Note: This Analysis was conducted prior to the release of Spotify’s Q1 Earnings Results.
Spotify is an interesting company and market leader in a recently resurrected and growing industry.
The stock is overvalued using traditional DCF valuation, as well as a unit economic based subscriber Valuation.
Using sensitivity analysis, the greatest lever the company has is Paid user growth. more users increase network effects which benefit Spotify's suppliers, creating a positive feedback that further benefits users.
Because of their positive LTV to CAC ratio, they should continue to invest in user growth to widen their moat.
Spotify is an incredibly compelling company, having helped raise the music industry from the ashes after the internet enabled widespread piracy and unlicensed forms of consumption, dramatically reducing revenues. I am a Spotify paid subscriber and love the service, so my hope was that the various valuation methods would point to an undervalued security that I could be happy to scoop up.
What the analysis suggests, however, is that there are significant growth assumptions priced into the stock that I have a hard time agreeing with, or in which I lack enough conviction to warrant investing. One unintended outcome was a reinforcement of the idea that traditional DCF analysis is not appropriate for some companies, Spotify included. As Bill Gurley argues, it is difficult to predict with any accuracy what the long term cash flows will be for a company that is relatively young or one with a new business model, and I’d argue Spotify qualifies as both. My DCF analysis values the company well below the current market price, but for reasons I just mentioned I have some concerns about the accuracy of the output.
As an alternative I valued the company using a subscriber model, which calculates the unit economics at the user level and then arrives at a valuation by assuming some trajectory of user growth. Using this method I came to a valuation higher than the DCF calc, but still below the current market price.
Lastly, I approached the valuation analysis in reverse, adjusting my assumptions up or down 25% to determine what levers Spotify has and which of those will positively impact the business the most when pulled.
My Excel model is available here.
What Does spotify actually do?
resurrecting the music industry
From 1999 to 2014, the recorded music industry faced continuous headwinds in the form of a movement from physical formats to digital, with revenues declining 40% over that period. In 1999-2002, illegal digital distribution increased in large part due to the rise of Napster and later the likes of Kazaa, Limewire, and YouTube. Music consumption didn’t decline, but revenue did because users could now access the content for free (illegally) and in an unbundled way (download one song as opposed to the whole album).
Spotify was “founded” in 2006 but launched its first product iteration in 2008 and mobile app in 2009, a fortuitous time to enter the space. Crackdowns of illegal platforms left consumers with market options that either had poor UX, or were unsafe from both legal (piracy = illegal) and digital security (downloads came with the risk of viruses) perspectives.
Incumbents Pandora and Soundcloud were already successfully streaming music to customers, but had different strategies. Pandora went with a fully (mostly) ad supported model and Soundcloud wasn’t making money, but Spotify was the first to offer a paid subscription model, allowing users to access any music in any quantity for one set price.
Fast forward to today - Spotify makes money two ways, by advertising revenue generated from ads displayed (and played) to users on the free tier, and by charging users a fee to use the premium service. On the flip side, Spotify’s main cost is the fees it pays to license out the content from music labels (mainly Universal, Sony, Warner, and Merlin), which equates to around 80% of revenues.
Historical Performance Overview
Spotify generates the majority of its revenue from subscriptions (90% in 2017) with the remainder coming from ad revenues (10% in 2017). Revenues grew at an impressive 53% CAGR from 2013 to 2017, although YoY growth dropped to 40% in 2017 in USD and has declined each year since 2015 in EUR.
Cost of Revenue
Cost of revenue is made up primarily from royalty and distribution costs paid to music labels and publishers in exchange for the right to stream content to users. As a result of new licensing agreements entered into in 2017, Spotify was able to lower their cost of revenue percentage to just under 80%.
Outside of Cost of Revenue, operating costs as a percentage of revenue have been increasing since 2015:
Monthly Active Users have been steadily increasing for both paid and ad supported tiers.
With an increase in users, total aggregate content hours consumed would be expected to increase, but Content hours per user is also increasing, meaning that on average, each user is increasing time spent listening.
Churn has been steadily declining since Q1 16, which the company attributes to increased engagement from new features as well as adoption of higher retention products like their Family Plan. The company also states that “Based on historical data from the beginning of 2015 to date, approximately 40% of the Premium Subscribers who churned rejoined within three months, approximately 45% rejoined within six months, and 50% rejoined within 12 months” which means that true churn is lower than the reported 5%-7%.
