The next great wine critic: Human, or machine?
No one, not even an omnivorous reader like myself, can possibly see everything that’s published in the world of wine, so it was that I missed “the news [that] traveled around the Internet so quickly it was seemingly everywhere,” in the words of Cyril Penn’s Wine Business Monthly. (Oh, well, better late than never.)
That “news” was the launch of a new app, from a company called Next Glass, “touted,” in the article’s words, “as the app that scientifically chooses your next wine or beer” by “understand[ing] the flavor profile of a drinker” and then “tell[ing] the consumer (on a 100 percent compatibility or ‘enjoyability’ scale) how likely they are to love it…”.
The idea originated with the company’s CEO, who had dinner with friends a few years back. “The diners were having trouble selecting a bottle [and] a suggestion from the waiter didn’t go over well,” so the CEO “realized that science could be used to make accurate and reliable recommendations.” The “science,” in Next Glass’s case, includes mass spectrometers, chemical analyzers and algorithms, all of overseen by a Ph.D. The company’s COO says Next Glass goes beyond crowd-sourced platforms like Yelp and Trip Advisor because it can tell the consumer whether or not he or she will enjoy the specific wine chosen, and not just what other people thought about it.
Well, this is obviously controversial stuff. Penn wrote, in an introduction, that the announcement from Next Glass “prompted one of the world’s greatest wine essayists to pen a piece…saying, ‘You think an algorithm can replace a wine critic? Think again.’” Penn didn’t identify the wine essayist, but Google did: None other than Matt Kramer. I was not surprised, because Matt really is one of the world’s greatest wine essayists and is always worth reading.
You can guess as to the substance of Matt’s argument: he summarized his opinion with, “It’s individual critics who are the real app…”. Now, you can say that Matt is hardly objective; he writes for a magazine whose strength is its bench of famous wine critics. But I think to dismiss Matt’s position based on that is not valid. Too often, we look for conspiracies and ulterior motives in critics, when really, there are none; and Matt’s credibility quotient is among the highest in the field. Matt is correct when he writes that critics “offer authentic thought and insight rather than data sifting with a ‘skin’ that makes it seem individualized and personal.”
I don’t doubt that Next Glass will have some success. They certainly got a lot of free media: coverage ranged from business publications to the San Francisco Chronicle, CNN and the Wall Street Journal’s WSJ.DLive, an online digital conference.
People, especially younger people, like apps; they seem to fit in with today’s fast-paced, iPhone-connected world. As one of my tattoo artist friends explained to me, “If you’re the guy at the table with the app, you’re looking cool.” And Next Glass, in particular, appears to be “scientific” in a way that might appeal to folks who resent being told what to do and think by others, particularly older men and women whose lives bear little resemblance to their own.
Far be it from me to suggest that wine critics are the end-all and be-all of wine recommending. But I was one myself, for a long time, and I know that world pretty well; and some voice inside me, which I have learned to trust because it’s usually right, is telling me that the individual wine critic, with all his flaws and virtues, is going to continue to be important, because–let’s face it–everything that purports to replace it isn’t as good. When a Gen Y’er uses Next Glass to buy her next glass or bottle of wine—and doesn’t particularly like it, or finds her friends don’t like it—Next Glass’s limitations will be clear.
Steve, take a look at this:
“Turning White Wine Into Red: Recommendation Failures The Alchemy Of Context” (http://recommendationinsights.com/?p=238)
and …
“What Problems Does Tribes Solve & How?” (http://recommendationinsights.com/?p=48)
and …
the other articles on the site that take a close look (backed by a fair number of scientific studies) about how current methods fail and how those are addressed by Tribes, an algorithm I invented. Patent filed last year.
Can a machine and software replace a wine critic? Of course they can.
But Next Glass isn’t doing this. A wine critic does three things:
1. Determines the quality of the wine
2. Describes the wine in a way humans can understand
3. Provides additional color to give the wine story elements that help distinguish it
Next Glass does zero of these.
