WEBVTT

00:00:00.080 --> 00:00:04.040
I've been running this podcast for a while now,

00:00:04.080 --> 00:00:07.240
and I still don't know how many people actually

00:00:07.240 --> 00:00:11.500
listen to it. It's not approximately, not within

00:00:11.500 --> 00:00:16.579
some sort of range. I don't know. I get a download

00:00:16.579 --> 00:00:20.640
count. That number goes up when an episode drops

00:00:20.640 --> 00:00:23.199
and I feel good about myself for a few days.

00:00:24.620 --> 00:00:27.899
But whether anyone actually hit play, whether

00:00:27.899 --> 00:00:30.760
they made it past the first two minutes, whether

00:00:30.760 --> 00:00:33.179
the episode just sat on somebody's phone and

00:00:33.179 --> 00:00:37.560
get deleted, completely unplayed. I have no idea.

00:00:39.179 --> 00:00:44.979
I talked about why that is in the RSS episode

00:00:44.979 --> 00:00:48.280
a couple of weeks back. The short version is

00:00:48.280 --> 00:00:51.640
that the open protocol gives you independence

00:00:51.640 --> 00:00:55.659
and takes away your measurement. That's the trade

00:00:55.659 --> 00:01:00.340
off. But you know, I've I've been thinking about

00:01:00.340 --> 00:01:03.960
the other side of it for a couple of weeks You

00:01:03.960 --> 00:01:06.980
know the streaming side the platforms own everything

00:01:06.980 --> 00:01:13.680
and they could Tell you so much more But they

00:01:13.680 --> 00:01:17.599
mostly don't This is the Tyler Woodward project

00:01:17.599 --> 00:01:21.439
everything here is my opinion and from my point

00:01:21.439 --> 00:01:24.879
of view Today I want to talk about streaming

00:01:24.879 --> 00:01:27.829
audio metrics and why they're broken in a way

00:01:27.829 --> 00:01:31.650
that most people aren't actually talking about.

00:01:38.870 --> 00:01:43.810
When people complain about streaming metrics,

00:01:44.510 --> 00:01:50.430
the conversation usually goes to one place. Spotify

00:01:50.430 --> 00:01:55.209
pays artists too little per stream, and that's

00:01:55.209 --> 00:01:58.439
a real legitimate problem. It's a legitimate

00:01:58.439 --> 00:02:06.219
complaint, but it's downstream of a more basic

00:02:06.219 --> 00:02:09.879
problem, which is that nobody outside of Spotify

00:02:09.879 --> 00:02:13.919
has ever verified what a quote unquote stream

00:02:13.919 --> 00:02:20.599
actually is. Think about that for a second. The

00:02:20.599 --> 00:02:26.110
number that determines what An artist gets paid

00:02:26.110 --> 00:02:31.289
is the fine counted and reported by the same

00:02:31.289 --> 00:02:36.789
company making the payment. There's no independent

00:02:36.789 --> 00:02:42.229
auditor checking it. There's no government body

00:02:42.229 --> 00:02:47.270
setting the standard for what counts. Spotify

00:02:47.270 --> 00:02:50.789
says you got four million streams. You got four

00:02:50.789 --> 00:02:56.009
million streams. Good luck. Now compare that

00:02:56.009 --> 00:03:00.969
to broadcast radio, which I work in the industry

00:03:00.969 --> 00:03:06.729
Radio audience measurement runs on Nielsen audio

00:03:06.729 --> 00:03:10.669
in the bigger markets It's the portable people

00:03:10.669 --> 00:03:13.569
meter a little bitty device about the size of

00:03:13.569 --> 00:03:18.990
a pager if anybody still remembers those Panelists

00:03:18.990 --> 00:03:22.419
carry those little things around It picks up

00:03:22.419 --> 00:03:25.780
the watermark baked into the station's audio.

00:03:26.560 --> 00:03:31.060
Nielsen counts how many panelists were near which

00:03:31.060 --> 00:03:36.560
signal and for how long. It's not perfect. The

00:03:36.560 --> 00:03:39.840
panel is small. The margins of error are real.

00:03:40.419 --> 00:03:42.759
And if you're, you know, if you've ever sat in

00:03:42.759 --> 00:03:46.900
a ratings meeting and watched people trying to

00:03:46.900 --> 00:03:50.139
draw conclusions from numbers inside the margin

00:03:50.139 --> 00:03:57.020
of error, You know exactly what I mean. But at

00:03:57.020 --> 00:04:02.360
least a third party is doing the counting. The

00:04:02.360 --> 00:04:05.639
station isn't writing its own ratings book. That

00:04:05.639 --> 00:04:11.740
would be insane. Streaming sold itself as the

00:04:11.740 --> 00:04:16.860
upgrade to all of that. Better data, more precision.

