Big data killed the political star. As the dust settles following one of the biggest political upsets in history, the masses are wondering where the skewed information originated.
As experts reflect on the path that resulted in Donald Trump securing the position as presidential-elect, questions are surfacing about how the consumption of big data failed to portray an accurate picture of the political landscape. Retailers should take note on how to avoid similar pitfalls when applying big data analysis to business practices.
“What we are seeing coming out of this week’s election is a failure to predict the outcome of the election via ‘self-reported’ behavior. Surveys and polling are subjective. By collecting very large data of observed activity through the Adobe Marketing Cloud for instance, retailers are able to view actual behavior,” said Tamara Gaffney, principal analyst at Adobe Digital Insights. With the rise of artificial intelligence and the corresponding reliance by humans, the election results are a gut-check for retail executives.
Data analysis is an integral part of business survival — as long as the survey isn’t simply a reflection of executives’ pipe dreams. As mainstream media comes under scrutiny for inaccurate projections due to insufficient polling practices, business leaders are best reminded that this also applies to revenue channels. Lori Mitchell-Keller, general manager of global consumer industries at SAP said, “Political polls are like consumer focus groups, you hope you’ve selected a random sample of your target segment and that the pursuant exchange provides insights on future product or service.”
Consider Survey Monkey’s interactive electoral map, which allows users to view the U.S. according to various voting demographics. For example, by selecting the Millennial (ages 18-35) segment; the map turns nearly entirely Democratic blue representing approximately 473 electoral votes.
Meanwhile the demographic of voters falling within the Education: No College section, the map turns almost fully Republican red showing President-elect Donald Trump earning 285 electoral votes, in step with the actual election results.
This is perhaps due to the digital accessibility of certain demographics. “Like the polling business, retailers must navigate an increasingly digital world. To thrive in an era of unprecedented innovation, businesses need to digitally transform their operations. This requires a digital core that gives the retailer a 360-degree view of their consumer,” said Mitchell-Keller.
With business — similar to electoral polls — the larger the data set to analyze the better. Gaffney said, “The key is to bring together as much observed data across channels like advertising, social media, mobile and from their own online sites and adjust ‘self-reported’ behavior findings when they are not aligned with observed behaviors.” This is a top signal that the analysis might be skewed. “Big data doesn’t fail or succeed on its own — that hinges on understanding the sources of that data, its currency and context,” Mitchell-Keller said.
Comprehensive understanding and sharing of big data is paramount for not only maximizing the potential of its information, but for future projections. Mitchell-Keller said, “To harness the potential power of big data, there are three critical steps retailers must take: Obtain a single version of the truth; share the wealth of your data anywhere, anytime, and use your data to meet ever-evolving customer needs.” By referencing this data and sharing its findings, executives and employees are best informed to strategize for upcoming initiatives and predict any speed bumps.
As polling techniques continue to be examined, retailers should be forewarned of incomplete data analysis that can lead to catastrophic errors in judgment calls.