I don’t know anyone at Next Glass and wish them the best of luck, but imho they are headed down the same dead-end that we’ve seen so many times before. People don’t just need to know which wine to buy, but *why* they should buy that wine. Imagine you went to Amazon to buy a, er, vacuum cleaner and there was a single product with a 94.7 compatibility rating. Are you going to press the Buy button? No effing way. Wine is no different.
There are a ton of other problems such as the inability to capture any context information that is fundamental to the assessment of a wine, the inability for humans to consistently evaluate a wine (which again is influenced by context), the reality that there are only ~ 8-10 sensory characteristics that really matter (fewer for beer – probably alcohol and IBUs serve 90% of the purpose), the statistical unreliability without a massive data set for an individual (the cold start problem), and so on.
BUT, what if Next Glass used their stuff for actually describing wines? For evaluating quality? Maybe then combined that with support information about the brand, vintage, varietals, etc. … and automated spinning together stories that sell wine. Then I’d be nervous if I were a critic.
The next great wine critic will be a social networking site – WineFaceTwit or something similar – it will be based completely on emotion with no logical or rational thought (oh? wait? just like other social media!) behind the wine ratings, such as “this wine sucks because I don’t like the blue on the label! it clashes with my socks!” or “I haven’t tasted it because it’s red and I only drink white wine, therefore it’s awful!” or “I heard the winemaker spits it out when he drinks it so I’m not buying any!” or “wine has a cork? who knew?”
I haven’t really delved into Next Glass yet, other than downloading the app and playing with it for 5 minutes. That said, there is something here. With all due respect Michael, consumers do make decisions based on “compatibility” — it is called Pandora. And websites like metacritic have replaced individual critics with the wisdom of the crowd. I can’t tell you the last time I reach an individual movie review from start to finish. I am not sure if Next Glass is purely based on the chemical analysis or if it also incorporates a human element — I think Pandora does both.
Having written a business plan back in 2007 that resembles what Next Glass does (but back in the ancient ages of building websites instead of apps), I am biased to think there is something here. Will they replace critics? Absolutely not. Will they potentially provide better recommendations for individuals and influence purchase decisions, I think yes. And considering the thousands of wines made, critics cannot cover the long tail as well as the crowd or the machine.
There was an article on Vivino’s score correlation with critics yesterday. Steven, you should reach out to Vivino and look at how your scores correlate with their users’ scores. And it would be interesting to know if Next Glass recommends wines to you that you have liked.
But Damon Levy, why would I care if my scores as a critic “correlate with their users’ scores”? I don’t. For a critic to be excessively concerned about that is to undermine the entire concept of criticism.
Steve, intellectual curiosity, nothing more. It shouldn’t change whether you critique and how you critique. For me, I am curious if the crowd’s ratings match a critic’s ratings. To take it a step further, I think these apps could create a few different groups of users that match up with different critics.
Because critics can’t taste every wine and my opinion is that every critic has a different palette, it would be useful if an app could say, “your preferences match up with the type of wines Steve Heimoff loves; while he didn’t taste this wine, our app thinks he would like this.”
Hey Damon… This isn’t really a Next Glass-specific issue for me. It’s just that it’s another in a long list of companies who have a good idea in abstract but then get sucked into the Bermuda Triangle of the wine industry (as always, I’m happy to be wrong and appreciate counter-arguments). OK, here we go…
Wine and music are nothing alike. The cost to a consumer to listen to a song recommended by Pandora is zero. The cost to try a bottle of wine is $20, $50 or more. Let’s change the rules a bit and now say that the way Pandora worked was that instead of playing songs for you, it displayed a Buy button where you could buy the entire release for $9.99 – but you couldn’t listen to it… nor could you read about it. “Trust me” says Pandora, “I’m science.” Really that’s what we’re talking about here and I think you can pretty clearly see why that doesn’t work.
