Business challenge

Most people buy wines based on the grape variety, region, or just the label they’re familiar with—but what if there was an objective way to pick a wine that perfectly matches your tastes?

Transformation

VineSleuth’s data model assesses wines objectively, by flavor alone. It asked IBM and the University of Connecticut to analyze how its results compare to data used by other wine industry leaders.

Results

Confirms

that VineSleuth’s methods yield results that are unique in the wine industry

Deepens

consumer insight to help drinkers enjoy a wider range of wines

Highlights

the value of partnering with IBM and academia to solve business problems

Business challenge story

The world’s first sommelier-as-a-service

VineSleuth, one of the most innovative businesses in the wine industry, was founded on a realization that the way the wine market currently provides information to consumers is deeply flawed. Wines are categorized and marketed by grape variety, by region, by vineyard, by alcohol content—but what the consumer really wants to know is how they actually taste.

“The University of Connecticut project has given us robust independent validation that our model offers something new in the world of wine.”

—Amy Gross, CEO, VineSleuth

Amy Gross, President and Founder of VineSleuth, explains: “The vocabulary that the wine trade has built up over the years is imprecise, and often impenetrable to the casual buyer. More than that, talk of regions and grape varieties can be intimidating to ordinary consumers, and put them off exploring new wines. The label on the bottle might use terms like ‘full-bodied’ or ‘fruity’, but it’s usually impossible for an average consumer to tell what a wine will really taste like until they actually pull the cork.

“I started VineSleuth to push back this complexity and help more people enjoy great wines. When buying wine at a store, I wanted to feel like I was being advised by a world-class sommelier, rather than relying on my own imperfect knowledge and intuition.”

VineSleuth realized that it could bring greater objectivity to the process of choosing wines by building a sophisticated data model that ranks each wine along dozens of different taste dimensions, developed by sensory scientists. A panel of expert wine tasters assess every major aspect of each wine’s taste profile, and the results are fed into VineSleuth’s proprietary algorithms to map out the similarities and differences between wines. IBM Watson cognitive technologies are then used to help communicate the results to end-users—including winemakers, retailers and consumers—using natural language.

“The idea is to help retailers and wine companies build up a data model that creates a more tangible link between their wines’ taste profiles and their customers’ preferences,” says Amy Gross.

However, while VineSleuth was confident in the robustness of its methodology, it wanted to gain a better understanding of how other wine companies assessed taste profiles, and validate its own findings.

“Most wine companies collect ‘tasting notes’, which summarize their own experts’ knowledge and opinions on each of the wines they produce,” says Amy Gross. “This data is typically stored as unstructured text, and not all of it is truly relevant to assessing the taste of the wine, but we wondered if it contained any dimensions that could be mapped consistently against our own results.

“Essentially, we wanted to know if what we had was just a more accurate, higher-quality version of the data that everyone else in the industry has, or if our proprietary analytics methodology gives us something truly unique. To find out, we decided to partner with IBM and the University of Connecticut.”

“From the data scientists who could already code in Python or Java to the MBA scholars who only knew Excel, they all learned how to use Watson Analytics within a week.”

—Girish Punj, Professor of Marketing, University of Connecticut

Transformation story

Finding the perfect pairing

Girish Punj, Professor of Marketing at the University of Connecticut, takes up the story: “As part of our MBA program, we offer an innovative course that brings our business and data science students together and gives them a real-world challenge to solve, using both their business acumen and their analytical skills.

“I was looking for a commercial organization that would be willing to let us partner with them, and when I saw Amy present at an IBM conference, I immediately put VineSleuth at the top of my wish-list. I spoke to my contacts in the IBM Academic Initiative, they provided an introduction, and Amy was enthusiastic about the partnership and quickly identified the perfect problem for our students to work on.”

John A. Elliott, Dean of the School of Business at the University of Connecticut, concurs.

“Our students were able to use their marketing and business analytics skills to help devise a sophisticated tool that will enable consumers to select a wine that satisfies their unique tastes,” Elliott says. “This is the type of hands-on learning that our students most enjoy—challenging, purposeful and directly relevant to the business world.”

The project itself involved the analysis of two highly confidential datasets—one containing data generated by VineSleuth’s own proprietary taste profiling methodology, and the other containing unstructured tasting notes provided by one of VineSleuth’s corporate partners. The students’ task was to analyze and attempt to correlate the two datasets, to see if it was possible to find consistent similarities between the results of VineSleuth’s assessments and those of the other organization’s wine experts.

