Churn in the era of dynamic retention

Kantar, a survey vendor, has been getting some attention by passing off consumer data as an actual measure of subscribers and suggesting that the music subscriber base actually declined in Q1 2022. It said the same in Q4 2021, but 2021 was a spectacular year for music subscriber growth, with the global base of subscribers growing by 118.8 million in 2021 – the largest ever increase in a single year – to reach 586 million. Of course, it would be obtuse to suggest that all is rosy in the world of digital subscriptions. After all, the attention recession has slowed growth and the actual recession will push up churn rates. But it is wrong to assume that digital subscriptions behave like their traditional counterparts, which is exactly why music subscriptions are well placed to weather the perfect storm of both recessions.

Digital subscriptions are different

Traditional subscriptions (pay-TV, internet, phone etc.) are slow moving, predictable beasts. Consumers are locked into contracts for fixed periods and must pay penalty clauses to exit them early. Which is why, when churn happens in these subscriptions, it is a big deal. It represents a hard break, the end of a subscriber relationship. But digital subscriptions are wired differently:

  • Churn doesn’t necessarily mean churn: Few have contracts, and most are as easy to leave as they are to join. They are built (if not necessarily designed) for hop-on / hop-off behaviour. When someone drops a Netflix subscription, the likelihood is that they will be back in a few months. The same does not apply for traditional subscribers.
  • Digital subscriptions are less critical: Most traditional subscriptions are utilities (phone, broadband etc.). Even a pay-TV subscription is a utility because the TV set may literally stop receiving signal without a subscription. So, cancelling one is a much bigger deal. But digital subscriptions usually just make digital entertainment better (e.g., an extra catalogue of TV shows to watch, music without ads etc.)
  • Many are still getting started: Even though music subscriptions growth is slowing in many markets, large numbers of consumers are still trying out subscriptions for the first time. This means there is always a high turnover of subscribers. Even more so in video and games where new services have come to market.

The last point is perhaps most important. MIDiA’s Q1 consumer data indicates that more people signed up to music subscriptions in the previous year (13%) than cancelled (10%) – both figures are as a share of all consumers that either had or used to have a music subscription.

The takeaway is that music subscriptions are highly fluid at the edges. They resemble a duck in water: elegant and slow moving above the water line, but legs pumping furiously below it. We can see this in Spotify’s reported numbers too. In 2020 Spotify added 25 million subscribers to its tally to reach 180 million. But it actually added twice as many subscribers as that before it also lost 25 million due to churn.

Churn is built into the model

Churn is quite simply part of the equation for music subscriptions. But at risk of sounding too Pollyanna-ish about this, there is no denying that dark clouds are building on the horizon. The cost-of-living crisis is accelerating, inflation and interest rates are going up, and wages are steadfast. As MIDiA’s recession data shows, around a fifth of music subscribers would consider cancelling their subscriptions if their everyday costs spiralled. A subscriber slowdown may indeed come. Those that do cancel should not be considered ‘lost’ but instead as taking a break. They will be there, ready to dive back in as soon as they can. 

DSPs will need to think in terms of what MIDiA calls dynamic retention. Instead of being focused on having a subscriber for all 12 months of a year, understand that in the coming economic climate, subscribers will likely require more flexibility. So, think instead of how many subscriber months can be had from that subscriber over a 12-month period, regardless of whether they are consecutive or not. It is certainly a shift in mindset, but this kind of pragmatic and flexible thinking will be crucial for navigating the times ahead.

Making Streaming Work (Fixing Playlists and Churn)

UPDATED 28/3/13 (see sections labelled ‘UPDATED’

During my SXSW panel I presented a slide that showed the distribution of paid, active free, and inactive users of the two big streaming services Spotify and Deezer based upon the latest data for both services.  What the numbers show is that inactive users is a big problem for streaming services, which in actual fact means that churn is a bid problem for streaming services.  (Something I discussed last week).

