First, Spotify's cost of sales looks too high. Spotify pays out about 70% of its revenue to rights holders, which should be well known by now. PrivCo has Spotify's cost of sales at 97.7% in 2011.
Yes, Spotify could have paid out more for content costs in 2011. But at SXSW in March chief content officer Ken Parks said the company pays out "in the neighborhood of 65% to 70% back to rights holders." That's a lot lower than what PrivCo thinks Spotify is paying rights holders. And it's hard to believe Spotify would sign deals that give away every dollar it earns.
Cost of sales is the heart of the debate about Spotify's business model: PrivCo interprets this 97.7% gross margin to mean Spotify will always hand over nearly every incremental dollar to content owners. This is wrong. I'll explain why below.
The second thing that stands out is a single deduction -- personnel costs -- under gross profit. This is related to the first problem: PrivCo has included in cost of sales many expenses that do not belong there. Costs of sales are items that are directly attributable to the revenue being generated.
A good example of proper accounting is Pandora's income statement. Pandora, a publicly traded company with audited financial statements that adhere to Generally Accepted Accounting Principles, breaks out content acquisition costs, product development costs, general and administrative costs, marketing and sales costs and cost of revenue (which is probably a catchall for items such as leases and equipment that help the company generate sales).
Like Spotify, Pandora's content acquisition costs increase linearly. As people listen to more music, its royalties increase. Similarly, as Spotify generates more revenue, it pays more to content owners.
A better way to assess Spotify's business model is to put the right numbers in the right places on the income statement.
First, Spotify's gross margin should be around 75% based mostly on its content acquisition costs. Spotify pays, or will pay in the future, around 70% of revenues to rights holders (I made it 72% to be conservative and to throw in a couple percentage points for performance royalties just in case they were being left out). Then I added 5% of revenue (it may need to be more) for the costs of the business directly related to running the service.
Next, the costs directly related to revenues should be separated from salaries, administrative and other items. PrivCo doesn't break out these numbers, but we know from Spotify's business model the company has significant sales and marketing expenses as well as administrative functions in over a dozen countries. Since we know these costs exist but don't know the details, I have condensed them into "G&A and Other Costs."
The bottom line is the same under both accounting methods, but under the new accounting the business model doesn't look nearly as bad. Most importantly, we see that Spotify doesn't hand over to rights owners anything close to $1 for every additional $1 in revenue it generates. As a result, its gross profit increased from $5.6 million to $31.8 million.
Nevertheless, the point should be clear: Spotify has about a 25% gross margin to play with after it pays rights holders. As the company gets bigger, that 25% gets bigger. If Spotify can grow to, say, $2 billion in revenue per year, it will have $500 million in gross revenue from which to pay its expenses.
That would be a lofty goal, but there's nothing about this company that indicates it is aiming for mediocrity.
Update: PrivCo CEO Sam Hamadeh tells me Spotify's financials - he says he has financial statements audited by Ernst & Young in his hands - define cost of sales as content acquisition costs plus credit card processing fees. That's it. So the cost of paying rights holders and processing payments took 98 cents of every dollar Spotify earned in 2011, Hamadeh says my assumptions were wrong about the way PrivCo calculated cost of sales.
The pieces don't fitting together too well. Parks says Spotify pays rights holders 65% to 70% of revenue. Anecdotal evidence suggests subscription services typically pay around 70% of revenue to rights holders. Case in point: Napster had a cost of revenue of 71.5% in 2008, according to its 2008 10-K http://www.annualreports.com/HostedData/AnnualReports/PDFArchive/naps2007.pdf. Yet PrivCo claims to have audited financial statements that say content acquisition costs are roughly 96% of revenue (with about 3% of revenue for credit card processing, says Hamadeh).
The wild card here would have to be free listening. Spotify's freemium business model means it incurs royalties for free listening and must generate advertising revenue to pay rights owners. A pure subscription model leads to more predictable expenses.
Whether or not Spotify's business model is sustainable depends on its ability to achieve typical gross margins of a subscription service - about 30% give or take. It does not need to hit 30% immediately, but it should aim for that target in the near future.
But will Spotify die a fiery death? It's hard to imagine right now. Labels and publishers need it to succeed and will have to accommodate its needs. Put simply, Spotify could be too big to fail.