00:04:17.959 --> 00:04:22.250
Finally, know your audience. And in some ways

00:04:22.250 --> 00:04:25.870
it is better. Spotify for artists will tell you

00:04:25.870 --> 00:04:28.750
skip rates, save rates, how many playlist ads

00:04:28.750 --> 00:04:34.709
you get. That's genuinely useful stuff, but the

00:04:34.709 --> 00:04:38.129
number at the foundation of the whole system,

00:04:38.269 --> 00:04:42.589
the stream count that determines what you get

00:04:42.589 --> 00:04:49.709
paid has never been independently verified. So

00:04:50.759 --> 00:04:55.259
What actually counts as a stream on Spotify?

00:04:57.379 --> 00:05:03.800
30 seconds if the client reports that 30 seconds

00:05:03.800 --> 00:05:11.319
of a track played It's a stream That's it Check

00:05:11.319 --> 00:05:15.220
Phone in your pocket volume was at zero. Maybe

00:05:15.220 --> 00:05:17.100
it was playing in the background while you did

00:05:17.100 --> 00:05:20.740
your homework made dinner maybe you were on a

00:05:20.649 --> 00:05:27.250
call 30 seconds past the stream is counted now

00:05:27.250 --> 00:05:34.110
for ads advertisements spotify applies a completely

00:05:34.110 --> 00:05:41.189
different standard the ad has to be unmuted above

00:05:41.189 --> 00:05:46.170
zero volume actually playing for at least two

00:05:46.170 --> 00:05:52.779
consecutive seconds There are independent guidelines

00:05:52.779 --> 00:05:57.939
for this. Advertisers demanded it and they got

00:05:57.939 --> 00:06:01.920
it. Because advertisers were spending real money

00:06:01.920 --> 00:06:10.500
and they wanted proof the ads were heard. Now

00:06:10.500 --> 00:06:14.689
artists are getting paid too. But the standard

00:06:14.689 --> 00:06:18.149
for what counts as a listen to their music is

00:06:18.149 --> 00:06:24.870
weaker than this standard for what counts as

00:06:24.870 --> 00:06:31.170
a heard ad. Same platform, different accountability

00:06:31.170 --> 00:06:36.430
depending on who's buying. And before you say,

00:06:36.709 --> 00:06:42.930
well, at least they have some standard. The system

00:06:42.930 --> 00:06:48.449
is gameable. A guy in North Carolina named Michael

00:06:48.449 --> 00:06:54.949
Smith figured this out. Go Google it. From 2017

00:06:54.949 --> 00:07:00.850
to 2024, he uploaded hundreds of thousands of

00:07:00.850 --> 00:07:04.569
AI generated songs under fake, you know, artist

00:07:04.569 --> 00:07:10.230
names. Then ran bot accounts to stream them.

00:07:11.079 --> 00:07:16.220
up to 10 ,000 bots running simultaneously at

00:07:16.220 --> 00:07:23.060
its peak. He pulled in over $10 million in royalties

00:07:23.060 --> 00:07:27.920
across Spotify, Apple Music, Amazon Music, Pandora,

00:07:28.259 --> 00:07:31.939
and YouTube Music before the DOJ caught up to

00:07:31.939 --> 00:07:38.980
him. He pleaded guilty last year. His sentencing

00:07:38.980 --> 00:07:48.209
is in July. Here's the Here's the little detail.

00:07:48.209 --> 00:07:51.529
I can't stop thinking about when I was doing

00:07:51.529 --> 00:07:58.689
this research Spotify paid him roughly $60 ,000

00:07:58.689 --> 00:08:03.089
of that 10 million the rest came from the other

00:08:03.089 --> 00:08:14.170
platforms So Spotify called most of it Amazon

00:08:14.360 --> 00:08:17.879
Apple and YouTube music paid out almost everything

00:08:17.879 --> 00:08:24.199
else for seven years. If this was a report card,

00:08:25.680 --> 00:08:29.600
well, most of the platforms get an F. The other

00:08:29.600 --> 00:08:32.919
thing worth knowing about streaming royalty math

00:08:32.919 --> 00:08:37.360
is how the pool works. And trust me, I freaking

00:08:37.360 --> 00:08:41.220
hate math. All the subscription. and ad revenue

00:08:41.220 --> 00:08:46.120
goes into one big giant pool. That pool gets

00:08:46.120 --> 00:08:48.759
divided by the total number of streams across

00:08:48.759 --> 00:08:54.379
the platforms. Your cut is proportional to your

00:08:54.379 --> 00:09:00.559
share of the total streams, which means every

00:09:00.559 --> 00:09:04.759
fake stream, every bot play, every background

00:09:04.759 --> 00:09:11.090
play that nobody actually heard. is diluting

00:09:11.090 --> 00:09:14.669
the pool for every real artist that's out there.

00:09:15.389 --> 00:09:19.289
Spotify scrubbed over a billion fake streams

00:09:19.289 --> 00:09:23.429
in 2024 and banned more than 10 ,000 artist accounts.

