What can ELT learn from Spotify?


There are few things that we at ELTjam enjoy more than mining Spotify’s immensely rich seam of content in an on-going quest to discover, share and devour copious amounts of well-crafted music. Spotify is on our desktops, laptops, phones and tablets; synced, sorted and ready to respond at any moment to a sudden musical impulse.

For those of you not familiar with Spotify, I’ll do my best to keep my gushing to a minimum in my description of it. Spotify is a web and mobile app that enables the user to search through millions of tracks from an enormous catalogue of artists. The user is then able to stream the music through their device, rather than purchasing the music and downloading it for keeps. Think iTunes’ music archives meets YouTube’s content delivery model.

Costs? It’s available for free on both mobile and web app as long as you don’t mind having your listening interrupted at regular intervals by adverts. The premium account gives you unlimited, uninterrupted listening privileges as well as the ability to listen to the content offline. This means that you download your chosen songs to your device-of-choice for access anywhere, anytime. Again, you won’t have purchased those tracks in the same way you would from iTunes; you don’t own them. You do own the right to access them … for as long as you pay your subscription at least.

There. Spotify described and virtues modestly extolled.

This post is interested in what ELT as an industry might be able to glean from Spotify’s MO and, interestingly (given previous posts and discussion taking place on the blog), we find ourselves talking about royalties. Spotify has been taking a lot of criticism in recent years for the model it employs in calculating an artist’s royalties. This is fairly understandable; there’s no real purchasing transaction taking place, as mentioned above. Intellectual property is not being traded, rather leased. The company has made public its formula for arriving at what an artist sees on their royalty cheques, and it’s far from simply being calculated on a simple ‘per-play’ basis. Instead, an artist’s royalties are dependent on the following variables:  the country in which people are streaming the artist’s music, the number of paid users as a percentage of total users in the territory, the pricing/currency value in the territory and an artist’s royalty rate with their label/publisher.

The result is that an artist can expect to earn somewhere between $0.006 and $0.0084 each time their song is played. Not a great deal, admittedly, and considerably less than Noddy Holder would expect to see arriving in his bank account every quarter. Radiohead’s Thom York publicly railed against Spotify’s apparent neglect of the artist in its business model by referring to their payment policies as ‘the last desperate fart of a dying corpse’. But if you take into account that a certain strata of artists on the platform is getting plays in the tens, if not hundreds of millions, then those paltry-looking payouts take on a whole new lustre. In fact, the company has paid out over $1 billion in royalties, $500 million of which were in 2013 alone.

So what can we learn from this? In a recent interview with Wired magazine Spotify’s cofounder and CEO, Daniel Ek, described his company’s mission in a way that seemed to resonate with some of the challenges and opportunities that are presenting themselves in the ELT industry. Firstly, Ek’s creation of Spotify was a continuation and evolution of how users were behaving in terms of their contact with the media they wanted to enjoy. By Spotify’s initial launch in 2006 file sharing platform Napster had been legally challenged and shutdown, and the iTunes store was demonstrating that music lovers were prepared to forego the physical object of the album; they just cared about the content. (Interestingly, before its courtroom demise, Napster was recognised as contributing to the chart success of Radiohead’s Kid A album due to the fact that several tracks from the album were leaked months in advance of it official release.)

Spotify seized the opportunity to cultivate that shift in music consumption by convincing record labels to sell access to their content rather than the shiny discs and inlays that they were producing in limited numbers to be displayed on shelves. This model has undoubtedly struck a chord with music consumers globally, as evidenced by the 24 million-strong usership and the fact that the company is experiencing annual growth in excess of 100%. As the user-based continues to grow, so does the artist royalty at the end of the formula.

Of particular interest is the fact that record labels have been gradually responding to this cultural shift, as Ek says in the Wired interview, ‘They are just waking up to what’s happening, but as with any large corporation they face the innovator’s dilemma’. Some labels, as Ek observes, are beginning to be more proactive and are creating their own Spotify playlists that include music from other sources rather than just their own in order to generate more interest.

Rather than focussing on how people consume music, Ek wants Spotify’s lasting contribution to be to change how artists create. With millions of users worldwide providing a living and engaged canvas for artists to experiment on and get feedback from, there is going to be a dramatic shift in how music is created and developed. Artists would be able to trial new material, share work in progress or release various iterations of album content to gauge the cohesion of their creative concept. Similarly, users would feel more connected and invested in the music they enjoy and the artists they follow.

