I believe that Pandora (P) has the best technology in the world for automated music recommendation: the innovative "music genome project (MGP) ".
The MGP is an algorithm that defines and organizes songs according to 450+ attributes (which are basically keywords, like "metal influences" or "romantic"). A song is described by a vector containing these attributes up to various extents, and these vectors are then used to create a list of similar songs using a distance function. The most important step is the initial one: describing a song according to the predefined attributes. This is why the process is carried out manually: a musician analyzes the song in a process that takes about half an hour.
(S: 2012 Annual Report)
Did the process described in the first paragraph sound ideal? Well, you are not the only one. Pandora is excellent at what it does. However, this could equally turn out to be a curse as there is a trade-off between quality and quantity.
High-quality products come at a price. Pandora is not an exception. First, to add a new song to the database, Pandora needs at least 30 minutes to describe it according to the pre-defined attributes, in case the song were not already among the 10% which enjoy the expertise of more than one technician to ensure quality. Second, considering that there are now over 900,000 songs from over 90,000 artists in the database, Pandora also needs an active army of music analysts to keep itself in the game.
Therefore Pandora will always have the smallest database amongst its competitors. At first glance, that should be fine, since as a compensation, quality standards are kept high. However the difference in database size between Pandora and its competitors could be significant enough to determine users to ultimately sacrifice quality for quantity and migrate to other services, because a bigger database also implies a higher probability of finding new music. For example, Spotify has a catalog that already includes more than 15 million songs. Grooveshark, on the other hand, has a catalog made up of user-uploaded songs with an unlimited potential for growth. Similarly, Last.fm has over 45 million unique tracks.
Thus, from both a cost and a quality-quantity trade-off perspective, Pandora is at a strong competitive disadvantage.
To make matters worse the highest cost isn't technological. The highest cost stems from the need to acquire content: paying SoundExchange (an organization that collects fees on behalf of labels and artists), several agencies and Rovi for song and artist information. Furthermore, the royalties related to SoundExchange and agencies are variable: they are negotiated on a regular basis. If things go north in the future, expect these costs to increase, because the reason why they were reduced in the first place (in 2008) was to provide the company with a life-jacket.
I also believe that there is little that Pandora can do in order to have more negotiation power in this matter. If Pandora turns out to be profitable in the medium run, SoundExchange and other agencies will re-negotiate content fees. Pandora, which could be providing as much as 37% to SoundExchange's figures, will use all the negotiation power and tools it has to prevent an increase. In the end, they won't be able to avoid the increase because the relationship with its content providers is not symbiotic. SoundExchange could live without Pandora, but Pandora cannot manage without SoundExchange. The best possible outcome for SoundExchange would be to keep Pandora in the game by allowing space for some profitability in the near future. But in the long run, as the graphic shows, fees will continue increasing in proportion to revenues. In this way, Pandora's margins could be doomed forever:
Finally, there are limits to traffic set not only by high royalty rates (e.g. Europe) but also by legislation. Currently, Pandora is only available in the United States, Australia and New Zealand. For a company that derives more than 80% of its revenue from advertisement, this is not just a painful fact, but might imply certain doom. So far, Pandora has been able to minimize the effect of this fact thanks to an increase in the average rate of daily users in the U.S., but once it saturates the national market there will be little space left for improvement.
The recent announcement of a 56% increase in revenue and the departure of CEO Joe Kennedy confirm my beliefs. To me, Pandora is still in the early phase of the growth curve, because it could represent as much as 15% of the total U.S. radio listening market share. On the other hand, the departure of CEO Kennedy symbolizes the forced end of an era for Pandora: the company is desperately trying to replace idealism with monetization. Unsurprisingly, those who saw the income statement also know that content costs are increasing, representing now 60.6% of revenue in the fiscal year which ended January 31 compared to 54.2% of revenue a year earlier. This is consistent with the fact that the power relation between Pandora and its content providers is not symbiotic. And yet, shares were up nearly 20%.
Considering that Pandora cannot possibly be profitable due to its business model and the cost structure, I think the 20% increase in share price is unjustified. An unprofitable company cannot be worth $2.38 billion dollars. Sooner or later, a correction is bound to happen.