Spotify and the Freemium Music System Dynamic Conundrum

Today Spotify (the Latest Internet Service du Jour) announced it would restrict it's free user base from May 1st - via PaidContent (love the irony in them reporting this story):

- Free listening hours will be halved from 20 hours per month to 10 hours.

- Free users can play a track up to five times only.

For new free users, both new limits kick in after six months. For existing free users who joined before last November, the new track limit will take effect from this May 1.

Apparently Spotify hasn't given any reason for the changes in its announcement PaidContent thinks there are three possibilities…

- Because it can: Spotify has proved it can convert one million out of its 10 million users to pay. But, with nine million left over, there’s an ample pool to try converting.

- It’s about cost control: This is not the first time Spotify has limited playing hours. To manage bandwidth during its first-year boom, it temporarily disabled new UK sign-ups and recast its unlimited free tier as the 20-hour Open it has today. A further clamp on play time could be another attempt at controlling bandwidth or royalty outgoings, which are about to grow when it launches in new countries.

- It’s a label stipulation: Though Spotify has signed two majors on its U.S. odyssey, they remain keener than their European counterparts on more pay-oriented music services. Spotify was also recently renewing its European label deals.

Can't see number 3 myself, Ad supported music is an industry staple since Marconi invented the radio. No 2 seems to have merit (but you can't cost cut your way to a 100 million dollar business), so they had better hope No 1 works - as PaidContent says:

Spotify has always said its free service remains a critical premium driver, from which it nevertheless makes “millions” in ad sales, and that any U.S. service would launch with a free element, albeit slightly different. Spotify’s New York-based chief content officer Kenneth Parks says today, somewhat defensively: “Our chief priority is to keep the free service, which is what has made Spotify so popular and successful.”

But, this also tells us that they can't make Ad based free music pay on its own (if they could, why limit Free users access?), one factor being that serving music doesn't get cheaper as volumes rise so scale economics are reduced.

This means they need to get paid users - but paid users come from the 10% conversion of it's Free user base, which puts Spotify smack bang into a fairly interesting piece of Game Theory - to wit, the Freemium Attraction-Retention Model (see chart above).

The "Steady State" - Today - is a fairly attractive Free service that takes X% of all music fans in the market vs its competitors ( it has attracted 1m + users pre it's US launch ) and converts c 10% of its Free (ie Ad supported) base to Paid subscribers. We don't know what it's Churn rate is yet.

Now, if you reduce the attractiveness of the Free deal vs competitors (of which there are many), you are likely to to recruit fewer free users (ie X % reduces), which means fewer new paid users can be converted at 10%, which means slower growth - and it is slower still if you factor in a churn rate (here shown at 20%, defecting to new free services). And the defection rate goes up faster if competitors then reply to this with a better Free (or even Paid) offer (which - I will hypothesize - will happen as newer services, still flush with VC money, willl discount to gain market share).

So the risk is a reduced capture of new entrants leading to a reduced conversion to Pay users and an increasing churn of existing users. This in system dynamics terms is called a negative feedback loop, or in the vernacular a "Vicious Circle" that cycles the business in ever decreasing circles down the plughole into the deadpool (see diagram above).

Of course, playing with those assumptions in the above model can also switch the state to a "Virtuous" Increasing Returns circle, but you then have to believe that one or more of the following must exist:

- There is no reduction in recruitment rate X%, or that access to the US with a larger market pool allows "X%" to drop but a larger market is addressesd, so numbers are still maintained or increased
- The conversion to "For Pay" stays the same (or at least stays high enough to fund the Free users) - ideally they fond a way to bias recruitment towards those likely to trade up
- That "Churn" % of For Pay does not go up due to newer deals
- Thart the labels will sign unique access deals to existing players (eg Spotify) and freeze new entrants out

In many ways you want your long term Free users to churn (or at least not incur costs) of course, hence the slow squeezing of the deal to the Free (down)loaders - but you run 2 further game theory risks, ie:

- Those Freetards join the New New service and enthuse about that, increasing its recruiting and driving X% and conversion to Pay % down and possibly driving existing user Churn % up
- The New New service grows and its buzz and scale allows it to negotiate better deals so when it launches a Pay service then its markedly more attractive (there being no real sustainable competitive advantage in Internet Music services)

Of course, if you do believe Spotify has some sort of major competitive advantage over The New New Others (or no New New others will emerge) you may choose to assume that the % do not shift, and given that Spotify was voted the Best Telegraph Startup just this week, one would like to believe they can pull it off - but it will be fascinating to see how the next 12 months turns out.

Update - good point from @Amanda:

Spotify premium for a month is less than the cost of a new CD. It saves me around £1000 a year. Why don't we talk about that?

My answer - the model essentially assumes those are the people who have already made the tradoff to pay to rent music, ie - in dynamic terms - they are "out the game" unless they decide another service is more attractive and they then churn. Others will calculate that if they must pay, they will pay to own, not rent. The Game Theory issue for Spotify now is that new users of the "pay to rent" bent may go elsewhere if the initial offer from other players is then seen as better.

(Update - 2317 comments on the Spotify blog post on the subject! )