Pandora’s Business Model Doesn’t Work, and Can’t Work

The music industry might have set up the subscription music market in such a way that nobody can make money, no matter what they do


We all know that Pandora is a great service. OK, sorry, we all know that Pandora is a really, really great service!

Pandora was the first real music subscription service: a simple user interface combined with super-clever recommendation algorithms provided endless hours of listening pleasure. Thankfully, the service remains as valuable today as it was when it launched.

Unfortunately for Pandora fans, this article is not about the excellent service provided by Pandora – it is about Pandora’s business, or more exactly, Pandora’s business model.

If you want to understand Pandora’s business model then do this: set up a stall in your local market and try selling 1 dollar bills for 80 cents each.  You’ll find that have no problem attracting customers. Actually, if you waited long enough, you’d have no problem attracting 200 million customers - which is where Pandora is right now.

But you’d be getting killed on each sale, and the more customers you signed up, the more you’d get killed.

Clearly, most people would not regard this as a business, but this article will show that this non-business has a name: Pandora.

Pandora’s variable cost structure

For the year ended 31 January 2013, Pandora derived 88% of total revenue from advertising. This figure has not changed much since 2008 when advertising accounted for 93% of total revenues.

Pandora is therefore an ad platform. In a way Pandora is just like Google, or Yahoo: for the last three years, advertising has accounted for an average of 93% and 82% of total revenues at these two companies respectively.

We can determine from Pandora’s SEC filings that the company has been earning an average of $0.0287 per listening hour in advertising revenues for the last 4 years – with the variation in per-hour ad revenue over that time being less than 9% (a maximum of $0.0312 in F/Y 2011 and a minimum of $0.0268 in F/Y 2013).

To see the problem here, let’s imagine that Pandora serves an average of 4 advertisements per hour (some ads served in the audio stream and some served on-screen/in-app). On average this means that Pandora earns about $0.0287 / 4 = $0.0072 in advertising revenue per ad unit.

Pandora’s business involves selling the ad inventory created when users listen to music: the more ad inventory is created, the more ‘products’ the company’s sales team has to sell. Ad inventory increases when new users sign up and also when existing users listen for longer periods.

Ad units are served to the available inventory. In this analysis we will assume that Pandora’s sales team sells all available ad inventory and so all available ad units are filled.

We can loosely compare a Pandora ad unit to the ad units that are served by companies like Yahoo or Google.

Although a gross simplification, Yahoo’s business involves selling the ad inventory created when users request pages on the company’s website, and this is also true of Google. (although both companies also serve their ads on third party sites by their affiliate programs, which Pandora currently does not). Like Pandora, Yahoo’s ad inventory and Google’s ad inventory increases when more users visit the company’s websites and also when users request more pages.

Now consider the variable cost that Pandora incurs in order to serve one unit of ad inventory, or one ad unit.

There are two variable cost elements: (i) Pandora has to stream the music; (ii) Pandora also has to serve the ad unit. These costs are analogous to the costs Yahoo and Google incur in serving their web pages and serving ad units on those pages.

Streaming costs

As shown in the table below, Pandora’s per-ad streaming costs per ad unit are likely to be larger than the costs incurred when Google and Yahoo serve a typical page on their websites.

This is not true for YouTube or Google Maps, which have very significant per ‘page’ data payloads but when one looks at the total volume of ad inventory created by Google and Yahoo (most of which involved serving text-based pages) then we would still say that the average content payload that has to be delivered to support one ad unit is substantially less than 16MB (Pandora Minimum Recommended), let alone 33MB (Pandora One).

Separately, when the whole online ad market is considered we note that the size of the average web page averages around 1MB (excluding ads).

The other way to look at this is to say that even if Pandora was not disadvantaged in regards to the data payload that has to be served per ad unit, then the company will certainly be disadvantaged when compared with Google and Yahoo when one looks at the cost incurred per ad: because Google and Yahoo have vertically integrated their content delivery networks solutions, their marginal costs of delivery are likely to be far lower on a per-MB basis than Pandora.

In summary, when compared with Google and Yahoo, Pandora suffers two incremental costs per ad unit served firstly because the company has to deliver a substantially higher content payload per ad unit and secondly because the company will be unable to achieve the same per-MB content delivery costs.

Pandora also then has to serve the actual ad units, but the cost for this is unlikely to be significantly different to the costs that Yahoo and Google are paying to serve their ad units.

So far, the per-ad variable costs for Pandora look significantly higher than Google and Yahoo.

But there is another important difference: royalties.

Royalty payments

Pandora has to pay the music industry a royalty to serve the ad and there is no equivalent cost for Yahoo or Google (This is not true in the case where the user is viewing a music video on, say, YouTube. In this case, Google has to pay the music industry a royalty for this view – in a similar way to Pandora. However, when viewed as a percentage of Google’s total ad revenues, the ad revenue from YouTube is quite small and so, when viewed at a company level, we can neglect this effect).

Pandora’s SEC filings show that the company has been paying an average royalty of $0.0182 per listening hour for the last 4 years – with the variation in per-hour royalty over that time being less than 2%.

If we return to our assumption about Pandora serving an average of 4 ad units per hour then the royalty cost per ad unit works out to be about $0.0182 / 4 = $0.0046.

So, in summary, for Pandora:

  • Average advertising revenue per ad unit served:       $0.0072
  • Pandora-specific music royalty per ad unit served:  $0.0046
  • Percentage of ad revenue that is royalty:                        63%

Music royalties can now be clearly seen as really huge part of the revenue earned for each and every ad the company serves.

Proportionate operating cash flow

The seriousness of this problem is best illustrated by looking at company’s proportionate operating cash flow (that is, the cash that the company’s day-to-day operations are creating expressed as a percentage of revenues).

