Data vs. Executive Bias...Round 1 (Fight!)
Big data is long past the hype cycle. Today, it is firmly entrenched in the mindset of everyone from the CEO on down. In fact, a business decision-maker who says, "We don't ever use data in our decisions" would be laughed out of the boardroom.
But there's an enormous amount of data—and insights—still waiting to be analyzed and used. That's because just the tip of iceberg, about 0.5% of data, is currently being analyzed, according to IDC.
But business use of data is changing—and one of the drivers is advances in artificial intelligence (AI) and machine learning applied to data. According to a Harvard Business Review article, big data and AI projects have become almost synonymous. And machine learning has become one of the most-used technologies for analyzing large volumes of fast-moving data—the kind of information that had been inaccessible to most companies.
Machine learning is a subset of AI in which the software teaches itself to perform better analyses. The software continually improves without the need for upgrades. But the availability of such tools doesn't mean you can set 'em and forget 'em. Becoming a data-driven business requires more than technology; the entire company must shift its outlook and practices so that it makes decisions based on real-world evidence.
Guided by Data
Duke University's Fuqua School of Business does a biannual survey of chief marketing officers. The Duke researchers found that, while spending on marketing analytics increases each year, the effect of analytics on company-wide performance has been modest at best. They attribute this poor performance to a shortage of trained data science professionals, as well as to a lack of access to measurement tools and processes.
While the Duke stats are specific to marketing, they have more universal implications. For example, only 36 percent of C-level executives told PWC they were highly data-driven. What's more, data analytics teams are still the smallest of any department in companies.
These are best practices, but all too often, a company's culture simply does not support implementing them. For example, KPMG's 2017 Global CEO Outlook found that the 1,300 CEOs surveyed didn't completely trust the data and analytics on which they based their decisions. When decision-makers aren't convinced that the findings of their analytics tools are reliable, they may let their own opinions and biases outweigh the results.
Business experts at the University of Toronto identified three common traps executives can fall into:
Confirmation bias: Paying more attention to findings that match your beliefs or hopes is natural, but it goes against the very point of analytics. For example, if your CEO has proposed launching a new service, the team tasked with analyzing its prospects may focus on positive findings and ignore negative ones.
Overconfidence: Successful business leaders are used to making decisions based on their experience or even gut reactions. It can be hard to let go of your ego and actually listen to what the analytics say—but it's essential.
Over-fitting: This is a data-science error that happens when analytics models conform closely to an existing data set. The model may be good at drawing the correct conclusions from this data set but fail at providing insights into the future.
It's notable that two of these pitfalls aren't inherent in the data or analysis; instead, they're the result of people interpreting—or ignoring—what the analysis says.
Characteristics of Data-Driven Companies
On the other hand, companies with cultures that support the use of information to drive decisions have several characteristics, says Irina Peregud, marketing director of FriendlyData.
First, they tend to have a CEO who's open-minded and curious. Top executives need to have a researcher's mindset: They have a strong drive to dig into data and find new insights.
Another characteristic of data-driven companies is that their employees understand data and how to use it. To facilitate this, organizations need to provide employees with broad access to data, so that any individual can immediately access the information needed to make a decision, without having to request a report from another department.
Finally, companies with a data-driven mindset are willing to learn from failure. Data analytics can help a company determine if a project produces enough value. What's more, data can help decision-makers understand why a project missed the mark. That knowledge can then be useful for making adjustments that help ensure a better return on investment in the future.
Becoming a data-driven business is a process, not an end. Analytics tools and services powered by AI give companies better access to actionable insights from their data. Once business processes and company culture integrate data analysis, business leaders won't just think they are part of a data-driven business, they'll know it.
LATEST POST : The Simplicity (and Complexity) of Auto-Feature Engineering