Research

“You May Also Like” author Tom Vanderbilt on the mystery of taste and the imperfection of algorithms

“You May Also Like” author Tom Vanderbilt on the mystery of taste and the imperfection of algorithms

Why people like the things they do has baffled marketers, product managers and customer experience professionals for ages. If you know what makes people tick, you’re one step closer to delivering products, services and experiences that customers will love.

The importance of understanding customer taste is why we’re excited to bring Tom Vanderbilt, author of You May Also Like: Taste in an Age of Endless Choice, to the 2016 Customer Intelligence Summit in Chicago. His keynote at the Summit will explore the mysterious, elusive and often messy nature of people’s preferences.

Vanderbilt sat down with us to discuss what drove him to explore taste and the strengths and limitations of algorithms in predicting people’s preferences.

What motivated you to look into taste and write You May Also Like?

Tom Vanderbilt TasteAs a journalist, I’m interested in the everyday processes people engage in but tend to take for granted. One statistic claims that we make more than 200 food decisions in a single day. These decisions are mostly programmed; as humans, we tend to fall back on habits.

Taste used to have a bandwidth restriction. Now we have a whole universe of options to choose from. In this global market, the internet has made us hyper aware of what others are choosing. Taste has been democratized.

How are brands like Netflix, Spotify and Pandora changing what we know about taste?

These companies have done great work to move closer us to understanding taste. They have robust statistical models predicting what artists, movies or shows you might like based on your previous behavior.

Netflix, for instance, recently started A/B testing artwork. The order of the movies and shows presented to users is important, but so are the creative aspects. If users see three characters instead of one, will it lead to a higher click-through rate? Netflix tests in real time, perfecting its feedback loop in which consumers make simple choices, the company gathers data and refine its approach and then consumers are given new options.

Where do algorithms fall short when it comes to predicting taste?

Algorithms can’t pick up on why people don’t like or select various options. There will always be people who, for instance, don’t like a movie because of a specific actor, but the algorithm wouldn’t show that.

Keep in mind that media tends to be more idiosyncratic than durable goods. If I’m in the market for bolt cutters, for example, I might look at the overall user rating but won’t bother with reviews. With 500 reviews and an overall rating of four and half stars, I can assume it’s a quality product.

With a movie, a lot of people may like it, but they could have different taste than me (or be in a different mood), so I can’t rely on user ratings alone. The challenge for companies that rely on algorithms is deciphering what turns people off.

Unlike Consumer Reports, which tend to be based on an objective set of criteria, user ratings are an imperfect guide to what’s actually good. In fact, there’s very little correlation between products best-rated by users versus those that get high marks from Consumer Reports. Price and other dynamics affect user-generated reviews, so they’re far from objective.

“Think of taste in terms of three opposing dimensions: novelty versus familiarity; complexity versus simplicity; and conformity versus distinction.”

How can customer intelligence pros get closer to understanding consumer taste?

To start, think of taste in terms of three opposing dimensions: novelty versus familiarity; complexity versus simplicity; and conformity versus distinction. These three “tensions” account for what we like and dislike.

Taste is always changing. Consumers are often barely aware that they make hundreds of simple choices every day. Our understanding of taste is like a moving target—you need to consistently gather and test customer insight to stay competitive.

Your last book, Traffic, is a national bestseller. How did it influence You May Also Like?

My latest book is not a sequel to Traffic, but it utilizes a similar methodology. Both mix academic, scholarly research with interviews with real world people. In Traffic, I talk to engineers and government officials. In You May Also Like, I focus on people doing interesting psychological research.

They both highlight things that may seem like common sense. I looked into whether there’s research confirming those assumptions. The way I approached research is the main connection between the two.

If you like this Q&A, you may also like…

Learn more about Vanderbilt’s work during his presentation at the 2016 Customer Intelligence Summit on September 20. His keynote has important implications for the way you use customer insight to drive product innovation, marketing and customer experience strategy.

To hear more from Vanderbilt, join 400+ customer intelligence pros in Chicago this September and save your spot for the number one customer intelligence event in North America.

2016 Customer Intelligence Summit



Subscribe to the Vision Critical blog

Get free customer intelligence tips and resources delivered weekly to your inbox.

By completing this form you consent to receive emails from Vision Critical. You can unsubscribe at any time. Learn more in our privacy policy.