00:09:24.830 --> 00:09:29.289
A billion with a B. That's not some sort of rounding

00:09:29.289 --> 00:09:32.250
error. That's a structural problem with the metric

00:09:32.250 --> 00:09:40.220
itself. And then there's the part that It has

00:09:40.220 --> 00:09:47.220
nothing to do with fraud at all. Spotify added

00:09:47.220 --> 00:09:50.879
audio book access to its premium subscriptions

00:09:50.879 --> 00:09:56.559
last year. 15 hours a month. Fine. Whatever.

00:10:01.019 --> 00:10:06.279
But when they did that, they reclassified premium

00:10:06.279 --> 00:10:11.470
as a quote unquote bundle a bundle of services

00:10:11.470 --> 00:10:15.690
rather than a music subscription that single

00:10:15.690 --> 00:10:19.230
little definitional change triggered a lower

00:10:19.230 --> 00:10:23.669
royalty rate for publishers and songwriters under

00:10:23.669 --> 00:10:28.769
the existing licensing agreements nothing about

00:10:28.769 --> 00:10:34.549
what users actually do changed no songs get removed

00:10:34.549 --> 00:10:39.279
nobody stop listening Spotify changed what it

00:10:39.279 --> 00:10:43.200
called the product and well, hundreds of millions

00:10:43.200 --> 00:10:49.980
of dollars in royalty obligations shifted. The

00:10:49.980 --> 00:10:55.159
mechanical licensing collective, the organization

00:10:55.159 --> 00:10:59.440
that handles those royalty payments sued. Of

00:10:59.440 --> 00:11:03.480
course, a federal judge dismissed the case earlier

00:11:03.480 --> 00:11:08.259
this year and it's still being appealed. Now,

00:11:08.620 --> 00:11:12.259
I'm not a lawyer and I'm not going to, I'm definitely

00:11:12.259 --> 00:11:15.440
not going to pretend I understand all the nuances

00:11:15.440 --> 00:11:19.639
of music licensing law, but you don't need to

00:11:19.639 --> 00:11:22.419
understand the law to understand what happened.

00:11:23.059 --> 00:11:27.779
The platform that counts streams also controls

00:11:27.779 --> 00:11:30.460
how those streams get classified and priced.

00:11:31.179 --> 00:11:33.559
When it's convenient to redefine the product,

00:11:33.799 --> 00:11:37.399
they redefine the product. And if you don't see

00:11:37.399 --> 00:11:39.740
the problem with that, I don't know how much

00:11:39.740 --> 00:11:44.940
more clear I can make it. The problem with streaming

00:11:44.940 --> 00:11:50.220
metrics isn't the data. You know, it's not that

00:11:50.220 --> 00:11:54.519
the data doesn't exist. It's that the data exists

00:11:54.519 --> 00:12:00.519
and gets used selectively. For selling ads, the

00:12:00.519 --> 00:12:05.490
platform supply real tangible standards for paying

00:12:05.490 --> 00:12:09.730
the artist the bar is lower and nobody's double

00:12:09.730 --> 00:12:14.870
checking the numbers or verifying anything now

00:12:14.870 --> 00:12:17.809
as i said earlier i work in broadcast and i run

00:12:17.809 --> 00:12:20.809
this podcast both of those measurement systems

00:12:20.809 --> 00:12:25.549
have real limitations but in broadcast there's

00:12:25.549 --> 00:12:30.330
a third party doing the counting and in podcasting

00:12:30.330 --> 00:12:33.820
i at least know exactly what I gave up when I

00:12:33.820 --> 00:12:38.779
chose to stay on the open protocol. Streaming

00:12:38.779 --> 00:12:44.159
is the only one of the three that tells you it

00:12:44.159 --> 00:12:48.220
has better data and then doesn't let anyone actually

00:12:48.220 --> 00:12:53.919
verify that. To me, that's worth thinking about.

00:12:55.799 --> 00:12:58.500
Thanks for listening to the Tyler Woodward Project.

00:12:58.799 --> 00:13:01.059
This is an independently produced and written

00:13:01.059 --> 00:13:03.440
podcast by myself. And if you'd like to support

00:13:03.440 --> 00:13:07.019
the show in the project, head over to tylerwoodward

00:13:07.019 --> 00:13:09.820
.me slash subscribe and become a member today.

00:13:10.220 --> 00:13:12.559
It goes a long way. Also make sure you leave

00:13:12.559 --> 00:13:15.059
a rating and review over on Apple podcast and

00:13:15.059 --> 00:13:18.720
yes, Spotify, iHeartRadio, uh, Amazon music,

00:13:18.980 --> 00:13:21.600
wherever you're finding the podcast today. It

00:13:21.600 --> 00:13:23.799
helps other people find the show. You can also

00:13:23.799 --> 00:13:27.580
follow me on Instagram threads and bluesky at

00:13:27.580 --> 00:13:31.889
tylerwoodward .com. I'll catch you next week.