Online Wine Shopping: Let the Algorithm Do the Picking

image by Jomphong via

image by Jomphong via



Would you trust a computer to choose your wine?
There’s a new generation of wine sellers counting on it.

Wine has been a tough sale online.
Wine shopping is daunting even in a traditional, bricks and mortar wine shop, where most customers wander the aisles a while and then end up grabbing an old favorite, an eye-catching label, or whatever’s on sale, with finger’s crossed that it won’t disappoint. It can be even more overwhelming online where the selection is inexhaustible and you don’t have store displays to cue you. Add to that a regulatory maze of interstate shipping laws, and by 2007, online sales were a piddling 3% of retail wine sales.

In the last few years, the internet has blossomed into a virtual vineyard.
Wine has benefited enormously from the rise of social media. There are thousands of online wine groups sharing tasting notes, alerting members to flash sale sites like Lot 18, and holding virtual wine tastings where on the count of three everybody uncorks and sips the same bottle. You can order wine for your Facebook friends through that site’s birthday reminders, and even Amazon, twice burned by failed wine-selling ventures, has jumped back in.

To succeed online, wine sites have to be more than just digital catalogs. Wine is consumed experientially, and in that sense its purchase has more in common with music or movies than with, say, a pair of shoes. That’s why the new generation of wine sellers looked not to Zappo’s but to Netflix for their sales model. And the secret sauce of the wildly successful video service is in the predictive algorithms that fuel their recommendations.

Online shopping has always run on recommendation engines.
The innovation was pioneered by Amazon, where now you’ll find them integrated into every inch of the shopping experience. From the home page through to the last click at checkout, Amazon beseeches you to consider ‘Frequently Bought Together’ items, ‘Customers Who Bought this Item Also Bought,’ and the less persuasive ‘Customers Who Viewed this Item Also Viewed,’ as well as ‘Sponsored Links,’ ‘Product Ads from External Websites,’ and a sidebar of  ‘More Buying Choices.’ Amazon’s algorithms skew toward building recommendation lists from items ordered by similar customer profiles. All the come-ons feel a bit like a traveling salesman with a foot stuck in your front door telling you about the vacuum cleaner your neighbor just bought.

Wine, like DVDs, requires more finesse.
Using its peer-to-peer comparative algorithms, Amazon derives a reported 10% of its book sales through recommendations on the site, while at Netflix recommendations drive 75% of the video viewing. Netflix accomplishes this through its algorithms, which turn an infinite buffet of data into a highly personalized, user-friendly experience. Instead of comparative recommendations, it builds individual profiles based on each customer’s individual preferences. It’s constantly throwing DVD titles at you, always asking your opinion about what you watch both on the service and elsewhere. Like Netflix, the new wine recommendation engines run on ratings. They build taste a profile based on what you’ve enjoyed in the past, and continually tinker with the profile as you rate your new wine purchases. And unlike Netflix, where the queue can get clogged with the entire Toy Story oeuvre, you don’t have to share this with your kids.

I’ll have what the MacBook Pro is having.
Try one of the new digital sommeliers:

Wine start-up Taste Factor, which compares the complexity of its recommendation engine to NASA, is like a custom wine-of-the-month club. Sign up for the subscription service and you get a starter pack of wine to rate. Your feedback establishes a tasting baseline, which is refined after subsequent monthly shipments, each of which is uniquely chosen for you.

Instead of NASA, Club W feels more like an online dating service. You start with a questionnaire—not about wine but lifestyle questions and details like how you take your coffee. The screen fills with potential matches, and you choose the ones that look good to you.

WineSimple also starts with a quiz to build each individual consumer taste profile. The geo-servicing phone app doesn’t sell wine, but it lets you know when you’re in a shop or restaurant that carries one of your recommended bottles.



Leave a Reply

Is it appropriate conversation for the dinner table? Then it should be fine.

Web Analytics