As an aside, I think Pandora is based primarily on collaborative filtering (you like A and B, I like A, therefore I will like B) whereas Next Glass is based solely on one person’s preferences (I like A which has features X1, X2, X3 but no X4; therefore I will like B which also has X1, X2 and X3 but no X4). Philosophically Next Glass doesn’t care what other people think – only about how your own unique physiology interprets wines. That sounds great on paper – now I don’t care about critics or amateur reviewers. Makes all the sense in the world.
Except that’s not how humans make reasoned purchase decisions. If you can’t explain *why* I should a wine, then I’m not going to buy it. “95.4” is not an answer. Nor is “high correlation of absence of 4EP.” The thing we cannot forget is:
WINE IS SOLD, NOT BOUGHT.
That is what makes wine so different from everything else. That’s why human sales reps and somms are still the best mechanisms for initiating transactions. That’s why Amazon Wine is going nowhere. Amazon is awesome at helping people buy stuff but doesn’t even begin to know how to sell.
BTW, I’m not saying recommender systems are useless in wine… after all, some information is better than none. But I’m saying that over-simplified models that come at it from an algorithm’s perspective and not a human buyer’s perspective have proven over and over and over again to simply consume lots of venture dollars and not further our understanding of how to crack the online wine nut.
The metacritic point is a good one – as you know, I’m a big fan of that for wine. But, again, it’s not an apples to apples comparison. A movie review takes several minutes to read. The real selling of that movie is a 30 second trailer plus a quick metacritic check to make sure it doesn’t suck. But if all you had was a metacritic score and a movie title, then that’s not helpful.
I can’t wait until someone invents an app that will just drink the wine for me
From the Los Angeles Times “Main News” Section
(September 3, 2006, Page Unknown):
“Danger, Will Robinson — No Good With Fish;
In Japan, robots have become wine taste testers.
But sommeliers’ jobs are probably safe.
The devices can rate just a few dozen vintages.”
Link: http://articles.latimes.com/print/2006/sep/03/news/adfg-winebot3
[Missing from online version: accompanying photo exhibit]
By Eric Talmadge
Associated Press
The ability to discern good wine from bad, name the specific brand from a tiny sip and recommend a complementary cheese would seem to be about as human a skill as there is. In Japan, robots are doing it.
Researchers at NEC System Technologies and Mie University have designed a robot that can taste — an electromechanical sommelier able to identify dozens of different wines, cheeses and hors d’oeuvres.
“There are all kinds of robots out there doing many different things,” said Hideo Shimazu, director of the NEC System Technology Research Laboratory and a joint-leader of the robot project. “But we decided to focus on wine because that seemed like a real challenge.”
Last month, they unveiled the fruits of their two-year effort — a green-and-white prototype with eyes, a head that swivels and a mouth that lights up whenever the robot talks.
The “tasting” is done elsewhere, however.
At the end of the robot’s left arm is an infrared spectrometer. When objects are placed up against the sensor, the robot fires off a beam of infrared light. The reflected light is then analyzed in real time to determine the object’s chemical composition.
“All foods have a unique fingerprint,” Shimazu said. “The robot uses that data to identify what it is inspecting right there on the spot.”
When it has identified a wine, the robot speaks up in a childlike voice. It names the brand and adds a comment or two on the taste, such as whether it is a buttery chardonnay or a full-bodied shiraz, and what kind of foods might go well on the side.
Shimazu said the robots could be “personalized,” or programmed to recognize the kinds of wines its owner prefers and recommend new varieties to fit its owner’s taste. Because it is analyzing the chemical composition of the wine or food placed before it, it can also alert its owner to possible health issues, gently warning against fatty or salty products.
That capability has other useful applications. Given three ripe, identical-looking apples to analyze, the robot was able without taking a bite to correctly single out one as sweet and the other two as a bit sour.
But sommeliers need not fear for their jobs just yet.
