but to poor decisions made with high confidence that, in turn, could lead to actions that could be erroneous and expensive. Let’s consider some specific examples.
When one has access to real-time data, it’s tempting to make real-time decisions. For instance, if you are a retailer and you have real-time access to sales data from cash registers from all your stores and real-time access to your inventory in your warehouse, you could be tempted to run sales promotions on the fly and manage your supply chain in tandem to support your real-time promotions.
However, this is unlikely to work because three types of events—your decisions, the ensuing customer behavior and supply chain events—operate in different timeframes, so making decisions any faster than the slowest-moving event could be useless at best and dangerous at worst.
Another problem with data and analytics is that they give you very fine-grained visibility into your business processes, and you could be tempted to overoptimize the processes. Highly optimized processes—just-in-time inventory being an example—are very fragile because circumstances beyond your control could arise, and there is little room for error.
A third problem is what’s known as “oversteering,” or making decisions when none is needed. So, for example, your data could tell you that a project is behind schedule, which, in turn, may lead you to berate the project manager or tell your stakeholders that the project will be delayed. Yet neither of these actions may be necessary if the project has contingency built in, if the status update has a different frequency than your sampling frequency or if perhaps the employees who are aware of the project delay will put in more work time to get the project back on schedule.
Businesses thrive on stability and repeatability. Stable and repeatable processes justify large-scale capital expenses; they justify large-scale employee training; and they reduce cognitive overhead because processes and decisions do not change and hence their rationale does not have to be explained repeatedly.
By contrast, an analytically based enterprise of the future will have to be designed around volatility rather than repeatability.
When you have fine-grained visibility into your processes, customers, suppliers and competitors, you have the ability to make very fine-grained decisions. In fact, your decision rules can capture subtleties such as “stock more beer on Sunday nights in locations where the home football team is on a winning streak.” Such decisions are highly context-sensitive and can change as rapidly as the fortunes of the football team.
Volatility—or rapidly changing decisions that are context- and time-sensitive—will be a big challenge for enterprises. Decisions are no longer easily explainable; capital investments cannot be based on mass repeatability but must cater to endemic volatility.
Today’s enterprises have more information than they can act upon because the information is siloed in so many ways: technologically (data in different systems that cannot be brought together), organizationally (data in different governance units that cannot be brought together) or by ownership (inside versus outside the enterprise). The enterprise of the future will be (or will be forced to be) “conscious” in the sense that it will know that it must integrate everything it has access to.
As an extreme example of “integrated awareness,” let’s consider pharmaceuticals, an industry that has traditionally relied on clinical trials data as a means of establishing the efficacy and the side effects of a drug.
A pharmaceuticals company today can legally and morally claim immunity from any adverse effect of a drug that was not revealed during clinical trials—in other words, any information that it did not explicitly collect as part of a clinical trial protocol. But in a world of blogs and social networks, where people share this information unprompted and in public, it will become both a responsibility and an obligation of pharmaceuticals companies to monitor public sources and integrate the public information with their own clinical data. (For more on the business impact of social media, see this article.) We lost our reference.
“I should have known” (either for regulatory or competitive reasons) will be the new normal, replacing the “I did not know” or “I could not have known” approach to awareness and information integration.
The end of analysis-paralysis
In the future, businesses will likely be run by managers and leaders who are no-nonsense empiricists; they won’t move a finger until after all the relevant data has been gathered and analyzed. A recipe for organizational “analysis-paralysis”? This is not an unreasonable fear. But though it may seem counterintuitive, an empirical enterprise with high analytical literacy is less likely to fall prey to this malady than today’s enterprises.
There are three very distinct ways that organizations can fall into the analysis-paralysis trap. One is a managerial tendency to “over-fit the curve”—a statistical term that refers to the diminishing value of additional data once a pattern (or curve, in the graphic sense) has been found. Data collection has a price, inaction has a price and an analytically literate organization will clearly understand the cost of over-fitting.
The second cause of analysis-paralysis is waiting for data that simply does not exist, which reflects an inability to design experiments to generate the needed data. As mentioned above, experimentation has a price and inaction has a price, so an analytically literate organization will be characterized by a clear understanding of data gaps and the value of experimentation to break the logjam.
The third cause of analysis-paralysis is the fact that most companies do not know or articulate their risk tolerance clearly and are much more likely to penalize failed action than inaction. As a result, many managers do not act unless there is enough data to assure them of successful outcomes. An analytically literate organization will have a firm grasp of its risk tolerance. With guidelines and models for action under uncertainty, it will restore the symmetry between how it treats failed action and inaction.
Intuition’s new pulpit
Empiricism and analytics sound a death knell for such vaunted business traits as intuition, gut feel, killer instinct and so forth, right?
Science is purely empirical and dispassionate, but scientists are not. Science is objective and mechanical, but it also values scientists who are creative, intuitive and can take a leap of faith.
Data, by itself, can be interpreted in many ways. Imagine a physical or business phenomenon that produces the following sequence of data: 1, 2, 6, 24, 33. Perhaps it’s a factorial sequence with 33 as noise, or a sequence where every fourth term is twice the multiple of the previous three. Or perhaps every fifth terms if the sum of the previous four.
All are indeed correct. To prove or disprove any theory, you need the next several terms of the sequence. A good scientist knows when there is enough data to warrant a theory, when there isn’t, what new data to gather and how to design an experiment to gather the right data.
Apple’s Steve Jobs is known to explicitly discount the value of surveys and focus groups for designing new products. How do you explain this apparent anti-empiricism?
One explanation is that, much like a creative scientist, people like Jobs recognize when there is not enough data or the right kind of data to form a theory. They recognize that, for completely new lines of products that will change a user’s experience or behavior, the only useful data is experiential data, not commentary and reactions from those who have never used the product.
Jobs and people like him are akin to scientists who recognize what type of data is needed to support a theory (in this case, whether a product will succeed), recognize that such data cannot be gathered through focus groups (one type of experiment) and boldly design new types of experiments (release the product and gather experiential data).
It should be noted that some products—in Apple’s case, it was the Newton—do not succeed and are terminated. Intuition, creative leaps and clever experimentation are not incompatible with empiricism; in fact, the value of these traits will be even better understood in the future enterprise by analogy to theoretical and experimental scientists.
The enterprise of the future, based on empiricism and analytical decision making, will indeed be considerably different from today’s enterprise. You may well ask: “Do you think this is true or do you know?”
About the author
Kishore S. Swaminathan is Accenture's chief scientist and the global director of Accenture Technology Labs' systems integration research. He is responsible for defining the company's vision for the future of technology and setting its research and development agenda. Based in Beijing, Dr. Swaminathan has spent his Accenture career researching cutting-edge technologies. Winner of the 2000 Computerworld Smithsonian award for the best application of IT, Dr. Swaminathan has worked on more than a dozen research projects and has as many patents to his credit.