How could this be applied in the ELT industry? Could content creators release iterative versions of their content in exchange for immediate and globally dispersed feedback? Learners would be able to contribute to the ‘sound’ of the learning content being created and shared by becoming vocal and vital partners in the transaction. But, as demonstrated above, with royalties being so dependent on the sheer numbers of subscribers, would ELT content creators ever see this model being more profitable than the traditional model, or will their hard work be sold for a song?

21 thoughts on “What can ELT learn from Spotify?”

  1. “Could content creators release iterative versions of their content in exchange for immediate and globally dispersed feedback? “

    Theoretically, yes. But this assumes that the kind of content that Spotify distributes (individual songs and/or album tracks) is commensurate with the kind of content that comprises a course of study. It is not. Educational content is neither uniform nor easily quantifiable: it can range across a whole variety of texts and tasks, both receptive and productive, some individual, some collaborative, some closed (or convergent), some open (or divergent), and of varying lengths and degrees of complexity, from one-minute warmers to one-hour simulations. Apart from the problem of how you determine what each type of activity is worth, royalty-wise, there is the problem that, unlike pieces of music, which are self-standing, educational tasks only optimally work when they are part of a cohesive sequence. Randomly downloading coursebook-type tasks, with no coursebook-like organization, would be a very different – and ultimately less motivating – experience than randomly sampling, say, hit songs from the 1980s.

    One solution might be to package the material into complete units, within which the individual activities are logically sequenced, but these would be more expensive and potentially less attractive to punters, and nor does this get around the lack of coherence issue: why, and in what order, would you choose between Units X, Y and Z?

    In the end, we come up against the same old problem: an educational program, least of all a language-learning one (whose objective is to develop skills rather than to transmit knowledge), cannot readily be broken down into the sort of discrete units that can be delivered and costed in the same way that songs (or hamburgers) can. A business model that attempts to do this is not likely to be educationally sound. (Not that that will cause your average edtech entrepreneur any lack of sleep).

    • Hello Scott,

      I’m replying to the following (from your comment above) with a few questions of my own:

      “unlike pieces of music, which are self-standing, educational tasks only optimally work when they are part of a cohesive sequence. Randomly downloading coursebook-type tasks, with no coursebook-like organization, would be a very different – and ultimately less motivating – experience than randomly sampling, say, hit songs from the 1980s.”

      1. Would you still consider a set of activities, downloaded individually for a particular purpose, to have “a cohesive sequence?” I’m talking about, for example, downloading a few activities to form a “skimming playlist,” or “a past-tense drilling mixtape” – a bit like Spotify’s “workout playlists” or such like.
      2. Would the role of an experienced teacher-as-selector, when trained carefully and applied reasonably, take away the “randomness” you mentioned and lead towards the more cohesive state of affairs here?
      3. Historically speaking, has there always been that cohesion in language learning materials – or did the format of the coursebook just bring about a more connected-up way of thinking about content (a bit like the LP bringing about the very idea of an “album” to music, I might add)?
      4. If that emphasis on whole, rounded-off sets of activities were to be taken away – by publishers’ decisions, by users’ changing preferences, by combination of whatever factors – what demands would that put in place instead on content creators and consumers?

      There’s more on my mind, but that will do for now. Thank you for the comment!

      • Hi Wiktor… sorry, but I didn’t notice your comment until now, and it deserves an answer – not silence! (Not that you may have had the patience to wait for an answer!)

        Regarding the ‘cohesive principle’ – of course, there’s nothing to say that the user’s criteria for bundling activities are any less valid than, say, an expert course designer’s – and may in fact be more valid, being motivated by the user’s perception of his7her own needs etc. So, if User A wants to collate a ‘playlist’ of preposition exercises, while User B would prefer a set of ‘deducing words from context’ tasks, all power to them!

        But I’m not sure all learners do have the self-awareness to make these kinds of choices. Of course, some kind of diagnostic software might help point them in the right direction (‘You have problems with phrasal verbs: go here!’). In answer to your second question – yes, a ‘human guide’ might be the ideal option, until at least algorithms are sophisticated enough to diagnose individual needs and to plot appropriate strategies and trajectories.

        Your third question (re the coursebook sequence) – again, there’s nothing written-in-stone to suggest that coursebooks have got it right either – in fact, I have expended quite a lot of energy arguing the contrary! And (your fourth question) – I think one demand that ‘spotification’ might impose on content writers might be to include suggested links in their materials, so that, in the absence of a master plan, the learner can follow a notional path through the maze. Hope this helps!

  2. Scott, I agree with your point here as long as we restrict the kind of content to very granular, “self-standing” tasks that are delivered in a business model where they are “randomly” downloaded by punters just like songs.

