Pandora Uses Data to Tune Up Its Customer Experience

Customer Experience
Customer Experience
The music company is analyzing granular sets of customer data to present a highly engaging and individualized experience.

Music is a personal experience. The songs, styles, and genres that individual listeners prefer derive from a combination of factors including personal histories, mood, and time and place.

While individuals self-curate some of the music they listen to for an average of 3.7 hours every day, the vast majority-80 percent-of music listening is serendipitous, noted Simon Fleming-Wood, CMO at Pandora during last week's Forrester Forum for Marketing Leaders. "Most of us turn on the radio."

Listening to music should be a relaxing experience. Ideally, everyone should turn on their music device and have a song they like fill the room-or their headphones. The Internet radio provider is well aware of the importance that its user base-which has grown from 8 million in 2009 to 75 million this past March-experience music that they enjoy, without having to put much thought into finding the most appropriate channel.

While Pandora was doing very well, with its share of overall music listening having grown from 2 percent in 2009 to 9 percent earlier this year, the Oakland-based music company wanted to do more to make its personal listening experience the best possible one for individual customers by helping them find the music that best fits their preferences. "Pandora is trying to redefine radio in today's connected world," Fleming-Wood explained. This sounds like a daunting task for a company whose users listened to 1.7 billion hours of music over multiple devices using Pandora this past March alone.

Pandora's business leaders knew that the answer to their quest for personalization lay in the huge amount of data that its customers share while they listen, giving a clear glimpse into their music listening habits in a very granular fashion. Over the years the company has amassed huge amounts of data about each of its individual users. Personal insights didn't stop with demographic information and whether the listener had given a song a "thumbs up" or a "thumbs down;" Pandora also knows the time of day when individual customers prefer specific genres of music, as well as their preferred devices. For example, an individual might enjoy listening to upbeat tunes while driving to work in the morning but seeks jazz when he gets home at night, and on the weekends prefers easy-listening.

The obvious step was to turn this rich data into targeted recommendations for Pandora's millions of customers that would add value to their day-to-day experiences. Tapping into information from the Music Genome Project, which identifies different characteristics for a song and identifies others with similar qualities, teams of data and music experts within Pandora began working on algorithms that purse the data to gain insights on customer preferences, even as they change throughout the day. "We are predicting which music you want to hear and in which context," Fleming-Wood said. Because customer preferences vary not only between individuals but also for the same person according to mood and time of day, among other variables, one algorithm wasn't enough. In fact, Pandora has different algorithms for various customer personas, and then another algorithm on top of that to identify which of the algorithms are most appropriate for that individual user at a particular moment and on a specific device. "We wanted to go beyond just the data," Fleming-Wood explained. "We wanted to take what Pandora does for people, that feeling they get when they listen to spot-on music, and use it in marketing."

Inspiring increased usage

Like all other companies, Pandora wants to continue growing. Pandora wanted to inspire its users to use Pandora more. Fleming-Smith said it was imperative to find out more about users in order to deliver the best possible music to them. The company's online properties have a feature that allows users to give a particular piece of music a "thumbs up" or "thumbs down," which, Fleming-Smith noted, is an extremely important way to gather information about individual music aficionados.

While users were aware of this function, Pandora wanted to increase its usage and recognized the importance to make users aware of the benefits of using the thumbs and how this would impact future music recommendations. Last year the brand embarked on a marketing campaign to encourage its user base to actively give a "thumbs up" or "thumbs down" to songs, providing Pandora with additional insights that can be used to get that individual user to a perfect experience. "The more they use the thumb, the more we can personalize their experience," Fleming-Wood explained. Pandora created video advertisements showing the delight of customers when the perfect song comes on the radio and incorporated the videos on its Web interface, while audio reminders to use thumbs were used on non-video enabled devices.

The campaign, using both audio and video depending on the device, led to some impressive results. Fleming-Wood noted that the thumbs per session increased. In fact, in the week after seeing, or listening to, the advertisement, thumbs per session went up by between 20 and 40 percent across all ages. The company also saw an increase in the weekly hours across all customer ages, with the highest being in the 13- to 17-year-old age group, which jumped up from just less than nine hours to more than 11.

Further, Fleming-Wood noted that those who were exposed to the advertisements were 5.5 percent more likely to come back to Pandora. Additionally, positive perceptions about the brand also went up.

Listening to music should not be a chore. Instead, each listener should be able to switch on the radio and the exact song he's in the mood for should blast through his speakers-or headphones. This is Pandora's aim and something that the company is succeeding in doing.