Of the thousands of wines on the market, the robot can be programmed to accurately identify only a few dozen at most. It also has more trouble with the task after the bottle has been opened and the wine begins to breathe and thus transform chemically.
“Wines are notoriously similar in their spectral fingerprints,” Shimazu said. “The variation this robot detects is very subtle.”
Some of the mistakes it makes would get a human sommelier fired — or worse.
When a reporter’s hand was placed against the robot’s taste sensor, it was identified as prosciutto. A cameraman was mistaken for bacon.
The 2-foot-tall robot also is expensive.
“Buying one of these would cost about as much as a new car,” Shimazu said. “We’d like to bring that down to 100,000 yen ($1,000) or less for the tasting sensor if we were to put it on the market.”
He said there is no plan yet to actually market the robot, though the sensor could be available as early as next year.
“We are getting a lot of business offers and a lot of interest,” he said. “But we see this more as a symbol of our technological ability than as a profitable product right now.”
Mie University engineering professor Atsushi Hashimoto, the project’s other co-leader, acknowledged there is much room for improvement.
But he said the robot could be used in the near future at wineries to test the taste of each bottle without actually unscrewing any corks.
“It’s still like a child,” he said. “But not a completely ignorant one.”
Industry experts note the shortcomings but agree on the robot’s possibilities.
“I see the potential to analyze expensive and old wine to say whether it is authentic or not,” said Philippe Bramaz of the Italian winemaker Calzaluga. “Auction houses such as Sotheby’s and Christie’s could use this technology to test wine without opening it.”
I have a wine friend who did a post-doctorate in chemistry at Caltech.
He took a job at Gallo in Modesto, tasked with bringing higher-tech to the wine world. (He no longer works for Gallo or any winery.)
Had access to all the “shiny new objects” and Ph.D.s that Next Glass has — and more. (Gallo has deeper pockets.)
I brought Next Glass’s website to his attention: http://nextglass.co/#/Solution
“We determine the beer and wine’s DNA by using Gas and Liquid Chromatography and Ion Trap Mass Spectroscopy, all processes that determine the chemical makeup of each beer and wine.
“These processes allow us to look at the beer and wine on a molecular level, uncovering the unique qualities of each bottle and helping us create the beer and wine’s overall enjoyment profile.
“Our team of prominent PhD scientists and expert lab techs developed and maintain the revolutionary Genome Cellar.”
His comment:
“Interesting. But the technology and application have been around for awhile.
“I don’t like the concept of wine ‘DNA.’ It is a big misnomer, but it is an attempt to explain what they are trying to do. It implies that there is a unique and non-changing chemical marker in every bottle of wine that if you can access it will tell you if you will like it or not. I don’t think it’s that simple. Does their technology tell you the interaction between the flavor compounds and the tannins in the wine? So they have to get both the flavor profile (GC) and the tannin profile (LC) and then figure out how they interact. The big issue here is that flavors and tannins change over time due to oxidation. So are they going to take a particular vintage and profile it over the years to see how the changes occurs? Can they model the change for every vintage? Also, do they look for flaws in wine, such as sulfur compounds? And then there is the interaction with wine and food, how are they going to figure that out?
“This technology is really useful for producers who are trying to make consistent and high quality wines. I really don’t know how useful it will be to consumers.”
I suspect that the day will come when an app will contain information that is more or less useful to a wine purchaser, but it will not be a machine that does it. Not unless the company selling the app has the ability to taste thoussands and thousands of wine and update that app containing comments on those wines on a near real-time basis.
Of course, the machines behind that app will need to be able to recognize the four thousand esters that make up aromas in wine and be able to determine if those esters are appropriate for the type. It will be one thing to be able to describe wine by style. It will be another entirely to be able to use a machine to judge quality.
I will admit to finding great faults in crowd-sourcing solutions, but I would trust crowd-sourcing commentaries before I would trust machine-based assessments of quality.
