- After 8 years in business Pandora has 3 million paying subscribers and 68 million non-paying listeners in the U.S. The company is still not profitable. The stock trades at 95 times next year's earnings.
- In our 3-year growth projections we either run out of potential customers (not enough people in the U.S.), or we have to factor in a dramatic unprecedented increase in advertising rates. Even if the company increases its revenues per non-paying listener by 136 percent and more than triples its paying subscribers, it needs to grow to 278 million non-paying listeners to justify its current valuation.
- The stock is near an all-time high. Within 2-3 quarters the number of registered accounts at Pandora will exceed the number of adults in the U.S., or it will stop growing. In either case, investors will be prompted to re-evaluate their expectations. News about the adoption rate of iTunes Radio may act as another catalyst for a major correction in the near term.
- The stock price needs to drop 75-80 percent to bring the market cap in line with the fundamentals.
Pandora (P) is a leading Internet radio service in the United States. The company has just completed a secondary public offering selling 15.73 million shares at $25 per share. The company's market cap reached $4.8 billion. This article looks at the fundamental value of Pandora by evaluating its assets: intellectual property, paying subscribers, and non-paying active listeners, and examining its growth prospects.
Any Internet radio service can use the same content that is available to Pandora and pay the same rates. These rates are set by the Copyright Royalty Board, a tribunal established within the U.S. Library of Congress. Alternatively, providers can sign deals with the music labels directly. All the major competitors already have access to more music than Pandora. While Pandora has 1 million tracks in its library, iTunes Radio is reported to have 27 million. Nokia Music, Spotify and Google All Access are estimated to have 20 million tracks or more in their libraries. This means that the mere fact of having access to 1 million songs has no tangible value.
In my opinion, Pandora has a very basic website compared to other popular Internet services. It is making slow progress, but the website still looks very 2005. A look at the source code reveals many deficiencies that do not instill any confidence in Pandora's technological prowess. Mobile apps are nicer, but they offer nothing groundbreaking either. I do not see much value in the user-facing part of Pandora's technology stack.
I did not see a strong evidence of in-car apps getting a lot of traction. Personally, I prefer to listen to music using my smartphone, which connects to my car's entertainment system seamlessly, and I suspect that many users do the same. Therefore I cannot put any value on Pandora's automotive protocol efforts, but I doubt it commands a significant premium over development costs.
Streaming music simultaneously to millions of people on various devices is a technical challenge. A company that is new to this business may find Pandora's technology interesting. Most likely buyers, however, already have streaming capabilities that are equal to or superior to Pandora's. Apple has iTunes, Google has YouTube, All Access and Hangouts, Microsoft has Skype and Xbox, Sony has PlayStation, Amazon has Prime Video, and so on. And there is no shortage of start-ups streaming all kinds of content.
Some of the potential suitors may find the Music Genome project (a recommendation engine) and the database of user preferences to be valuable. Apple, of course, has its own Genius technology, and it certainly has more data on user music preferences than Pandora. Google has no shortage of smart engineers and a lot of user data between All Access and YouTube. Microsoft probably feels that a combination of Xbox Music and Nokia Music gives it enough in-house expertise. Other buyers may be more amenable to put a higher valuation on these assets, but within reason.
Finally, I should mention that Pandora is currently defending itself against four lawsuits that allege violation of nine different patents by the company. As of January 31, 2013, the company had three patents of its own. On June 5, 2013, Pandora purchased a number of patents from Yahoo! for $8 million.
The bottom line is that most likely buyers will value Pandora's IP at somewhere between $20 million and $50 million - an immaterial number compared with Pandora's current market cap.
Pandora One Subscribers
Pandora reported that in August it had 72.1 million active listeners, an increase of 28 percent from the previous year. Listening hours, however, increased only by 16 percent, suggesting that listeners spent 9.4 percent less time with the service. This decline may be partially explained by changes in skipping limits (how many songs a user can skip per station or per day), but it's a worrying sign nevertheless.
Pandora offers two levels of service. Pandora One is a premium service that costs $36 per year or $4 per month. In May the company reported that the number of paying subscribers surpassed 2.5 million. Results for the second quarter did not mention paying subscribers, although the number was provided during the conference call: 3 million. After stagnating at the 12-13 percent level for several years, revenues from subscriptions increased to 21 percent of total revenues in the most recent quarter.
Paying customers are usually considered more valuable than free customers who generate advertising revenues. Pandora One subscribers represent only 4 percent of active listeners, but they generated 21 percent of total revenues - five times the revenue per user of a free service.