Professor Punj comments: “We gave the students a free choice of using whatever analytics tools they liked—including SAS and R as well as IBM products such as SPSS®. It was notable that although we had a wide mix of skill levels in each student team, almost everyone picked up IBM Watson Analytics and used it to explore and get to grips with the datasets quickly.

“From the data scientists who could already code in Python or Java to the MBA scholars who only knew Excel, they were all able to learn how to use it within a week.”

Following this initial exploration phase, several of the student teams decided to use IBM SPSS Modeler Premium as a powerful statistical modeling engine for deeper analysis of the data.

“Several of our students also took the opportunity to close the loop by importing the results from SPSS back into Watson Analytics and taking advantage of its visualization capabilities,” says Professor Punj.

Results story

Vintage results for retailers and consumers

The findings of the seven student projects differed in their details, but one outcome was clear: there was no consistent way to correlate the results of VineSleuth’s objective taste profiling methodology with the more subjective data found in the tasting notes.

Amy Gross comments: “I think some of the students saw that as a failure on their part, and were disappointed that they couldn’t connect the two datasets—but that absolutely wasn’t a disappointment from VineSleuth’s perspective. It provided robust independent validation that our model is unique, and that our data provides something that the rest of the industry currently cannot match. That’s a very healthy indicator for our business.

“It means, too, that we have a genuinely new tool to help retailers market their wines—and consumers expand their palette. For example, a retailer could embed our service in interactive kiosks in the wine sections of its stores. When a customer scans their loyalty card, the kiosk could ask them whether they had enjoyed the bottles of wine they bought last week. It could then use the customer’s preferences to recommend a wine they have never tried—or thought of trying—before.

“Or, to take it one stage further: they could integrate our service into the recommendation engine of their online store. When a customer has, say, salmon and chili peppers in their basket, it could recommend a white wine that pairs well with that flavor combination. Or vice versa—if they have a full-bodied red wine in their basket, it could recommend picking up steak and other ingredients that would complement it. Ultimately, that’s going to translate to bigger basket sizes and a higher average spend per customer.”

She adds: “It has been a great experience working with Girish and the University of Connecticut team, and it’s something we would definitely recommend. If you’re a small company and you don’t have an army of data scientists that you can throw at every problem, it’s great to have access to a team of intelligent, motivated students who can augment your in-house resources.”

Professor Punj concludes: “We would like to thank VineSleuth for giving us the opportunity to work on this fascinating project. The students really enjoyed the challenge of taking a deep dive into a subject area that is so innovative and unusual from both a business and a big data perspective.

“By bringing data science and MBA students together, we created a peer-learning environment that helped each student gain new skills—and using IBM tools like Watson Analytics really helped to bridge the gap between the technical and business scholars.

“The lessons we’ve learned from this initial partnership are now being put into practice: we’re already working with IBM to line up another corporate partner to work with over the next few semesters.”

10981561

About University of Connecticut and VineSleuth

Founded in 1881, the University of Connecticut is one of the top public research universities in the United States, with more than 30,000 students across its four campuses. Its MBA program brings students into contact with the world of Big Data, teaching a new generation of business leaders how analytics can help to drive organizational success.

VineSleuth uses cognitive computing and sensory science to help the wine industry reach out to new audiences by providing truly objective insight into how wines taste, giving consumers a better basis for their purchasing decisions.

Take the next step

The IBM Watson Analytics Academic Program gives educational institutions access to leading-edge Watson Analytics technologies for classroom teaching and research. The software will not only allow students to explore datasets, get automated predictive insights, and create dashboards; it will also allow them to analyze social data from Twitter to achieve a more complete view of their data. For more information, visit watson.analytics.ibmcloud.com/solutions/industry/education/wap

IBM Analytics offers one of the world's deepest and broadest analytics platform, domain and industry solutions that deliver new value to businesses, governments and individuals. For more information about how IBM Analytics helps to transform industries and professions with data, visit ibm.com/analytics. Follow us on Twitter at @IBMAnalytics, on our blog at ibmbigdatahub.com and join the conversation #IBMAnalytics.

View more client stories or learn more about IBM Analytics