Charlie, funny enough, measuring 4,000 esters is probably what Next Glass is (or will be) good at! To the extent that there are prototypes for a style, comparing the two *seems* pretty straightforward. So it’s likely going to be good at judging typicity.
Theoretically it might be able to judge quality as well where quality is defined as similar characteristics to well-reviewed wines. Enologix has done this for years – albeit with a handful of attributes.
If Next Glass can provide sufficient value to wineries, then wineries would even pay to have their wines analyzed, just like they do with labs today. So they could overcome the coverage problem they have today.
Once they deal with all that, then the only challenge they have is translating all of it into a format optimized for human persuasion. Having spent a little bit of time playing around with natural language generation from highly-structured content, this is definitely doable.
If they pull all that together before the money runs out, then they may really have something. But it’s a big if.
This is actually one area where I think machines will beat crowds. Let’s mark this thread to look back on in 2018.
Phenomenal topic and thread here.
Can’t agree more that showing consumers the ‘why’ rather than ‘what’ they should buy is critical.
As well, that professional critics’ knowledge and expertise would always be valuable.
In addition, that the voice of the community (consumers) will keep on getting louder, not lower, more important, not less important as consumers begin to follow their individual palates more and more, rather than blindly follow others’ opinions.
Intimidation is going out the door, being replaced by a more confident consumer who trusts their own knowledge, likes and dislikes. This new knowledge is being enabled by technology. Not stars and thumbs, nor by machines and chemistry. Technology that captures individuals’ preferences and experiences with wine in great detail, and shows them the ‘why’ that nurtures and builds the confidence and knowledge they need.
If I were satisfied to drink only one “type” of wine all the time the machine approach might have some value, but just like the foods I enjoy, it varies with season, mood and environment. The organic human taste/smell system may not be quantifiable to the degree that a machine is, but with experience professionals and civilians alike get to know what floats their boat, in many styles.
Being in the former category, I taste at least a thousand or more wines every year, and most are soundly made, with some that are exciting, nuanced or interesting for various reasons. The rational for explaining “why” as pointed out in the discussion, can be accomplished better (I think) by a human with some experience in the subject.
If you are partial to oaky California Chards, then by all means buy them, but you shouldn’t need a machine to find them….the info is on the label. If you are partial to Bordeaux, look for Cabs and Merlot, etc. There is typicity by grape varietal, and growing region; if you can read a label and have depth of experience, then you should have an idea of what’s in the bottle. Does that sound like something a professional could write about or teach in a class? Sounds like it to me, but I’m biased in that regard.
Downloaded Next Glass and was disappointed that it couldn’t read a majority of wine labels I enjoy; Williams Selyem, Saxum and Alban to name a few. A manual search in the app didn’t yield a result either. So how can it tell me what I will like if it can’t register what I like? Hopefully they are still building their wine database because I was having a little fun building my beer profile on it.
So there are around 3 million wine SKUs in the marketplace today with ~ 150K added every year. Let’s say they’ve tested 15K wines? That’s 0.5%. Oh, and a bunch of those 15K million wines have changed since they were evaluated. So it’s almost like starting from scratch each time. You can see why this begets a strategy of focusing on 80/20 supermarket wines. The problem of course is that someone reaching for a bottle of Santa Margherita or Yellow Tail is the *last* person who is going to download and run a wine app and populate it with their preference data.
Which is why a pivot to beer makes more sense. There are far fewer beers, there aren’t vintages and the beer doesn’t change over time. And it seems to me that there are far fewer attributes that are relevant in beer – so you need less personal data to make useful recommendations. The question then is whether it solves a problem for anyone. I completely get how people are confused about wine, but I just don’t see that same level of confusion with beer. Anyone who cares enough to download a beer app can already discriminate at a category (e.g., IPA, pilsner, etc.) level. Then within those categories, don’t people want diversity… to try new things?