We have compared Pandora’s proportionate operating cash flow with a few other representative companies (Google and Yahoo – both ad platforms, like Pandora), ITV (a UK-based commercial television broadcaster who, like Pandora, operates an ad-supported, content-based business) and Sirius Satellite Radio (a subscription-based radio service provider).

As can be seen in the chart above, the results of this exercise are stark: Pandora is a clear outlier with a level of performance that is way below what a rational investor would expect: over the last three years only around 2% of Pandora’s revenue makes it through the company’s day-to-day operations to emerge as operating positive cash flow. But other representative companies typically achieve around 20% or, in the case of Google, about 30%, which is between 10x and 15x better than Pandora.

Worse, as can be seen from the above chart, the trend in this key operating metric is heading in the wrong direction – with the company’s operations having been generating proportionally less cash for the last three straight years.

There is a simple reason why Pandora is performing so much worse than these other companies: music royalties are dragging down Pandora’s business. 

It’s worth remembering that Pandora is not exactly a start-up: the company was launched in 2005. We would have expected that after 8 years the company’s core business operations would have stabilised.

It seems to us that if the company was ever going to achieve the level of profitability that other comparable businesses mange to achieve then we would have seen signs of it already.

But we have not.

Pandora has never made a profit and we think that the company will never make a profit, unless there is a major change in strategy. 

Things could get even worse for Pandora

We think that Pandora’s current loss-making situation could get worse:

  • Losses will increase with each new user and with each incremental listening hour: As noted above the downwards trend in proportionate operating cash flow is very worrying: the fact is that in spite of strong growth in users and  listening hours per user, the company is heading deeper into the red with each year that passes.

    It is clear that the problem lies at the heart of Pandora’s business model – its day-to-day operations are simply not profitable. Without a major change in strategy, this means that as the company grows, losses will increase.
  • Royalty payments are likely to increase in 2015: Pandora will need to renegotiate its license agreements with the music industry in 2015 and it seems very unlikely that there will be any appetite on the part of the music industry to lower royalty payments.

    If anything, the music industry will be keenly awaiting an opportunity to further milk a growing market (Pandora backed a bill presented to Congress on 21 September 2012 campaigning for lower royalty fees but this was not enacted, probably because lobbyists working on behalf of the music industry pushed the debate in the opposite direction.). We project that music subscriptions will grow strongly in the coming years to account for 47% of all digital music revenues earned by record labels worldwide in 2017, or about USD 4.1 billion.
  • Market trend is that music subscriptions will be given away for free: The arrival of Amazon, Google and Apple in the music subscription market is not good news for Pandora. The signs are that these companies are happy to run their music subscription businesses almost as a marketing expense: Apple is bundling iTunes Radio with iTunes Match (which costs $25 per year – about the same as a Pandora subscription).

    If this becomes a trend then this either means that Pandora will be forced to to invest in developing features to compete with iTunes Match (without being able to generate any incremental revenue from these new features) or it means ceding a significant portion of its addressable market to these new heavyweight players (which it clearly cannot seriously contemplate – music subscriptions being a volume play).
  • Pandora’s listening hours have flicked from online to mobile, but mobile is harder to monetise with ads: After 6 years of steady growth, mobile now accounts for 76% Pandora’s total listening hours. 

    This is a problem because mobile is harder to monetise using ads than online because:
    (i) the small physical size of the mobile device means that there are fewer opportunities for serving ads (think in terms of the number of available ad units per page); 
    (ii) the mobile ad market is far smaller and less mature than the online ad market, meaning that as Pandora’s ad inventory grows then its ad sales teams will find it harder to sell inventory to advertisers most of whom still view mobile advertising as ‘experimental’; 
    (iii) the mobile advertising market requires competencies in serving personalised and local ads to a greater extent that is presently the case with online advertising and Pandora has yet to develop the technology and expertise to meet these unique requirements.

Stock price

In spite of our negative outlook on the music subscription market as a whole, Pandora’s stock has actually performed remarkably well since it’s IPO in June 2011, and especially over the last 12 months:

As of 11 November 2013, Pandora’s market capitalisation was USD 5.18 Billion. We find this simply staggering. Pandora:

  • Is an 8-year old company that has never made a profit;
  • Is operating a business model that is intrinsically unprofitable, which means that growth in users simply means growth in losses;
  • Operates in a sector – the music subscription services sector – which is fundamentally unprofitable: all music subscription services are losing money;
  • Has not hinted at any change in strategy which will have the effect of dramatically increasing the top line, which is what is needed to drive the business into a profitable operating situation;
  • Will have to overcome some very strong headwinds simply to survive, one of which will be upcoming negotiations with the music industry over royalty rates.

We think that the market has either:

  • Not fully understood just how precarious Pandora’s business really is;
  • Is assuming that the company will monetise its 200 million user base by entering undefined new product and service categories (not music-related);
  • Assumes that the company will be taken off the market by a privately-funded structure that is able to extract USD 5.18 billion in value from the company’s user base, because Pandora’s current business and assets are incapable of doing this job.


Pandora is a really wonderful service, but it is very hard to see anything wonderful about the company’s business model, or even its future prospects if it carries on as it is.

Nevertheless, this doesn’t mean that Pandora cannot find success in another way – as noted above the company could be taken private and the company's present investors might in that case see a tidy return, but let’s not fool ourselves into thinking the company is ever going to be profitable – unless, that is, it changes strategy (and we have some clear views on what those changes might entail...).

But at this point, if we were professional investors (which we are not) then this company seems a strong candidate for a short position or, for the risk adverse, a long-held put option.