The company does not disclose two crucial metrics that are vital for estimating the value of its customer base: churn rate and customer acquisition costs. The company lumps customer acquisition costs with its costs to sell ads in its own service into a single line in the income statement. Surprisingly, not a single analyst asked about these metrics, that often make or break a business like Pandora, during the latest conference call or at the latest Investor Day conference.
I could not find any estimates for customer churn at Pandora. An estimate for Netflix puts its churn rate at between 40 and 50 percent per year. Sirius FM churn rate was reported at between 1.7 and 2 percent per month. If we optimistically assume that Pandora retains 80 percent of its subscribers each year (i.e. better than Sirius FM despite being a web-based service with plenty of competition only a click away), the average lifetime revenue from a paying subscriber totals $180. Subscribers with a monthly plan pay more, but they are much less likely to stay with the service for 5 years, so their lifetime revenue is likely to be lower than $180.
We know neither gross nor net subscriber acquisition costs, but they are higher than zero - the company spent $7.6 million on advertising in its latest fiscal year. The company also paid at least $4.8 million in subscription processing fees - more than 9 percent of its subscription revenue. Finally, the company has to pay for content: approximately $4.64 per active listener per year (based on the latest quarterly results), or $23.20 over a lifetime of a subscriber.
If we pin a lifetime value (difference between lifetime revenue and costs) of a paying subscriber at $150, 3 million subscribers equal $450 million, or 9.4 percent of the current market cap.
Note that the churn rate is always lower during the time of rapid growth in the number of subscribers. Churn rate will increase when this growth stalls, and it will jump further if the total number of subscribers starts to decline. A move in a churn rate to 25 percent, for example, will reduce the value of paying subscribers to $366 million.
Users of free services are usually more fickle than paying customers. Consider, for example, that the company boasted of 175 million registered users in its latest annual report, but it only had 65.6 million active listeners in January. The company reported that by the end of July the number of accounts reached 200 million, and the number of active users - 71.2 million, i.e. 129 million users - or 65 percent - were inactive. I doubt that most inactive users take months-long breaks from the service and then come back. It is more likely that a significant percentage of registered users stopped using the service without closing their accounts.
I believe that out of 200 million registered accounts at least 50 million accounts can be excluded as inactive for more than a year. This adjustment still implies that in any given month more than half of users do not listen to any music, so the real number of dead accounts is probably higher.
The company added 25 million registered accounts in 5 months, while the number of monthly active listeners increased by 5.6 million over the same period. These numbers indicate an annual churn rate of at least 37 percent.
Each active non-paying customer currently generates approximately $7.18 in advertising revenues per year. This means that the company makes $2.54 per non-paying active listener per year in gross margin. Assuming that all of these users cost nothing to acquire, and that the churn rate is 37 percent per year, the current lifetime value of each non-paying active listener is $6.86.
The total value of non-paying active listeners currently stands at $468 million, or 9.8 percent of the market cap, but only if we ignore the fact that the company has to keep a network of sales representatives in order to generate advertising revenues.
If we take SG&A expenses into account, the current lifetime value of a non-paying Pandora customer is negative.
The company can grow its revenues in one of the following ways:
1. Increase the number of hours users spend with the service
The number of hours per listener per month increased from 17.8 in 2012 to 20.6 hours in 2013 (fiscal years), or by 15.5 percent. This year, however, the number of hours dropped to 20.5 in the first quarter, and then to 18.3 in the second quarter. The number for August was 18.85. Overall, I do not see a clear upward trend here. It makes sense as Pandora competes not only with the other Internet radio services, but also with the other forms of entertainment, and with music libraries that its own users have on their devices.
2. Increase the advertising rates
Advertising revenue per thousand listener hours (RPM) has actually declined from $33.22 in 2012, to $29.13 in 2013 (fiscal years). While the mobile RPM increased slightly (from $21.05 to $22.53), this increase was more than offset by the decline in the desktop RPM (from $62.68 to $53.73). RPM improved to $28.02 in the first quarter, and $38.87 in the second quarter of this year.
Last year total advertising revenue was at approximately $6.87 per active non-paying listener. In the first 6 months of this year it increased to approximately $7.18 (annualized). It is possible that the company maintains its recent momentum, and the advertising revenues continue to rise. I have to note, however, that:
- the business is getting more competitive
- advertisers have plenty of other avenues for their online ads
- a significant increase in the number of ad slots per hour of service can push many listeners away
3. Increase in subscription price
I doubt that the company can substantially raise its subscription price without losing a significant number of paying customers given the very competitive landscape, and the fact that Pandora has a tiny music library compared with its biggest competitors.