The sad thing is that if it worked and people used it, then at its very best it simply enforces homogeneity of consumption.
BTW, none of this is damning of using machines or even Next Glass as a company… hopefully they’ll figure it as they seem like smart and motivated folks.
Michael, I agree … but they would use Tribes if they knew they would be getting a reliable recommendation crowdsourced from many, many people who tried a wine and … would buy it again (minus usual drama and awkward prose) That’s how you remove the gambling part from the Vino Casino (http://wineindustryinsight.com/?p=55012)
I had a dream last night that I rolled over and my wife was replaced with you, excited and wide-eyed with a pitch on Tribes. 😉
But Lew, their entire argument is that taste is a personal thing and that recommendations from others aren’t useful because of the human physiology that is unique to us.
Your argument (that others share) is that if there is enough correlation between the likes/dislikes of two parties, then A can use B’s experiences as a proxy for their own preferences.
Two different approaches and they both have strengths/weaknesses. But my point (as I’ve made elsewhere) is that effective wine buying advice is only 20% an algorithm issue… the other 80% is about presentation or *selling* of that advice. I honestly believe that no matter how good your underlying reco algorithm, it will always be beaten by a system that can sell the *why* – even if that system is powered by a monkey rolling 4d20.
Michael,
I have been SOLD wine and it’s often a bad experience.
Read that link (http://wineindustryinsight.com/?p=55012). It describes having been SOLD … and how the process can suck a lot of the time.
For more than 40 years, I have been SOLD — have tried to rely on recommendations from informed wine store retailers, talented sommeliers, and wine reviews. And I know a little more than the average consumer about wine and how to describe it.
But regardless of how long I spend trying to describe my taste preferences, the wines I prefer and specific organoleptic characteristics of wine that I find pleasing, roughly 1/3 of all wines which I have been sold or winds which I would not buy again.
And some of them, even though they are well-made wines, I have poured down the drain because life is too short to drink wine I don’t like.
My experience is not unusual among wine drinkers. And the average consumer doesn’t care about all of the detail contextual descriptions – – and is even confused and put off by them.
Like the average consumer, what I want is a wine that I like. And I’m tired of being SOLD wine that I don’t like.
That is why I invented the Tribes algorithm.
My Tribes algorithm is intended to work *alongside* selling and context … not intended to replace context, critics, experts, and other valuable and beloved information and institutions.
But by crowdsourcing preferences that are expressed very clearly as whether or not someone would buy a wine, and grouping people buy those preferences, new recommendations can be made much more accurately than by context alone.
Michael, I do think you are correct that some form of crowdsourcing is the answer.
I believe that the correct information to be crowdsourced is an expression of purchased content that is clear: That expression would simply be “I would buy this again,” or “I would not buy this again.”
This gets away from being SOLD wine someone won’t like.
In other words, I can tell a retailer or sommelier that I like Château La Plonk because I prefer an intense, big wine. The person selling me that wine says, “I’ve tried that wine and like it very much. Because of that I think you would like the Bung Vineyards Petit Verdot.”
At least a third of the time that I am sold that one, I don’t like it. This has happened over and over and over until I simply stop buying wine.
It particularly hurts when that wine cost $30 or $40.
That frustration is the prime motivation for my invention of the Tribes. It’s also the reason why I don’t spend more than $15 or $20 for a bottle of wine … Because I find wine I don’t like at about the same rate at that price range as I do at $30 or $40 and more.
Like the average consumer, I would spend a lot more per bottle of wine if I knew that I was being sold a wine that I like.
Getting past all the winespeak and drama, the average consumer knows whether they like a wine or not.
But it’s more accurate to know whether you’d buy a wine again (or not), then it is to try and interpret another person’s contextual description and match it with your own description … Especially because each person’s experience of context is different, because it is colored by experience, culture, genetics, psychology, and other factors.
The problem with context, especially with wine, is that it is simply too full of drama. One does not need contact to know that they like Coke versus Pepsi.
