The Magical Equations Behind the Brands

Filed Under Mindjet

The Magical Equations Behind the Brands
Peter Beland

by
June 26, 2012

Algorithms are increasingly providing a more satisfying and personalized user experience on the Internet. Notice those concert suggestions on Facebook that pop up when you check out a favorite band’s page? Algorithms. Wonder how Google drums up hundreds of results for your cryptic “curling + Canada + futons + death metal” search request?  Algorithms. Ashamed of those anime romance recommendations from Netflix? Blame algorithms (or blame your little sister when your friends ask.)

Algorithms transform the Internet’s infinite buffet of data into a highly personalized, user-friendly experience. The process happens so seamlessly that few users are ever aware that they’re the subject of a complicated equation. How does the magic really work? Algorithms are essentially recipes—step-by-step directions for accomplishing a specific task. Step one, do this. Step two, whisk that. Step three, simmer. Step four, repeat. But in practice, these recipes are quite sophisticated. Think Heston Blumenthal’s chilli, not Rachael Ray’s meatloaf. And they’re of critical importance to businesses, who rely on them to crunch the massive amount of data produced by millions of web users worldwide.

“We have discovered through the years that there is tremendous value to our subscribers in incorporating recommendations to personalize as much of Netflix as possible,” writes Netflix engineering manager Xavier Amatriain. Indeed, the company contends that 75 percent of users’ content choices are informed by recommendations—either from a friend or the company’s proprietary algorithm, which notes customer viewing habits and ratings. Netflix’s algorithm is a critical component of the company’s customer retention strategy, and the video-streaming service hires the best minds in the business to constantly tweak and optimize it.

In 2006, Netflix went so far as to offer a $1 million prize to anyone who could improve the accuracy of their existing recommendation system by 10 percent. Although nobody could meet that benchmark, one team did win a “progress prize” after attaining 8.43 percent improvement with a combination of 107 algorithms. Two of those algorithms remain part of the company’s recommendation engine, and thus may be responsible for those Toradora! DVDs you keep getting in the mail.

To be successful, algorithms have to deliver not a “personalized” experience, but a satisfying one. Case in point: the most famous algorithm on the web—Google. The company’s self-imposed mission is to “organize the world’s information and make it universally accessible and useful.” Its remarkably efficient algorithm provides the sorting-and-retrieval framework behind that effort. The sophisticated formula is—of course—proprietary property fiercely guarded by the company and constantly analyzed by would-be search engine optimizers. Google notes that they look at 200+ different “signals” and assorted analytical metrics for each page to determine ranking. The company’s PageRank™ algorithm analyzes which sites have been “voted” quality sources by other pages.

Like Netflix, Google’s algorithm is the key to the company’s success. If a new start-up should ever come along with a buzz-generating search engine that blows Google out of the water, the Silicon Valley giant will go the way of Goliath. Thus Google never stops tinkering with its algorithm and metrics. To stay in the game, the company has to stay several steps ahead of known-and-unknown competitors.

The influence of algorithms extends well beyond Netflix and Google.  They, for instance, are even revolutionizing the undergarment game. True&Co, a San Francisco startup co-founded by a former Netflix engineer, uses algorithms to help female web shoppers find the best fitting brassiere without a fitting room. The company breaks bras down into 20 components and 50 points of data. As shoppers know, sizes vary across brands, so True&Co’s algorithm “translates” body-type information from brand-to-brand. The website then recommends and ships multiple items to shoppers, who only keep and pay for the ones they like. Feedback about the appropriate “fit” of the product helps True&Co fine tune its service, much like the star ratings on Netflix provide the company with input to help craft future recommendations.

Algorithms provide brands and marketers an opportunity to not only reach but connect with new and existing customers. As all of these examples demonstrate, the personalized experience that an algorithm simulates is just a means to an end. What customers want is results—to reach their preferred destination as quickly as possible. In fact, algorithms are literally providing this service in many new office buildings. “Destination” elevators that swap the classic two-button (up / down) arrangement for a more efficient “bin-packing” algorithmic system. Instead of stepping in the elevator in then choosing a floor, the destination system requires you to enter your destination on a panel in the lobby and then wait for your assigned car to arrive. The bin-packing formula reduces the number of stops between you and your exit by more efficiently routing passengers.  It might seem like the elevator has grown fond of you and prioritizes your trips above-all-others, but that’s just the magic of algorithms.

Image Source: www.iStockphoto.com

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>