    But I think restricting Tim’s question to these definitions veers dangerously close to a straw man argument, and I think there are Spotify-type models that will work in ELT, pedagogically as well as financially for authors. Of course I’m about to describe the business model we developed with English360, so I’m really interested in whether you find this “educationally sound” or not…we believe it is anyway.

    So basically we’re an open web platform with LMS stuff inside, used to design and deliver blended learning and flipped classroom programs. We sell to language schools, universities, and independent teachers. Like Spotify, we have a monthly per student subscription price, and like Spotify, there is no limit to the amount of content that can be accessed – you could do 2 pages or all 90 courses and resources in the system.

    The content in the system comes from our publishing partners, and we define “publishing partner” as anyone who authors content in the platform that passes our quality control. Larger partners include Cambridge, HarperCollins, Richmond, and Garnet. Some schools author their own content as choose to give open access to other teachers, and we have various independent authors and teachers who self-publish with English360, using the authoring tools in the system. Everyone is paid using exactly the same mechanism: we take a percentage of our net sales and put this percentage into a “content pool” and then this amount is divided up among all content suppliers by page view (e.g. if you are an independent author, and your content garners 3.54% of total page views in a month, you get 3.54% of the content pool). So it’s a level playing field for everyone. To answer your question about how to determine how much each activity is worth, we let teacher usage decide that amount.

    The content can be used as a whole, in the complete course as published, but it is also “disaggregatable” in that teachers can pick and choose pages, sections, units, or even individual tasks from any source and build personalized courses out of that, using the content in the system and (usually) adding some content of their own to this curated remix course. Basically this is exactly what most BE or ESP teachers do with a photocopier, and with the same intent: personalization. So as with Spotify, you can listen to an entire CD (the whole course), or just individual tracks (pages or tasks) in a customized playlist. In fact some call this approach “playlist course design”.

    So for your question about coherence, our model lets the teacher or school provide that. And the results will usually not be as perfectly consistent and coherent as a shiny pristine course from Oxbridge University Press, but because the course is customized for each class or each student, it’s more relevant, meaningful and engaging…values I’d prioritize over the coherence of a one-size-fits-all global coursebook.

    Thoughts? Have I morphed away from 20 years of teaching into the “average edtech entrepreneur” that doesn’t care if the product is educationally sound?

    • So for your question about coherence, … Thoughts?


      I appreciate you were addressing this question to Scott, but if I may …?

      our model lets the teacher or school provide that [coherence]. And the results will usually not be as perfectly consistent and coherent as a shiny pristine course from Oxbridge University Press, but because the course is customized for each class or each student, it’s more relevant, meaningful and engaging

      I can see that E360 has a great deal to offer, though a comment I’d make (and similar to the one made below) is that I wonder if it might not be a) ahead of its time and b) a model that might be more attractive to independent learners than schools?

      Basically, private language schools often run to quite tight margins and so staff tend to be rather time (not to mention cash) poor.

      A system that requires them to have one member of staff set aside time to do a needs analysis, work out a rough syllabus, then find suitable materials to match from an online source … that’s a potentially major time commitment that the school cannot afford.

      Perhaps the online provider could offer to do all that on the school’s behalf, but then that would require an outside consultant of some kind which, needing to be an expert teacher to do effectively, would ramp up the cost of the offering?

      In that situation, I imagine many private language schools would opt to live with the current system of buying in a shiny ‘off-the-peg’ Oxbridge course and letting the teachers work it out for themselves.

      Alternatively, where a language school – such as a university language centre – may have budget as well as a commitment to giving teachers time off from the classroom to devise materials it’s more than likely that they will opt to devise their own tailor-made materials from scratch.

      That at least has been my experience, both in university language centres and in the British Council.

      So as above, maybe this kind of model should be a B2C rather than B2B?

      • “Alternatively, where a language school – such as a university language centre – may have budget as well as a commitment to giving teachers time off from the classroom to devise materials it’s more than likely that they will opt to devise their own tailor-made materials from scratch.”

        That’s exactly what we found after we first launched…wish we had talked to you first ;-). Only the larger schools and university departments really have the resources to make courses from scratch. So we pivoted our efforts to focus on them, and are doing quite well with that segment.

        That said, lots of our smaller schools use content out of the box and supplement just a little. They will use, say, CUPs Business Benchmark at 90% of how it is written and sequenced, and add some custom activities here and there. This takes very little effort and it’s a quick and easy way to deliver an online component.

        Regarding your “ahead of its time” concern, yes, maybe. But teachers already do this with photocopiers, so the concept isn’t new. And almost every teacher makes some of his or her own content. We grew slowly the first few years but now have over 10K paying students, so no need to worry too much about our fate.

