How Does Spotify Pay Artists? Interview with Spotify’s D.A. Wallach | Music Business and Trend Mongering

Q:
A:
Very Little.

Did not like that he did not answer direct questions about indie vs major payouts.

That's nonsense. If we assume the generous $.004 amount per play, and DSOTM has 10 tracks, that's $.04 per full album play. In order for the label to make $10, he would have to listen to the full album 250 times. I think in general people vastly overestimate how many times they actually listen to their favorite music. 250 listens is insane. That's listening to DSOTM in its entirety once a month for almost 21 years. Or once a week for 5 years. Or every weekday for an entire year. I'm sure there are a few sad fanboys who have done that but that's obviously not the case for all of the 50 million people who bought the album.

If you check the play count in iTunes of your favourite album, you could calculate how much the artist would get if you listened to it on Spotify!

My top played song was April 26th 1992 by Sublime and I played it 26 times. Sublime would have gotten little more than 10 cents!

Yeah, my top track is 39 plays of I Can Only Give You Everything by MC5. In my teens I would have racked up some serious repeat plays of certain albums but I doubt it would have reached 250.

If you look at last.fm, there are 631,335 people who have listened to DSOTM, for a total of 16,499,167 plays. Meaning that the average is about 26 plays per listener. Or about 10 cents.

Q: How much does Rack Gear know about the music business?

In general last.fm is a pretty fascinating resource for musical statistics. You can look up any artist or album and figure out the ratio between the number of listeners and the number of plays, giving you the average number of plays per listener. I've calculated this for a lot of different artists and it rarely goes above 2.5 plays per listener.

To me that means that the average music consumer has no idea how much they've actually been paying for music. If you listen to a 99 cent track 2.5 times then it cost you 40 cents per play. But can you imagine somebody paying 40 cents to stream a track one time? It wouldn't happen. The illusion of ownership is really powerful.

Maybe you could constructively argue the case for Spotify, rather than destructively picking on one of the other posters?
I'm all ears.

if you actually do some metrics on potential listening experiences like BB is doing,
you can soon find an upper limit of human listening potential in the round.

a question is how many (new) tracks per month would different people be comfortable listening to.
obviously that would be different for different people in different contexts.

so part of the question is what kind of contextual listening audience would you like for your music.
do you want an in depth listener or a superficial listener or both.

if you take the 2.5 listens per track as a general context, you have a superficial listener.

on a superficial reading from a marketing potential for add impressions, something like that
might read out as a general trend towards massive fragmentation of attention experience.

to me the internet as a medium so far, looks to be about maximizing scale and minimizing cost.
That translates to lower willingness to pay as well, because no one really values superficiality.
This doesn't work in favour for Artists who value in depth attention but works well for those who
can leverage the components which work well in a context of superficiality and fragmented attention.

This is also why I think the overall trend will be towards organizations using Music as a vehicle
to piggyback advertisments and marketing campaign strategies.

probably a lot more than you... BTW... I've had lunch w/ DA and heard the rap in person... so...

as for me...

https://gearspace.com/board/5850833-post2.html

Spotify is a huge scam period!! Just try to see how many songs you can listen offline! ....

thanks for this - I'm going to have to check it out.

but it also makes the argument that people might be listening to streaming services to see what they like, and then buying it... and they will listen to the purchased copy more? I don't know. Certainly interesting stats.

Well lastfm started out as something called audioscrobbler several years ago which is an itunes plugin that tracks all of your itunes listens. And when you join it imports all of your past playcounts. Now they track spotify plays as well.

So it's not just people listening to last.fm's streaming radio service. For example when I listen to stuff in my car on my iphone it scrobbles everything I listen to and then syncs it back up to last.fm when I plug in my iphone. CDs played on your computer through itunes would be tracked as well.

Of course it's not a foolproof system. There have been a few times that I accidentally left my ipod playing an album on repeat with the sound off and racked up a ridiculous amount of plays. but if you're nerdy and picky about it you can go in and delete errors like that even.

I don't think that's true at all. With anything album there are going to be people who love it and fanatically listen to the whole thing over and over. There are going to be people who are curious and listen to the whole thing once. There are going to be people who listen to two tracks and decide they hate it. And there will even be people who are obsessively fanatic about only a couple of songs on the album which they listen to a bunch of times while ignoring the other tracks. So all of that averages out. The problem is that traditionally all of those people would have paid the same price for the full L.P. and now they don't have to.

The problem hasn't changed, it's just switched sides. It used to be a problem for the purchaser, now it's a problem for the seller.

And we probably need to find a balance in the middle.
Which would probably be official YouTube channels and some auditioning at retail sites (both already in place by the way).
I can hear most or all of dance tracks before I purchase them. iTunes was pathetic at 30 seconds, and is improved now, but has a way to go before rivalling Juno and Beatport.

It's not really about auditioning before you buy, it's about which music has staying power for the listener. Even among the things that you audition and choose to buy there will be music that you only listen to a few more times and music that you can't stop playing over and over.

I was just addressing the specific point made that no one has to buy a whole album without having heard it (thanks to Spotify it is claimed).

That's not what I claimed at all. It goes beyond that. Before the internet existed there were albums that I played at in-store listening stations and decided to buy but ultimately only listened to once or twice. I've bought music on iTunes on the strength of the short preview and then never really gotten into it later.

Under a purchase model there's a certain threshold that the buyer needs to pass. You hear it a certain number of times and decide that it's worth paying for, then what happens after that nobody knows or cares.