One doesn’t need context to know that they don’t like brussels sprouts, or that they do like raw oysters.
Trying to describe contextually why someone likes raw oysters and Pepsi, is only wading into a thicket of genetic, experiential, emotional and cultural context that is absolutely of no use in accurately communicating the experience to another person who have not tried them.
Lew, that link does not describe being sold… after all, all of the wines were purchased at Safeway where there is no one to sell. It complains that *winery education* doesn’t work and I’d agree with that. But nowhere does it even imply that being persuaded at the point of purchase is a bad thing.
“My Tribes algorithm is intended to work *alongside* selling and context” – Good, now we’re getting somewhere. I am not saying algorithms are unimportant. I am simply saying that any reco system that does not include persuasion (including the “why” and setting buyer expectation) won’t get traction. Context is sometimes very important (e.g., ordering wine to go with my klav kalash), sometimes not (casually browsing at a wine store). But selling is always important for wine. Always.
“Michael, I do think you are correct that some form of crowdsourcing is the answer.” – I didn’t say that. It may be true, but not necessarily. On paper, wine critics could still remain supreme for years to come… alas, their general reluctance to see their primary role as facilitating wine purchases means that they don’t look at all this from a wine buyer’s perspective and adjust their offering to support them. Instead, they simply point out the flaws in crowdsourcing and hope it’ll go away. So, if we stick to that vector then, yes, crowdsourcing content OR machine-generating content will win. But like I said before, if Next Glass could use their data for *describing* wines and, for extra credit, even providing some rough quality measures, then I can see that as basically eliminating critics AND crowds. In the long-run, the machine will win and almost everyone will be replaced. Maybe there’s a little crowdsource cul-de-sac, but 5 years from now we will all get our guidance from a machine combined with software that can tell stories… and frankly it will be a much better experience than critics or crowds currently provide. I am 100% certain this is possible (actually just 98% because I don’t know much about organic chemistry). Pretty much every industry is going to undergo this transformation and it’s not clear until it’s too late and you’ve been made redundant. Wine is on its way. Here, let’s look at a popular example:
5 years ago in San Francisco the way you went from point A to point B (without a car or public transportation) was to call a taxi, wait 10-15 or more minutes, hop in and then pay the professional driver $18 and done. Expensive, not a great experience. Then today we have Uber, Lyft, etc. Now we have an order of magnitude more drivers, software optimizes their location and now we wait 3 minutes and pay $9… or we take Uber Pool which takes us 2 more minutes but we only pay $6. That’s great for consumers… but the “problem” is that we still have extra costs in the system – namely the human driver. 5 years from now when the first autonomous cars are ferrying us from A to B… our cost is now $3, our ride is safer and maybe by the time we walk out the door, the car drives up and we wait 0 minutes.
The parallels are obvious. Pros>Crowds>Machines. Pros lasted for a hundred years… a slight blip for crowds and then an eternity of machine.
So, yeah, I guess there are two separate issues… (1) how do humans interact with systems to get guidance on buying wine, and (2) what is the source of that guidance. I do believe these are separate issues and how you deal with #2 makes no difference until you deal with #1… and, AFAIK, nobody is working seriously on #1 (except of course traditional human sales people/somms).
Michael you missed the point onto scores:
number one: my wife who is an average consumer, was recently SOLD A bottle of a horrible Chardonnay by the wine aisle guy at Safeway. That’s the most common type of sale you will find.
Number two I have described my experience as an educated wine buyer being SOLD and how that doesn’t work wait too much of the time.
Are you Safeway’s example number one because wine is SOLD there and because I tend to buy there now because I don’t get any better advice from just wondering the labels that I do from the experts who try to SELL me.
For whatever it’s worth I’m making a lot more progress dealing with people at Pandora another music sites that I am in the wine business simply because they recognize the need and are willing to open their minds to a new idea.