        We may try a B2C product at some point but for now we prefer to focus on supporting teachers and schools.

    • Yes, Cleve, that makes sense. (I sort of actually knew you did this, but I didn’t join the dots). I guess I was imagining a system that ‘sold’ directly to the learners so that they had to do the ‘cohesion’ thing. If it’s mediated by teachers, then I think it does meet the ‘educationally sound’ criterion – assuming that the teachers themselves aren’t just picking and choosing in an arbitrary, even random, fashion – but what self-respecting teacher would do that? 😉

  3. How could this be applied in the ELT industry? Could content creators release iterative versions of their content in exchange for immediate and globally dispersed feedback?

    For more or less the same reasons as Scott, as attractive as the model seems I’m not sure it is transferrable.

    I would add that one thing that a Spotify model doesn’t take into account in an ELT context is the convenience of the end user.

    Teachers are actually far more likely to benefit from receiving a limited range of content rather than a seemingly infinite stream of it for the simple reason that an effective teacher needs read (watch or listen to) the content in order to evaluate how effective and/or relevant it is likely to be to the students.

    This is the chief reason why very little CD-ROM content ever got used – there was already so much in the SB and the WB and often the TB Resource book photocopiables – and then any other content of the teacher’s own – that asking them to plough through every screen on a CD-ROM to then decide which tasks would be worth their students doing …. forget it, there’s just not enough time in the day for all that.

    Once when I was doing research for a publisher both you and I know quite well, I investigated the concept of an ESP ‘kit’.
    As a general concept, it was greeted with a great deal of enthusiasm from the teachers I interviewed but then I pushed further:

    How would the teachers imagine this actually working in practice?
    How would did they see themselves fitting content from such a kit into your daily teaching workload?
    How would they know what they needed and were looking for?

    And so on.

    Sure enough, the more they thought about it the more it became clear that such a ‘kit’ would only really work if they could devote time to sitting down and going through all of the material to then pull out the ones that (as Scott suggests above) could make up a coherent course – it would not only take a huge amount of time, but the time taken would rapidly expand as more and more content was added.

    So far from being a selling point, having loads of content is actually unhelpful.

    The conclusion, possibly ironically, is that for all their failings (actual or believed) a coursebook with a restricted amount of content is more effective, especially when the teacher is at liberty to use it/ not use it adapt it, add to it etc. as circumstances dictate.

    Less is more, basically.

    At least for where teachers are concerned – if you’re talking about a self-study market only, that might have more legs.

    PS Thanks for the summary of where Spotify came from and how it developed and also for the tidbit about Radiohead’s ‘Kid A’ album ; – )

    • Hi Nik,

      Thanks so much for your comment. You raise a really good point there; providing too much choice can potentially remove an individual’s ability to navigate through it in a meaningful and effective way.

      As I was putting the post together I was thinking of it from a self-study perspective, though. A learner could access the material they wanted as and when they needed it. Would that W2L (writer to learner) channel, coupled with the scale of the platform, shape how the content is created and what it looks like at all?

      • I could easily listen to certain pop songs hundreds of times. I definitely wouldn’t want to do the same exercise on the subjunctive hundreds of times. Even a reading passage or conversation would lose its attraction after I’ve used it a bit. I’m not sure what the comparison is here: it feels like apples and pears.

        Granted, a teacher may use the same exercise hundreds of times over the course of a few years. But that limits the audience to language teachers, which count for nothing in number compared to people who like music.

  4. “Could content creators release iterative versions of their content in exchange for immediate and globally dispersed feedback?”

    Don’t we already see this in open educational resources (OERs)? Granted, how users interact with OERs can vary greatly depending on the type of (Creative Commons) license used, but we are already seeing a shift from traditional modes of publishing to a more transparent, open, public, and interactive form of authorship, which I suspect will be a trend that continues. It’s not a stretch to think OER users become those educational stakeholders (teachers, students, etc.) who collectively benefit from not only having the access to the various forms of materials (input), but also are involved in OER creation, reuse, remixing, and distribution. Thus, feedback becomes more of an iterative and reciprocal process of assessment at all levels.

  5. Turn it around. If Spotify were like traditional EFL publishing then 90% of today’s music listeners would still be listening to the Beatles. Maybe not the actual Beatles themselves but rather just new groups singing the same old songs.

  6. Ha! Ha! And teachers would be building their students’ playlists. On the other hand: which would you prefer – a classic Beatles track or an utterly forgettable boy band hit?

