Business Forum: What your company needs to do to get analytics right

"Yeah, well, you know, that's just, like, your opinion, man."

– "The Big Lebowski"

Analytics has never been sexier in the world of business. Big data, artificial intelligence, and machine learning are all terms that fill executives with excitement at their potential — or dread at falling behind. For companies investing in these technologies, how can they raise the odds of success?

Broadly, there are four critical factors for success in analytics. Obviously, these may be applied differently depending on the sort of analytics you are implementing, but the core principles are the same. Of the four, three are operational, and one is deeply cultural, even emotional:

1. Technology. A well-designed, flexible IT platform is critical for successful analytics. Most important is that the data be available to users in a consistent and reliable fashion. In the real world, a majority of resources (technology and people) are spent assembling the data for analysis.

The field of information management for analytics is about 25 years old — a youthful, immature engineering discipline compared with many others: civil engineering (2,000 years old; the Romans could grade roads); electrical engineering (150 years old); and computer engineering (50+ years).

2. Lean processes. Lean, Total Quality, Six Sigma — whatever methodology you use, a company's information is only as usable as the consistency and standardization of its processes. How can different departments agree on what numbers mean if they don't agree on what they are measuring? This pursuit of "a single version of the truth" has always been critical to business intelligence, and it continues to be critical in more-advanced forms of analytics such as data science and machine learning.

A common misperception is that artificial intelligence will discover new insights on its own without direction from the user. In reality, the majority of the time data scientists propose a hypothesis based on observations or gut instinct that the system then investigates, often taking the user down new paths. But it is rare for genuinely autonomous analysis to be valuable.

4. Fact-based leadership. The most important and difficult success factor for becoming a fact-based organization is fact-based leadership. Let's assume an organization has built the ability to gain insights by analyzing data. Will executives make decisions based on this data? The simple answer would be "Sure, why not?"

The more complex reality is what we have all seen and experienced in every organizational setting we have ever experienced — business, government, not-for profit and others. Leaders want the facts, and will use them — unless the facts contradict their past experience, or their gut instincts, or their political aspirations, or have the ability to derail a pet project they have invested their reputation in.

When faced with these temptations, the deciding factor will usually be the organizational culture — have executives been trained to follow the facts as a matter of habit? Are workers who unearth difficult facts celebrated or diminished?

This is another commonality with a successful lean organization: Everyone has to answer to the facts and satisfying customers. As an interesting historical note, after World War II all the major Japanese car manufacturers adapted Total Quality Control. But the best-quality cars came from the newer companies such as Toyota and Honda, which were more open to change. In contrast, Mitsubishi, an older Japanese manufacturing conglomerate (Zaibatsu), with roots in WWII weapons manufacturing, lagged behind the newer firms for several decades in adoption of lean culture ­— and quality of cars.

The organization itself must encourage this fact-based truth-telling, or its senior ranks will be filled with survivors who rose as yes-men. After all, it was a fact obvious to all, without advanced analytics, that the emperor had no clothes. Yet who would say it aloud? Only a powerless child with nothing to lose.