How A Netflix-Style Recommender Is Vital to Reversing Wine’s Market Marginalization

Articles in this series include:

• Article #1: How A Netflix-Style Recommender Is Vital to Reversing Wine’s Market Marginalization

• Article #2: Why solving the “Paradox of Choice” is a major reason Netflix’s recommender is worth $1 Bil/yr. And why wine fails at that.

• Article #3: Reviews and 5-Star ratings are so useless for recommendations that Netflix ditched its prized $1-million algorithm.

• Article #4: Wine profile matching fails because genes determine that no two people taste the same wine the same way.

• Article #5: Netflix Added Big Data To Amp Up Recommendations. Why Wine Needs a Touch of That Too.

• Article #6: The Path To Netflix-Quality Wine Recommendations Leads Through The Doors of Perception

---

This is the first of a series of articles on why and how the wine industry can break out of its marginalized market position by using a Netflix-style recommender system adapted to the specific perception and intent issues demanded by the senses of taste and smell.

This dive into Netflix’s recommendation system was prompted by our series of articles late last year looking into Winc’s exaggerated claim that it is “The Netflix of Wine.”

Netflix retains 93% of its customers from year to year. Wine? 30% to 80% (Maybe.)

To write that Winc series required extensive research into the Netflix recommender system, its result successes and the globally unmatched ability to create a system that enables it to retain 93% of its customers. By comparison, the best winery member clubs manage, on average, to retain 80% of their customers for a full year.

The retention is far worse for standard subscription clubs like Winc or Tasting Room which have an annual retention rate estimated by wine industry financial insiders at about 30%. Conversely, that means they lose 70% of their customers from one year to the next.

Netflix customer retention and income soared when it got rid of customer ratings & reviews

Achieving that astounding 93% customer retention rate, according to two of Netflix’s top recommendation architects — was possible only after the company eliminated the use of ratings and reviews which they found questionable and inferior. We will revisit that article in more detail in upcoming articles.

 Wine’s Market Marginalization: The least-preferred AlcBev in America

Netflix does has an overall advantage because it’s not a heavily regulated and restricted alcoholic beverage. However, wine is the least-preferred of all alcoholic beverages. Here’s how the math works against wine:

• The legal age for consumption is 21 years, which includes about 70 percent of the 328 million people in the U.S. population.
• Of that proportion, about 65 percent consume alcohol, the rest being abstainers.

• Of those who drink alcohol, only 25 percent prefer wine

Why don’t AlcBev drinkers prefer wine?

Sadly, most people in the wine industry can’t answer the question of why 75% of AlcBev consumers don’t buy their wine.

This is because the wine industry, as a whole, has very little, current, valid data on which to base decisions. It’s almost as if wineries are afraid of knowing the truth and prefer to conduct their strategic planning from what they can extract from their individual gluteus maximii. There may be decent research done by large organizations, but those are locked away behind paywalls.

However, one bit of research indicates that people don’t drink wine because it’s hard to choose one they like.

Why do AlcBev Consumers have trouble finding a wine they like?

Because all the available information — reviews and ratings — fail. And those who try a wine are likely to be dissatisfied with their choices.

As a result people are not motivated to try wine. What’s more, those who do give it a try  rarely continue because the lack of an accurate, recommendation system makes it difficult to find wine that suits their individul taste.

Too many choices

As we will see later in this series, the presence of too many choices  — “The Paradox of Choice” — also prevents people  from even considering wine. Because of all these issues, wine does not retain its customers. This means that efforts to attract new customers are unsuccessful and fail to expand the overall market base.

Ratings and Reviews (as Netflix learned) Are A dead-ends on the path to customer retention and financial viability

Because reviews and ratings still dominate wine’s recommendation methods (despite their known failures), wine — and other “products of taste” — must follow Netflix, abandon ratings and reviews and develop a method that provides more accurate recommendations which can increase customer growth, satisfaction and and retention.

Importantly, the sensory differences between even Netflix (sight and sound) and wine (smell and taste) mean that Netflix could nit be tye Netflix of wine without the addition of a significantly difficult new capability: the ability to capture individual consumer perception and intent.

This series

This series of articles (see topics below) have not all been completely drafted  at the time of this publication. In the two months we have been researching and writing this series, we have found that the completion of one article often requires additions and/or corrections in previous articles.

Because of this, each article in the series may need adjustments after publication. Updates and edits will be noted in the body and headline of all articles.

Topics to be covered in this Series

The following bullets represent articles that are finished or in the editing stages. The order of the titles represent the current assessment of chronological order. That order and the final number of articles may to change.

Some articles may be too long for online reading and could be split into two or more articles. Alternately, a listed topic may be incorporated into another title depending on the amount of information or other reorganization.

Articles Finished:

• Netflix’s recommender earns it $1 billion/year by solving the paradox of choice (Article 1)
• Netflix users and videos versus wine drinkers and brands – Data and calculations (Sidebar to article 1)

Articles in Progress:

• Why reviews and ratings fail so supremely. Reviews and ratings were the main recommendation methods in the early online days. Netflix had to use them when its started its operation with physical products in the CD rental and sale business.
  
• Then came Ringo: The 1990s music recommendation paradigm that popularized Collaborative Filtering (CF) –still underpinning today’s recommendation systems. Major original illustrations.
• How CF and streaming media changed the Netflix recommender. The structure of the current system as described by its architects and developers.
• Netflix and Collaborative Filtering and the personal data the company gathers, stores, and uses to make recommendations. Original illustrations.
• Videos vs wine … sight and sound vs smell and taste. Why wine is a tougher recommendation nut to crack than movies.
• Why even Netflix probably couldn’t be the Netflix of wine without a adding perception and intent to its recommendation paradigm. Collaborative Filtering relies on observable human actions but cannot answer “why?” (perception and intent).
• Finally, to make perception-based recommendations work, wine needs a system to work with its data silos, fractured, physical-product distribution, and sales system. Netflix has the advantage of a mostly self-contained ecosytem. An all-digital, unified, instant recommendation and feedback system can solve that issue.

Article access

The time and resources required for this series demands that they be offered on a premium subscriber basis.

However, the significance of the information and the public interest demands that the articles be widely and freely available.

Because of that, premium subscribers to Wine Executive News will be the first to receive the full text of each article. Full, free article access will be made available after an appropriate delay: possibly a week, but the length of time may vary.