But a streaming model is entirely different. The whole lifespan of the music is what matters, not getting the listener to pass that initial threshold that convinces them to buy. So it's really unknown territory.

My claim is that in the pre-internet music consumer landscape, most people had no idea how many times they listened to things, and therefore how much on average they were paying per listen. I think it was a lot higher than people realize. Therefore getting people to pay the proper amount for streaming may be difficult because "ownership" seems more attractive even if it doesn't really work out that way financially. Witness the guy in the article who thinks he's listened to Dark Side of the Moon enough times that his spotify royalties would make his purchase price "trivial."

To be clear, I'm not making that tired old argument that people used to have to buy a bunch of filler or anything like that.

It's a pretty basic fact: if streaming took over and sales went away, listeners would essentially be paying different amounts for each album, depending on how many times they listen to it.

So there has to be some kind of baseline for which albums get the old retail amount and which don't. Is it only your favorite few albums that should get the equivalent of the purchase price? I don't think anyone wants that. So rates would have to be figured in such a way that the average album you like would make what they used to make retail. Then your favorite albums would actually be making more. And there would still be a segment of albums that you would have happily bought under a retail model but which will make less money from you under a streaming model.

For example if I look back at my itunes playcounts for my favorite albums of the past decade, they all top out at about 30-35 listens. So to get to $15 from 30 plays of a 12 track album, that would be 4 cents a play. But there are albums that I really like and happily paid $15 for that only got 15 plays. So a 12 track album with 15 plays would need a streaming rate of 8 cents a track to equal the retail price. And there are albums with even fewer plays that I bought because they were a new album by a band I already like for example. If you listen to something only once it costs you about a dollar per track per listen.

So even if the perfect sustainable streaming rate were achieved there are still going to be artists who lose out with a streaming model who would have done ok with sales.

10 cents per stream seems ideal. I can't imagine that any artist or label would turn that down. But that would probably make a subscription service prohibitively expensive. Based on my 4 year lastfm history I listen to about 8500 songs a year. Assuming that's only say 2/3rds of my actual listens since it doesn't include vinyl, I'll round it up to 12000. So to sustain that level of listening at 10 cents per stream I would need to pay $100 a month.

The problem is that the people who are older and able to afford more probably listen less while the people who listen the most are kids who can least afford it. Or maybe that's not a problem but it's the way that the system will subsidize itself.

I think 5 cents per play works, even if split between the performers (master rights) and writers (publishing).

yes, I'm not suggesting that 2.5 listens per track (is) a proper way to accertain a general context.
I'm suggesting that it's a danger if this is taken as a superficial reading by marketing people.
because it suggests massive fragmentation into superficial listening experiences.

so aids undermining music which is made to be in depth. it's an example of what you get
if you let the add industry dictate value by making superficial statistical assumptions about listening patterns.
to me Spotify is built on these assumed consumption patterns, but promises scale in the future
to make up for the shortfall in price per listen.

it's an add driven model, not an Artist and listener driven model imo. whatever they try to say.
they would say it's a consumer driven model. I say it amplifies the superficial because
of the sales marketing driven motives and generalised naive assumptions. Then they would say
it's what the consumer wants.

no kid will pay 50 per month. if you assume 5 cents per play to be within some normal bound,
then 0.04 is a what % reduction. ? 125% reduction ?

I think these things should be looked at more per-household rather than per listener. I don't think it makes sense that spotify only allows one simultaneous user. A teenager is not going to pay $50 a month for streaming, but for a family with a couple of teenagers and a parent who can listen to streaming music in their car during a commute or something, that might be doable. That way the teenage heavy users are subsidized by the older, light users. Most media consumption statistics seem to have traditionally been reported per household. I would be interested to know what the average american household spent on music in 1999.

Also I don't really understand what you're talking about re:undermining, etc. If anything a fair streaming model would reward depth since it obviously rewards repeat plays instead of a one time upfront purchase price that ignores longevity.

anyone who is within earshot can hear it... be it one person or a million...

That's a grey area. How wide is "earshot"?

- A set of speakers in a room in your house.
- Multiple sets of speakers in multiple rooms in your house.
- Multiple speakers as above, but wirelessly connected so you can have a set of speakers out in the back yard.
- As above, but IP enabled so you can use non-proprietary wireless on your home network.
- As above, so you can connect to your home music server over the Internet from your phone while out jogging or driving in your car.
- As above, so you can listen on your PC while at work.

OK with me... as you've illustrated there's no need for a "family account" one per person works just fine...

I think it's a pretty simple marketing and pricing issue. My wife might want to listen to something for an hour once or twice a week. It's not enough to justify getting her own $10 a month account. The thing is I would gladly pay $20 a month for Spotify, and if I could use it on multiple computers simultaneously like I can with Netflix, I would install it on her machine and she could use it if and when she wants. I wouldn't worry about whether she was getting her full month's worth because it would just be part of the package. But I wouldn't bother to buy her her own account now because inevitably she would say that she never uses it much and we should cancel it.

It's all about bundling. Look at cable. How many millions of people want to buy just HBO but they can't because they have to buy the whole package? People justify it to themselves because it seems like they're getting all of this extra stuff even if they never watch it.

And of course the same was true with sales. Nobody bought music per user. They shared or made copies among family members. The total household spending is what is important, particularly since parents are inevitably going to be the ones paying for a kid's spotify subscription.