By “you” I don’t mean “one.” I mean you. There are people out there running their companies with big data. But chances are you are not one of them.
In recent years data and analytics have become a business ideology. IBM runs ads during football games to tout how it is creating “a smarter planet.” A few choice anecdotes circulate, such as the one about how Target knew a teenaged girl was pregnant before her father did, simply by studying her purchase history. Executives want their companies to become “data-driven” in the certainty that this orientation will lead to more customers and greater profitability. Vendors promise magic results, and the promise of magic results is used to justify big implementations, featuring the usual cast of large vendors, global integrators, and confused clients, as well as the usual phenomenon of budgets gone wild.
Do not assume you are reading yet another piece attacking big data hype. I happen to believe that much of the hype around data is understated, and that those who dismiss big data as yesterday’s news are sleeping through the most essential phase of the hype curve. We are truly in a new era of unprecedented data availability and cheap processing power. This era will transform business. I am skeptical about neither proposition. What I am skeptical about is the ability of most traditional companies to participate in that data revolution, let alone lead it. There are sharp, data-centric players already flourishing out there. But who are they? The case studies seem to always be about the usual suspects like Google, Amazon, and some edgy startups, with some scattered examples from traditional industries thrown in to enrich the story.
Look around your own company and be honest with yourself. If you work in any sort of traditional corporation, you are probably not “leveraging big data,” and the odds are you never quite got a handle on your own small data either. Your company likely lacks three essential characteristics of a data-driven organization:
- Data literacy, meaning a shared competency in mathematics and statistics that lets teams study, learn from, and apply data to real-world problems.
- Flexible processes and systems which can incorporate changes rapidly in response to new information, in critical areas such as pricing, product development and the customer relationship.
- A data-focused culture in which facts, information, and assumptions can be analyzed, debated openly, and ultimately used as the basis for action.
Rather than these qualities, the typical corporation more likely has what I call the three H’s: hierarchy, hunches, and hubris.
The archetypal data-centric company, Google, is a very different place than where you work. Not only does it hire engineers, mathematicians and computer scientists; it is run by them. Data literacy and an innovative mindset are assumed qualities of Google’s employees, which it calls “smart creatives.” The company is built for responsiveness: it is relatively flat and is an interlocking set of literal and metaphorical “services” designed to share and apply information and insight across product lines. It has a meritocratic culture in which the best-formulated and best-supported argument is supposed to win, no matter whom it came from.
This is not to say that Google always lives up to its ideals, that its methods would work in every industry, or that it is a perfect place to work. The point is rather that as a true child of the information age, the company is built to profit from data from top to bottom. Its competency with data is a first principle, not a feature grafted on after the fact. Conversely injecting plentiful data and analysis into a traditional company will not turn it into Google. In some cases applying these tools in an old-line organization will only cause frustration. Employees will begin seeing vast gaps between what they could accomplish and what they are actually permitted to do. They will gain insights that no one is empowered to act upon. And a fledgling data function may become nothing more than an elaborate internal science project, a curiosity that generates interest but does not drive real results.
Given that you are not a Google and not likely to become one, what should you do? Given that I do not believe this is a solved problem, I do not have any glib advice. But I will say that most of what I see in the analyst literature about creating a data-centric organization is not very enlightening. Recommendations to involve stronger business sponsors or institute more governance or roll out incrementally seem both elementary and inadequate. We need to start thinking more seriously about what a data-centric organization should actually look like. In that spirit, I have three suggested avenues for exploration.
First, combine focused data projects with process projects, with a goal to solving very specific business problems. Find improvement opportunities that could be addressed with better information. Identify, select, and begin governing the information needed to support that process. Simultaneously, engineer flexibility into the process itself so that the people who execute it will actually be able to use and apply insight to improve results. Recognize that both halves — the right data and the process flexibility — are equally important.
Second, build a data-literacy program within your organization, spanning levels and focusing first (of course) on the process areas in which you want to make initial progress. Don’t assume that all your staff are mathematically or statistically literate; but likewise, do not assume that they cannot achieve improved levels of competency. Educate them through a tiered program, much as companies have done for decades with Six Sigma and Lean. Make sure that a culture of data literacy takes hold and that data and information start being used to drive decisions. Highlight, reward, and publicly applaud cases in which this actually happens.
Third, bring data analysts and data scientists into the mainstream of your organization. Embed them into teams who can understand and act on the insights they generate. Incentivize them not to preach (which they can sometimes end up doing), but rather to get directly involved in discrete issues, study and improve critical processes, and share in the gains achieved. This type of approach needs to be governed (you do not want analysts working to improve one metric while harming another), but as a means of engaging and motivating analysts, it is sound.
As the sociologist Anthony Giddens argued in Modernity and Self-Identity, the strange thing about scientific method is that it tends to undermine certainty, not strengthen it. The old BI vendor promise of a “single version of the truth” is not only impossible, it’s irrelevant. A data-centric company does not need a single version of the truth; it needs tools, people, and culture that support a culture of learning and experimentation. Those qualities, not simply piles of data, are what might make you the Google of your industry.
Rob Lawrence 