Mike On Ads

I was introduced to the online radio Pandora this afternoon. Pandora is the human interface into the “Music Genome Project”, a 5 year project that had human music experts classify every single piece of music created (except for latin & classical) since 1900 with various characteristics. Specifically: arrangement, beat, form, harmony, lyrics, melody, orchestration, rhythm, syncopation, tempo, vamping and voice (source). The Pandora way it works is remarkably simple. You start your own radio station by seeding it with a song. Pandora then analyzes all the genome aspects of that songs and finds songs with similar aspects to play. Even straight off the bat I found the selection to be absolutely amazing. Each song you hear you also have a chance of ‘training’ Pandora by giving it either a thumbs up or down. I threw 7 very varied and rather random songs at Pandora and was absolutely shocked by the quality of the selection.

Now here’s my question — to me music seems so much more complex than online ads. Lets think about this for a minute… lets say on average 50 billion ads are shown a day. That means that this year alone we’ve shown about 17 trillian ad-impressions. Lets be somewhat pessimistic and say an average click-rate of around .1%, or 17.8 billion clicks. How is it possible that we as an industry haven’t been capable of showing users more relevant ads?

One of the interesting things of course is that Pandora requires on a “seed” and then subsequent feedback from a user. Is the future world one where you start your day with a search on Google, which then gets treated as a ‘seed’ for your days ad-stream until you search for something different? Can we ever convince consumers that with only a little bit of their help we can greatly improve their online experience by showing them ads that align with their interests?

As an added note, I must say Pandora does a terrific job of integrating advertising with the site. Check out this “Bose Ad” that I received midway through my stream: