Taking a Shot at AI: What Companies Can Learn From Jack Daniel’s and Data

Print Friendly, PDF & Email

In this special guest feature, Martin Kurpiel, Vice President at Valid, discusses how the maker of Jack Daniels recently launched a new brand of rye by leveraging its data analytics team. Martin Kurpiel is the Senior Vice President of Technology Solutions and IT at Valid. He has more than 40 years of experience in solution architecture, product development, and application and system integration, with a strong emphasis on project management. In previous roles, he supported project managers, managed internal and external relationships, provided resources, engaged stakeholders and directed client relationships. At Valid, he actively leads IT, logical security, data solutions, data analytics, project management and product engineering.

Artificial Intelligence (AI) is exploding across industries. Manufacturing, retail, healthcare and more are using AI to improve customer experiences, automate processes and perhaps most importantly, unearth lucrative insights hidden within their data.

Earlier this year, Brown-Forman Corporation (the $3 billion company behind Jack Daniel’s bourbon) used their data to support the launch of an entirely new product. For three years, the corporation took inventory of and integrated vast stores of consumer, production and sales data siloed across the global brand. Then, by combining the data with internal financial information they uncovered valuable insights into pricing, barrel yield, competition — and market share. Despite the brand already producing two other rye products, the data showed that there was room for another. And in February, Brown-Forman launched Old Forester rye.

Having been convinced of the value this newly integrated data can provide, the brand recently stated that it’s ready to “dabble in AI” to unearth future insights. Even experimenting with AI requires a certain degree of preparation and meeting certain prerequisites, especially with large supplies of data.

Leveraging Data as a Competitive Advantage

In a recent report from PwC, the majority of executives agreed that consumer data is “critical or important” to gaining a competitive edge — but only 15% have the data necessary to do so.

The correct kind of data is important to the success of AI solutions. Data is the raw material that AI depends on and the technology cannot operate effectively if data is incomplete or inaccurate. Because AI uses data to look for patterns and make recommendations, any flaws in the data risk critically skewing the decisions and recommendations made. For example, if bad or incomplete data erroneously identifies a medical patient as high risk, they could be denied insurance coverage.

How to Prep Data for Success

Not preparing the data before implementing AI solutions is one of the most common reasons that AI fails. Here are some steps to ensure that data is ready to use:

  • Clean the data. Organizations must ensure that data is scrubbed clean before attempting to glean any insights from it. That means, making sure that formatting is correct and streamlined as well as identifying and removing any anomalies. It is also important to evaluate the data as a whole to determine if anything is missing from the data sets that would render them incomplete.
  • Evaluate the outliers. If analysts identify any outliers, then they must take a closer look to determine if the outliers are accurate and if they are likely to recur. Outliers may need to be transformed or converted to a categorical value to reduce the potential impact on averages, which AI uses to create its algorithms.
  • Use balanced samples. When it comes time for training and evaluation, it’s important that organizations collect the data proportionately. That means collecting across different groups and timeframes in order to reduce any accidental biases.

What Does the Data Say?

AI predicts outcomes. Therefore, it’s important that organizations understand how algorithms reach conclusions to safeguard against false positives or potential noise. To verify the effectiveness of algorithms and protect against biased results, enterprises must continuously review the output. While AI technology is adept at identifying correlations between data attributes, it has no way of deciphering if those correlations make sense — human reasoning is necessary to interpret results and draw conclusions.

It’s estimated that by increasing data usability by 10%, Fortune 1000 companies can increase their annual revenue by more than $2 billion. By preparing data and joining forces with AI, organizations can expect years of return and a competitive edge.

Sign up for the free insideBIGDATA newsletter.

Speak Your Mind