Paul Resnikoff at Digital Music News and Stuart Dredge at MusicAlly have since written pieces about the data and something of a debate is emerging.  But the important point is not whether Deezer has more inactive users than Spotify, but that streaming services as a whole have a problem with churn.

To illustrate the fact that this is not a Deezer problem I have created a new chart (below) that uses the latest available official numbers for all three types of users from both services.  The most recent total user numbers for Spotify are Facebook app users and are therefore not official stats.  The most recent official Spotify data for all three metrics is from year-end 2011 – when these numbers were filed in a company report – and for Deezer this means early 2013 when the numbers were quoted in the press.

UPDATED: Note that Spotify only ever mentions its total registered user number in company reports while Deezer have quoted theirs more frequently. So the Deezer numbers below are a more accurate representation of the current scenario whereas Spotify’s user base dynamics have changed markedly since end-2011.  (Whether that translates into more or fewer inactive users we’ll have to wait and see.)

deezer and spotify

Retention is a Freemium Issue not a Spotify or Deezer Issue

What is clear is that both services have essentially the same distribution of users, with the vast majority of both services’ installed bases being inactive users.  If you spread Spotify’s 2011 numbers over the course of the year, from the end of 2010 base numbers, this translates into Spotify acquiring 1.9 million new users every month but only keeping hold of 200,000 of them.

Again, this is not a Spotify problem, it is a fundamental issue with the freemium music model: many more people decide its not for them than continue using the service.  Over time this effect will soften, as people become more familiar with the idea of on-demand streaming.  But it will always be a key part of the mix for both free and paid users.

UPDATED: It is also not even just a music service issue.  As I discussed in a previous blog post about Facebook, social networks like Google+ and Twitter also have a big issue with inactive users, as this chart illustrates. In fact only a quarter of Google+ users are active, as are less than half of Twitter users.  As Daniel Ek correctly identified on Twitter, this is a problem that affects all businesses that have a free tier that requires registration.

Currently Deezer and Spotify are in growth stage and are more focused on acquisition than retention, but sooner or later they’re going to have to recalibrate their metrics if they want to move towards sustainable financial models.  It can be done, as Rhapsody shows us, but it is not an easy task, and it also doesn’t leave a lot of spare cash in the kitty for aggressive growth.

Any established subscription business – such as a cable or satellite TV provider – will tell you that managing churn is the overriding strategic objective.  Any subscription service – especially a nice-to-have like music – is going to be vulnerable to churn.  But this does not mean that the music subscription business is fundamentally flawed, rather that the industry needs to start thinking in terms of a much more fluid movement of users than was ever the case for downloads.  In the download model Apple locked in its customers with devices.  Streaming services have no such asset – at least not yet.

Playlists Belong to Users not Services

With time, clear blue water will emerge between the value proposition of streaming services, and this should be considered not just as a loyalty driver, but also as reason for people to swap and change subscription services just as people swap and change cars.  And for that to happen streaming services need to stop thinking about users’ playlists and libraries as the property of the streaming service to be used as velvet handcuffs and instead as the transferable property of the user, and by extension, the communal property of the marketplace.

Locking music consumers into devices sort of made sense for companies like Apple that were largely using music to sell hardware.  But for companies like Deezer and Spotify that are just in the business of selling music – or at least access to it  – there is no such justification.  The subscription market is only just getting going and there is far too much green-field opportunity for services to get bogged down in internecine conflict.  As MusicAlly’s Dredge correctly identifies, opening up Playlists could prove to be crucial to the long term validity of streaming and subscriptions (and Tomahawk is a great first step). But to really work, streaming services need to stop thinking about Playlists as their property and instead as the property of their users.  That’s when services like Tomahawk could come into their own and it is when mainstream music fans will view streaming services with less scepticism.   In the words of ShareMyPlaylists: Long Live The Playlist!

Why Losing Free Customers is a Good Thing for Spotify’s Business Model

In my Future Music Forum keynote last week I discussed some Spotify metrics which were picked up by Paid Content and have stirred up a bit of a debate.  Here is a little more context to those numbers.