  7. Thanks for the reply, Scott.

    I completely agree with your point about breaking educational content into discrete, individual chunks. The aspect of the Spotify model that I was finding particularly compelling, however, was the fact that it had the potential to change how contributors created content, rather than how users would be consuming it. In an ELT context, would such a platform motivate content creators to be more experimental, or to develop learning content in new ways?

    I like your point about pre-packaged learning items that would help counterbalance the apparent lack of cohesion; learning journeys that are compiled into ‘albums’.

  8. I am not sure Spotify is able to ‘change how contributors created content’. I realise Mr Ek wishes they could but from an artist viewpoint I can’t really see the point of sharing ‘work in progress’ on Spotify – I’d rather do this on my own website where I could have some feedback from my fans, as I assume this would be the point of sharing ‘work in progress’. I don’t think Spotify gives you any sort of feedback on your content except for the number of plays? Also, I am not sure if artists would readily confess that they ‘wrote this song in this way because it’s what people demanded on Spotify’? Spotify does affect content distribution, but I don’t think it will affect content creation in a meaningful way – does it really offer any tools that artists haven’t had so far?
    Also, another controversy surrounding Spotify’s model it’s that they have never made a profit so far?

    • Pawel, thank you for this intelligent post. I think we all get so caught up in the romance of new ideas, and intoxicated by the very concept of future change, especially when such sexy terms as ‘experimentation’ and ‘ creative collaboration’ are used, that we typically forget to ask ourselves how this will actually work.

      You are absolutely correct in noting that Spotify is a poor tool for collaborative creation, as fans can only endorse through listening to the music: there aren’t any feedback tools. In this sense, the oldest of old methods – playing to a live audience – would be so much more effective, because the artist can physically see and feel the response of the audience, and talk to them afterwards. A bit like a teacher experimenting in a classroom, really …

      Additionally, new delivery and feedback mechanisms do not in any way guarantee creative or novel content – you can employ state-of-the-art distribution and communication tools to test or promote derivative and cliched materials, ad nauseam: we hugely underestimate the potential of new media to reinforce old preferences. In fact, one of the most interesting (and defining) features of edtech is how ‘potentially’ exciting, high-tech tools gel so easily with conservative agendas and deeply behaviouristic methodologies.

      But, here’s the thing. Given that teachers (as well as rock bands) have been able to exploit technology to initiate dialogues with customers/learners in the interests of experimentation, innovation and content development for nearly a decade now, shouldn’t we be asking why they haven’t done so? Hence, the responsible question (in my opinion) is not whether ELT will become like (or learn from) Spotify – but, rather, why hasn’t it? And not ‘How will edtech change the educational environment?’, but ‘Why hasn’t it?’

  9. All the ELT publishers, big or small, have built their business on the scarcity of any kind of content (Texts, images, video, voices,…) and charge teachers, schools and learners, high prices accordingly.
    Spotify has built its business on the basis of an unlimited supply of music and songs.
    New business ventures in the ELT field will also have to take into account this new reality. That means that publishers will have to sell experiences and results to learners if they want to thrive in this age of new abundance.

    • Having worked for commercial (for profit) as well as non-for-profit ELT publishers in the past, my experience is that creating new content is much more expensive than adapting existing content. It requires a different approach, and greater and varied types of resources, processes and systems. However, I believe that the main issues and problems that new business ventures in ELT need to solve, are intrinsically related to the fragmentation of content formats as well as the ever increasing amount of delivery mechanisms. While end-users may have the desire to make their own choices from this vast pool of options, at some point they are going to need some type of guidance and support. I believe that the role of the teacher will not be replaced, but rather re-placed (re-positioned) to assist the learner in making these types of choices. There is plenty of room for trying and testing new models, but we need to apply the learning of those experiences to improve upon them in an iterative process.

  10. I wouldn’t quite agree – I would say that ELT publishers built their business on skilful adaptation of the content, not on its scarcity. There’s been a wealth of free content available, but learners need somoeone to lead them through it so they actually learn something along the way.
    While unlimited supply of music might make sense to a music fan – though, as I wrote above, Spotify supposedly hasn’t made a profit as of yet – I would say that unlimited supply of learning material is useless to an average student. They need the content formatted to their needs, and in this respect I don’t think we’re in for a big change.
    ‘Selling experiences and results’ sounds great but it’s vague – ELT publishers and teachers have always been providing learners with experiences, so it’s nothing new, it’s just that hopefully new technology would make language learning faster and more enjoyable.
    And results – they mean so many things to different people and depend on so many factors, that the only way to demonstrate that the given company delivers results it promised is for them to devise a framework against which these results would be measured. So, again, it’s nothing new, it’s just marketing?


Leave a comment