Wine4.Me co-founder Amy Gross wasn't always into wine. In fact, if it wasn't for an early insecurity, she might still be comfortably drinking beer and whatever mix drinks are easily ordered. But now she has an app that is not only personal, but possibly the most scientific and objective personal experience in an app out there.
"I started drinking wine when I first met my husband because he was six years older than me," the VineSleuth blogger confides. "I was trying to be mature, so on our first date I decided that I would drink Chardonnay ‘cause that would be much more sophisticated than a beer. After weeks and months of going through that, I started to actually figure out what I liked."
Did it work? Did she impress the boy? The short answer is yes: "We got married," she laughs.
A couple years later, Gross was still enjoying wine; it had even become an integral part of her and husband Gary's marriage. "There’s a shop that used to do 12 under 12," Gross says. "Every month, Gary and I worked through those, and try other wines as well."
Eventually, the casual wine drinking evolved into a greater curiousity, and soon they found themselves in wine country. "Our neighbors were going to Napa, and we joined them [in 2009]. We went to the places they told us to go — they made all the appointments and we just showed up. Being quite an inquisitive person (I was a former journalism major) I asked tons of questions, but got home and still felt like, 'gosh darn it, I still don’t know what wines I like.'"
"If I don’t know what’s in the bottle, I don’t know what’s in the bottle."
Wanting to take their wine experience to the next level, they began searching for ways to do that, sure that there had to be an app. But they couldn’t find one. They found apps that told them what the experts liked, what their friends and other people liked, but nothing that told them what they would like, as individuals. Mulling over the void, they decided that something needed to be done.
"I got to work on really thinking about how it should be created," Gross explains. "We reached out to his two brothers (one knew much more about wine than we did at the time, the other was a programmer) and they helped us with ways to characterize the wine, and then created the first rendition."
But simply characterizing the wines wasn't enough. They needed to make it undeniable — they needed to make it scientific and objective.
"We called our neighbor [Michael J. Tompkins], who really understood statistics," says Gross. "We asked him if he knew anyone who could help us. He responded that he had been thinking about it for awhile and knew how to create the algorithm; he simply didn’t know if anyone would use it."
By 2010, Tompkins, a research and consulting scientist who is now the company's Chief Science Officer, had jumped on board as a co-founder. Together, they got to work. Their first and foremost priority was to make sure all their data was substantial and robust. They talked about scraping tasting notes off the web; they talked about crowdsourcing reviews; they talked about all these ways other sites and apps were doing this. But none were doing it as scientifically as they had wanted to do it.
"We wanted our own team of scientists characterizing the wine — we wanted to come up with the system ourselves by working with sensory scientists," Gross explains. Through her blog VineSleuth Uncorked, she had interviewed and enjoyed some winemakers and winery owners in Finger Lakes Wine Region in New York, and come to trust them; so in an effort to fill in gaps that she and Mike had, she took a trip to the region. Gross thought that wine could be characterized subjectively, and their research showed it. And her excursion not only encouraged the team, but led to significant introductions that would heavily influence the app's development. One of the winemaker's wives worked in the enology department at Cornell University. Through them, Gross and Tompkins met Anna Katharine Mansfield, an assistant professor who focused in the sensory science.
Yet, there were obstacles to overcome in determining precisely what the algorithm would do. They wanted sommeliers to understand and appreciate the system, and they wanted scientists to understand and appreciate it, but they were not going to train every person on every individual characteristic as that would take them way too long. Instead, with Mansfield's expertise, they came up with a definitive list of traits that should be observed. Then they determined parameters for their evaluators, deciding that their team first needed agree, then, more importantly, have a repeat rate of 90% or higher.
"There are some really fantastic palates out there that can do great things in evaluating wine, but they may not hold up to what we need," Gross says. "Our evaluation is telling you what you’re tasting in the wine, what you’re enjoying in the wine. Obviously, whenever you have a relationship with somebody or you’ve been to the winery, there are going to be things that stand out to you, and bring that warm, fuzzy feeling about in wine. But the bottom line is: what you enjoy tasting, you enjoy tasting. We are scientific, but yet we’re also using the human element in tasting. You could analyze for characteristics in a lab setting but you can’t tell what you’re tasting. We can’t factor in the warm, fuzzy feeling, but we can tell you what you like and what other things you’re going to like."
Basically, by entering traits you like in your wine, or wines you've enjoyed in the past, Wine4.Me will put it through the algorithm and tell you others you'll similary enjoy. According to Gross, there are currently around 1250 wines in the database, wines acquired by themselves based on the 1,000 best selling wines in addition to personal/team picks. Their major focus was widely available wines, those that people can find in grocery stores and larger shops. As for the handful of rarer, boutique wines? Websites and online stores are also listed. And this database is only growing.
"There really isn’t anyone doing what we’re doing," Gross continues. "I know a lot of people say that, but we’re the only ones that are using sensory science. We don’t care what the critics like, and what your friends like — it’s what you like individually. I’ve seen systems that put people in my categories, but nothing that is really personal. Delectable does a great job with showing you what other people like, but that doesn’t help the casual wine drinker that doesn’t know what he or she likes."
"We want to empower casual wine drinkers to find wines that they like. The app is not for the person already out there trying all sorts of wine. It is for the person who knows one or two or three wines they like. The current marketplace leans on information from people buying $50, $60, $70, or $80 bottles of wine. Truth is, you don’t need to apologize for liking a $12 bottle of wine; different bottles of wine are doing differnt things for different people. Sometimes you want something rich and silky and lush, and sometimes you want something that goes well with pizza. To me, I’ve been surprised by how intimiated people get by something they’re meant to enjoy."
The ultimate goal is to instill a confidence to ask more questions, making more wine choices wihtout intimidations. And if they are able to predict what one user is going to like, they might someday be able to predict what areas will like, in turn using that critical mass to help stores and wineries. But, for now, Wine4.Me is simpler than that — it is about the individual.