The headline statistic is that in 2011 Spotify had to acquire approximately 1.8 million users per month to retain just 400,000 a month (i.e. ‘losing’ 1.4 million a month), resulting in a total monthly churn rate of approximately 20%.  These estimates are based upon the following reported numbers:

  • Spotify’s end of year accounts for 2011 reported a total of 32.8 million registered users.
  • In December 2011 Spotify reported 10 million active users on its developer blog.
  • In March 2011 Spotify reported 1 million paying subscribers, representing 15% of active users, which put the active user count at 6.7 million.
  • In September 2010 Spotify held a press event to announce 10 million registered users.

The headline numbers give a ‘gap’ of 22.8 million between registered users and active users at the end of 2011.  Using all of the reported numbers and applying flat rate growth assumptions for intervening months we can calculate the total number of active and registered user gains throughout calendar year 2011 (see figure 1).  All of which gives approximately 1.8 million new registered users per month but only 400,000 active users per month.

Figure 1

Now of course there will be monthly and seasonal variations in those numbers so the exact count will be different for each calendar month.  Also many of those 1.4 million new monthly inactive users (i.e. the gap between new registered and new active) may well become active later in the year.  But the headline trend remains that Spotify has to gain a lot more users than it holds onto (or at least did in 2011 – though I would expect similar metrics to apply in 2012).

Losing Low-Value Free Users Actually Helps Spotify’s Business Model

None of this is necessarily a reflection of a flawed business model for Spotify.  In fact, in my view, it reflects positively.  Let me explain.  Spotify’s business is all about selling premium subscriptions.  That’s where the money is for Spotify, labels, publishers and artists alike. The free tier of its business is simply a marketing funnel.  Ultimately it doesn’t actually matter that much how many of those free users stay on board as free users, what matters is how many convert to paid.  In fact, it benefits Spotify if those users who have no intention of paying churn out early on from the free service as it means less cost to Spotify’s bottom line.  As challenging a path towards profitability as Spotify may find itself on, it would be a dramatically more difficult road if all of those 32.8 million users were active.  So Spotify’s business model and margins actually benefit from the majority of those new free users churning out of the service early, allowing Spotify to focus on migrating the remaining engaged free users to paid.

Figure 2

Free Churn Does However Raise Questions About the Wider Streaming Market

All in all Spotify has brought a huge amount of value to the digital music market and has achieved many credit-worthy milestones (see figure 2).  But as much sense as the free-user-leakage makes sense to Spotify’s business model, it does raise challenging questions about the streaming model more broadly.

For so many users (two thirds of Spotify’s 2011 total) to effectively say “no to free” indicates that streaming audio, even when free, does not resonate strongly enough with mass market music fans.  There are multiple potential reasons that Spotify free users churn out, such as: usage caps, advertising, being PC only, not being able to burn to CD, even just being a stream rather than a download.  Many of those can be fixed with a 9.99 subscription, but the simple fact is that most consumers do not spend that kind of money on music.  9.99 is actually the average monthly spend of the top 20% of music buyers. So it is a price point for the aficionados not the mainstream, which means that most consumers will never get a proper taste of the ‘complete’ streaming audio experience.  Which is why I continue to argue strongly that subsidized subscriptions and cheaper price points are the crucial routes to the mainstream music fan that need pursuing with haste.

Spotify, Rhapsody, Deezer, rDio etc are all doing a great job of trying to take premium subscriptions to the masses, but until they can work out a way to get cost-to-consumer price points down, the addressable audience remains a subset of that top 20% of music buyers.

The Elephant in the Room

And all of their cases are challenged further by an uneven playing field.  While all those music services have to charge for mobile access and have some gaps in their catalogues, YouTube provides unlimited access, on all mobile devices, with the world’s largest music catalogue, with video, for absolutely no cost at all to the consumer.  As far as streaming goes, there is one rule for YouTube, and another for the rest.  Until that anomaly is fixed, the rest will be swimming against the tide.