I assume YoY growth slows over the next 10 years to a terminal run rate equal to the long term risk free rate. I assume margins improve, flipping to positive by year 5 and trend towards the industry average by year 10. Using a cost of capital of 9.24%, I value the equity of the business at $94 per share, which at its current price of $159, SPOT is overvalued based upon discounted cash flow analysis.
Revenue Growth: Revenue has grown at a CAGR of 53% from 2013 - 2017 but as expected with most companies that achieve scale, YoY growth has been declining for the past two consecutive years with 2017 at 39% EUR and 41% USD (fx rates are the cause of the difference). I assume total growth rates continue to decline, less quickly over the next 5 years and then more rapidly as competitor offerings catch up and market saturation increases.
Are these growth rates reasonable? According to Goldman Sachs estimates, by 2030 global streaming revenues will total $34 billion, with paid subscriptions accounting for $28 billion and ad revenues the other $6 billion. In the prospectus, Spotify states that their “global streaming market share was approximately 42% in 2016 as determined by revenue” which would put the global streaming market at $7.8B - $7.9B, larger than $4.7B estimated by IFPI.
Assuming IFPI and Goldman are accurate, then a linear interpolation between those two datasets would put total global streaming revenues 10 years from the base year at roughly $27 - $28 billion by 2027. At my growth rates of 33% next year trending lower to 18% by year 5, and then steadily lower going forward, that puts total Revenues for Spotify at $23.7 billion by 2027, which would equate to ~85% of all global streaming revenues.
Given that in 2016 Spotify’s global market share was around 42% (according to the company), combined with the traction that Apple music has seen of late and the continued success of the likes of Tencent Music and Amazon, 85% global market share seems highly improbable.
How do I reconcile this? I think the IFPI data is understated (both MIDiA and Spotify’s estimates of 2016-2017 streaming revenues are larger than IFPI) and the Goldman Sachs estimates are conservative.
Over the past 20 years the music industry has felt the pain of disruption and evolution from physical sales to the format that will dominate for the foreseeable future in streaming. Until recently music was consumed through a mixture of formats that were all lacking in one way or another - none had the complete package of fully comprehensive catalogs of licensed content (Soundcloud and YouTube were mostly unlicensed), a high quality UI (none were great), and portability via the internet (physical formats were limited by what you could carry with you).
Because of streaming, any user with a connected smartphone or computer can access nearly any piece of recorded music via an enjoyable UI for relatively cheap. And yet as of today, global streaming subscribers only total less than 200 million, and while the number of global MAUs of streaming services (which include users on the ad supported tiers) is larger, it is still a fraction of the 3.6 billion internet users globally. Consider that even after several years of strong growth due to streaming, global music revenues are only ~75% of what they were 15 years ago.
Because of this, I believe there is a much greater chance that global music revenues will be larger than Goldman’s estimates in 10 years as opposed to smaller.
Operating Margin: Operating Margin has been negative since 2013, but improving in 2017 due to restructured royalty and distribution agreements. S&M, G&A, and R&D are all increasing as a percentage of revenue, a step in the wrong direction if the company has profitability as a goal.
Improvements in cost of revenue is the largest lever but the one over which Spotify likely has the least control. Universal Music Group, Sony Music Entertainment, Warner Music Group, and Music and Entertainment Rights Licensing Independent Network (“Merlin”) account for more than 2/3rds of global market share and platforms like Spotify are dependent upon those rights, for without them there is no product to offer users. The platform with the largest catalog is in the best position to win over potential new users, and because of this need, the record labels have the leverage in the negotiations. As streaming becomes increasingly ubiquitous and new competitors enter the market, this leverage will only increase.
I assume the company successfully moves to profitability in the next 5 years with a combination of improvements in SG&A spend and potentially better licensing deals. By year 10, I assume the company operates at an industry average operating margin of just over 13%.
Other assumptions: I assume the company will continue to reinvest cash generated into the business at high levels in the early years, and lower levels in later years as the company matures. I have converted R&D operating expenses to CapEx from the view that research costs are intended to produce future growth and thus are more like an asset that should be amortized. Because the company has no debt, no operating leases that could be considered long term debt, and no preferred equity, the cost of capital is equal to that of the equity, which I estimate at 9.24%.
Subscriber Model Valuation
The subscriber model has many more assumptions than the DCF (I’m not sure if that makes it more or less accurate, but more on that another time), but in a nutshell I am assuming churn continues to decline slightly, customer acquisition continues to slow YoY, content costs decline but level out at ~70%, and both ad revenues and corporate (non content) costs grow but at a declining pace.