End of my efforts it’s a waste of all of our time
I agree we’ve reached a point of diminishing returns here… but it certainly wasn’t a waste of time for me. Thanks for the back and forth… I always learn something.
http://sethgodin.typepad.com/seths_blog/2013/11/the-generous-skeptic.html
Bernard,
“If you are partial to oaky California Chards, then by all means buy them, but you shouldn’t need a machine to find them….the info is on the label.”
I demur.
The California Chardonnay producers could instantly raise their wine unit sales if they simply used their front or back label to declare two things:
Was the wine oak barrel fermented or aged (to inform enthusiasts who prefer that style), and
Did the wine go through malolactic fermentation (to inform enthusiasts who prefer that “buttery” style).
I meet countless wine shoppers in stores who ask for those types of wines, and the store salespersons are stymied: they don’t know which wines have those characteristics (outside of personal drinking experience).
Consumer packaged goods studies repeatedly declare that a consumer spends no more than 3 second reading labels in grocery stores before forming a “buy/not buy” initial impression.
Wineries control a valuable piece of “real estate”: the wine label.
Too few wineries use their labels effectively.
Bob
http://image.slidesharecdn.com/stevejobsvsbillgates-130528160028-phpapp02/95/steve-jobs-vs-bill-gates-8-638.jpg?cb=1370343218
On the role of wine reviews in “moving the needle” . . .
Excerpts from Wines & Vines
(February 9, 2015):
“Wine Consumers Thirsty for Other Beverages;
Sobering data from Wine Market Council paints wine drinkers as fickle”
Link: http://www.winesandvines.com/template.cfm?section=news&content=145997
. . .
In August 2014, ORC International surveyed a representative sample of all U.S. adults adjusted to current Census Bureau demographic data to segment consumers into non-drinkers, beer and spirits (but not wine) consumers, all wine drinkers, and high-frequency wine drinkers. They obtained 2,920 completed surveys.
The surveys determined that 36% of the population are abstainers, 27% occasional drinkers, 24% non-adopters (don’t drink wine) and 13% high-frequency wine drinkers.
By generation, 41% of baby boomers drink wine, followed by 29% of Millennials, 18% of Gen X and 12% of older people.
The survey of 1,001 high-frequency wine drinkers was conducted by Illuminate Research of Seattle, Wash., with a sub-segment of 321 high-end wine buyers. The respondents were provided by the Survey Sampling International panel of U.S wine consumers and focused on the wine-drinking population.
[Bob’s aside: Around one-third (321) of “high-frequency wine drinkers” (1,001 obtained surveys) were deemed “high-end wine buyers.”]
. . .
One of the most interesting questions was whether high-frequency wine consumers pay attention to REVIEWS. The highest number (31%) considered them somewhat important, while 13% said they were extremely important and 16% said not important at all.
More to the point, 63% of high-end wine buyers consider REVIEWS extremely or very important, which only 22% of other buyers do.
Among the frequent drinkers, Millennials depended most on REVIEWS, with 56% considering them very or extremely important, with figures dropping by age. Forty-two percent of Gen Xers said REVIEWS were important, 21% of baby boomers and only 15% of those age 69 and above felt REVIEWS were important.
[On the subject of where reviews might be seen . . .]
Younger consumers are tied to SOCIAL MEDIA, and the study found that 62% of Millennials and 40% of Gen X consumers use Facebook, 38% and 21% Twitter. Still, 46% of Gen X and 25% of Millennials don’t use SOCIAL MEDIA.
Let me be a “honey bee” and cross pollinate between Steve’s wine blog and W. Blake Gray’s web blog on the subject of wine bottle labels.
“What People, [Wine] Critics Hate in Wine Back-label Copy”
[Citing Harvard psychology PhD candidate Mark Allen Thornton wine blog titled “Buying Wine from the Back Label; Using data to decipher winemakers’ language.”]
Link: http://blog.wblakegray.com/2015_03_01_archive.html