4. Increase in the number of users
We should start by noting that 200 million registered accounts equal 86 percent of U.S. adults, while 71 million active listeners equal 30 percent of U.S. adults. The potential for further growth is limited.
The company claims that the number of registered accounts and active listeners may overstate the number of unique individuals, as some people access the service from multiple accounts. It is obviously the case given that the market penetration numbers referenced above start to look dubious.
Assuming that despite the increasing competitive pressures,
- the annual revenue per listener, net of content costs, improves to $6 from $2.54 in the latest quarter (136 percent increase),
- the stock appreciates at modest 10 percent p.a.,
- there is no change in subscription price, and
- churn rates remain stable,
in 3 years the company has to increase the number of paying subscribers from 3 million to 33 million (13.7 percent of all adults in the U.S.), or the number of non-paying active listeners from 71.2 million to 342 million (more than the entire population of the U.S.) to justify its current valuation.
You are welcome to combine these data points (lifetime value of $150 per subscriber and $16.21 per non-paying active listener) in any combination. For example, at 10 million subscribers the company will need 278 million non-paying active listeners.
It took the company 8 years to reach 3 million paying subscribers and 68 million non-paying active listeners. I find it hard to believe that within 3 years the company can reach any combination of paying and non-paying customers necessary to support its valuation.
International expansion is a possibility, but so far there is no evidence that business outside of the U.S. can be more profitable. There is little operating leverage in overseas expansion: new markets will require new licensing deals and new marketing/sales teams to generate advertising revenue. New markets will require substantial investments, depressing the earnings for years.
So far we looked at the company's assets. Let's take a different approach, and estimate Pandora's earnings potential.
Analysts following the stock expect, on average, that the company will increase its revenues by 50.6 percent this year, and further 38.5 percent next year. They expect the company to earn $0.03 per share this year, which implies a very strong second half (the company lost $36.4 million in the first 6 months, or $0.21 per share). Their expectations for next year are much higher: $0.27 per share.
The company is currently valued at 850 times its expected earnings this year, and at 95 times next year's earnings. Even if analysts are correct, this is hardly an attractive stock.
Assuming 10 percent stock appreciation over the next 5 years, and a down-to-earth P/E of 15 at the end of this period, Pandora will have to generate $2.74 in earnings per share in 2018, or 10 times (!) the projected earnings for the next year. Its earnings in 2018 will have to eclipse the company's sales in 2013. This is a tough goal for a company that already counts 30 percent of the U.S. adults as its active listeners, and that is still not profitable even with this large number of customers.
In any longer-term projections we quickly run out of people in the U.S. to support the required revenue growth, or we must project an indefinite fast growth in revenues per customer.
The biggest problem with Pandora's business model is that two of its largest cost items - cost of content and cost of procuring ad revenue - grow in line with the usage of the service. So the only reliable way to grow the bottom line is to stick more ads into each hour of music, or to increase advertising rates.
I estimate the current liquidation value of Pandora at $0.8 billion, including cash from the recent public offering, or up to $1.3 billion if an acquirer can generate the same advertising revenues at no cost, which is unlikely. This means that the value of the company's assets account for 17-27 percent of its current market capitalization. The remaining 73-83 percent are based on the expectations of investors regarding the prospects of a company. I believe that these expectations are greatly exaggerated.
Even if we completely ignore the customer base valuation approach outlined above and look exclusively at earnings projections, I see no evidence that the current P/E of 95 (next year's earnings) is justified.
I see no evidence that despite the growing competition the company can dramatically increase its revenues per customer to the point where it would need a less incredulous growth in its customer base to justify its current valuation. The operative word here is "dramatically": even a fantastic 136 percent increase is not enough to push the required customers number to a more realistic level.
The number of registered accounts at Pandora will exceed the number of adults in the United States within two or three quarters, or it will stop growing. In either case, the company will have a lot of explaining to do, and investors may be forced to reexamine their assumptions about its growth potential. News of adoption of iTunes Radio may serve as another catalyst for a large correction in the near term. I believe now is a good time to initiate a short position with a 6- to 12-month time horizon.
Pandora is the best short opportunity I have seen in a long time. A drop back to a 52-week low of $7 per share is very likely, even if the company continues to grow at its current pace. If the company stumbles, there will be little interest from major competitors to acquire it, unless the stock price goes down at least 75 percent.