I arrive at a value per share of $115, which is above the simple DCF but still below the current market price of ~$159.
Lifetime duration of average customer: 15 yrs. At a churn rate of 5 per year, that would suggest a tenure of 20 yrs (1/churn), but because there are very few consumer technology services (especially in the content space) that last more than 10-15 yrs, I went with 15yrs.
Revenue per Subscriber (and growth of): Equal to 2017 avg of ~$78, growing at the rate of inflation. The part I feel the least comfortable about here is the growth. Increasingly evidence points to improvements in technology over the past couple decades being deflationary (increasing output/value at constant price), and a reduction in cost-push inflationary pressure due to the removal of distribution costs across the music industry because of the internet doesn’t help.
Content Costs per Subscriber: Declines from current levels towards 71% by year 10 as licensing deals with record labels improve. This one is hard to predict. On one hand, as Spotify grows and increases market share, so does its negotiating leverage to reduce costs, which it has started to prove it can do. On the other hand, as competition increases, Spotify’s leverage decreases because the record labels can pull their content and sell to the other platforms, to Spotify’s detriment. As market leader of the streaming category, which will surely be the record label’s primary source of revenue in the future, I am willing to bet there are more concessions to come on the part of the labels, albeit not many.
Customer Acquisition and Churn: Acquisition steadily decreases from existing levels to 10% by year 10+. Churn declines slightly in the near term as Spotify continues to improve the customer experience but increases over time as competition increases.
Strong customer acquisition has accompanied quarters in which Spotify offered heavily discounted trials, but acquisition totals outside of those quarters was much lower.
Inactive subscribers (paid subscribers plus ad supported users minus MAUs) made up a small percentage of the total in 2017 but grew 4x from the beginning of 2016, a trend worth watching.
The long term acquisition growth at year 10 is one of the main reasons that the subscriber model produces as valuation larger than the DCF, but in a universe of 3.6 billion global internet users, I don’t think 10%+ growth is unrealistic and in fact, the lower long term growth in the DCF is more to prevent crazy permanent growth after the terminal year.
Advertising Revenues: Growth is fairly strong, averaging nearly 30% per year in the first 5 years and declining to 10% by year 10. The longer term shift towards programmatic channels should continue and audio formats should benefit.
Corporate Costs: Growth declines from the recent 50%+ YoY levels to a more reasonable 10% by year 5 as the company focuses on profitability
Reverse DCF Analysis
Using my assumptions as baselines and adjusting various inputs while holding others constant derives what levers are available to Spotify and which will produce the most outsized results.
Lifetime duration of average customer
Adjusting up 25% to 19 yrs changes valuation by: +18%
Content Costs Per Subscriber
Assumption: Current level of 78%, declining to 4% by year 5 and long term level of 71% by year 10
Adjusting costs down 25% to 53% by year 10 changes valuation by +101%
Assumption: Declines to 4% in the next couple years then increases as competition increases, hitting a long term level of 5.5% in line with recent years
Adjusting churn down 25%to a long term level of 4.1% changes valuation by: +23%
Assumption: Declining from 38% in year 1 to just under 10% by year 11
Adjusting up 25% to arrive at 12.5% growth by year 10 (thus increasing all growth rates in between year 1 and 10) changes valuation by: 107%
Ad Revenue Growth
Assumption: 35% initial YoY growth declines to 10% by year 10
Adjusting growth up 25% changes valuation by: +5%
Assumption: Costs continue to grow at 20%-40% over the next several years and hit a long term rate of 5% by year 10
Adjusting costs down 25% changes valuation by 26%
As suspected, the two largest levers to improve the valuation are content costs and growth rate of acquired users - the former because of the impact on free cash flow, the latter because of the effects of compounding. One of these levers Spotify has little to no control over (content costs), the other is well within their gift to positively impact (user growth).
As a platform business, Spotify has built a competitive advantage by leveraging network effects - Spotify generates incremental revenue for the record labels and artists with each additional revenue generating user, thus strengthening their negotiating position to both reduce costs and expand their content library, which in turn they can use to lower prices while offering an increasingly wider selection for consumers. Because of this, and their positive LTV to CAC, they should rightfully run cash flow negative, plowing cash back into user growth to widen their moat.
I built my valuation models utilizing the methodologies of Aswath Damodaran and for this piece used the following resources:
Goodwater Capital Spotify Thesis (http://www.goodwatercap.com/thesis/understanding-spotify)
IFPI State of the Industry Report
Goldman Sachs Research