1
00:00:00,000 --> 00:00:02,760
Brian Moffitt is with us now from NPR.

2
00:00:02,760 --> 00:00:08,120
And Brian, can you tell us what was announced today on Tuesday and how long it was in the

3
00:00:08,120 --> 00:00:09,120
works?

4
00:00:09,120 --> 00:00:14,160
So, oh boy, I think we have been working on this concept for, I would say, almost two

5
00:00:14,160 --> 00:00:19,800
years now, starting from the earliest iterations internally of how we kind of started to think

6
00:00:19,800 --> 00:00:25,040
about how great it would be if we had listening metrics across all the places our podcasts

7
00:00:25,040 --> 00:00:26,040
were enjoyed.

8
00:00:26,040 --> 00:00:29,400
So, I can kind of take you all the way back there.

9
00:00:29,400 --> 00:00:34,440
So, NPR has an app called NPR One, which you may be familiar with.

10
00:00:34,440 --> 00:00:38,280
And we have a reasonable audience for that, but I think about half a million people every

11
00:00:38,280 --> 00:00:41,200
month use it, monthly active users.

12
00:00:41,200 --> 00:00:46,440
And I think it was around two years ago, we started introducing podcast content into NPR

13
00:00:46,440 --> 00:00:47,440
One.

14
00:00:47,440 --> 00:00:52,200
Previously, it had been a lot of radio shows broken into segments and kind of pieced together

15
00:00:52,200 --> 00:00:54,560
based on listener preference.

16
00:00:54,560 --> 00:00:58,520
You could mark things as interesting or you could take signals from the user as to what

17
00:00:58,520 --> 00:01:01,400
things they listened all the way through, what things they skipped to kind of create

18
00:01:01,400 --> 00:01:05,100
a personalized flow of stories for users.

19
00:01:05,100 --> 00:01:07,480
And we put podcasts in there.

20
00:01:07,480 --> 00:01:13,040
And we realized pretty quickly that because of NPR One being pretty much an always on

21
00:01:13,040 --> 00:01:18,240
platform where you're essentially streaming the content as you're listening to it, we

22
00:01:18,240 --> 00:01:23,540
got very detailed, anonymous information about how people were listening.

23
00:01:23,540 --> 00:01:27,480
How long into an episode did they stay before they had skipped?

24
00:01:27,480 --> 00:01:31,120
Things like that, which things got skipped more than other things.

25
00:01:31,120 --> 00:01:35,840
And we learned that that was tremendously helpful as a publisher when we thought about

26
00:01:35,840 --> 00:01:40,640
what kind of content we're going to create or how we even generate content.

27
00:01:40,640 --> 00:01:44,560
And just to give you a quick example, I think one of the cases that I'm not the expert

28
00:01:44,560 --> 00:01:48,320
in if there's interested in this, we can hook you up with the editorial folks.

29
00:01:48,320 --> 00:01:53,720
But I know when we were launching the NPR Politics podcast a year or so ago, there was

30
00:01:53,720 --> 00:01:58,280
kind of a question of how do you open a podcast to keep listeners interested the most?

31
00:01:58,280 --> 00:02:02,360
Do you sort of just start with some banter around the table and people talking?

32
00:02:02,360 --> 00:02:04,560
Or do you kind of dive right into the issues?

33
00:02:04,560 --> 00:02:08,100
So with NPR One, they were able to test two different versions and see which one actually

34
00:02:08,100 --> 00:02:09,100
resonated.

35
00:02:09,100 --> 00:02:10,160
And there was a very clear winner.

36
00:02:10,160 --> 00:02:13,360
And that was kind of the format we chose for the politics podcast.

37
00:02:13,360 --> 00:02:20,080
So that kind of insight into how people are listening to the content helps us be a better

38
00:02:20,080 --> 00:02:21,440
publisher.

39
00:02:21,440 --> 00:02:23,960
Not getting too technical in the weeds there.

40
00:02:23,960 --> 00:02:26,280
Tell us how this new technology is going to work.

41
00:02:26,280 --> 00:02:27,280
Okay.

42
00:02:27,280 --> 00:02:30,960
So the challenge in the podcasting space over the last decade has been all of us have been

43
00:02:30,960 --> 00:02:34,200
circling around a better definition of a download.

44
00:02:34,200 --> 00:02:37,800
And this is largely because most of the places where people listen are not places that we

45
00:02:37,800 --> 00:02:38,960
control the distribution.

46
00:02:38,960 --> 00:02:43,600
So it's Apple Podcasts is the largest and you've got a bunch of other apps that distribute

47
00:02:43,600 --> 00:02:45,280
our content for free.

48
00:02:45,280 --> 00:02:49,060
And so all we know is that somebody requested the file.

49
00:02:49,060 --> 00:02:53,040
And we know if the bytes were delivered successfully to that user and that's kind of where we are

50
00:02:53,040 --> 00:02:58,540
with the IAB and the common definition known as IAB v2, that we can all speak the same

51
00:02:58,540 --> 00:03:01,720
language about downloads.

52
00:03:01,720 --> 00:03:02,720
That's been great.

53
00:03:02,720 --> 00:03:06,200
It's I think a huge step forward for the industry that we all have that definition.

54
00:03:06,200 --> 00:03:08,600
But we know there's downloads that don't get played.

55
00:03:08,600 --> 00:03:10,200
All you got to do is look at your own behavior.

56
00:03:10,200 --> 00:03:13,440
There's so much good stuff out there that you don't quite get to everything.

57
00:03:13,440 --> 00:03:16,880
We have good research that says people listen to most of what they download and that's been

58
00:03:16,880 --> 00:03:17,880
great.

59
00:03:17,880 --> 00:03:22,000
So what we want to get at is a way to understand from all those places where our podcasts are

60
00:03:22,000 --> 00:03:26,000
played which files that were downloaded actually got listened to.

61
00:03:26,000 --> 00:03:28,440
And I think it's really important to interject here.

62
00:03:28,440 --> 00:03:30,920
We're not worried about who listened to it.

63
00:03:30,920 --> 00:03:34,800
We're just worried that someone listened to it because that's the information we don't

64
00:03:34,800 --> 00:03:38,040
have and that's all we're really looking to get it.

65
00:03:38,040 --> 00:03:40,760
Same things we learned from NPR 1.

66
00:03:40,760 --> 00:03:43,360
What kind of engagement are we getting with this content?

67
00:03:43,360 --> 00:03:47,000
Is it worth the distribution deals we're making with all these different platforms?

68
00:03:47,000 --> 00:03:48,480
Are people listening to what they download?

69
00:03:48,480 --> 00:03:54,680
There's a tremendous cost to NPR and time and energy and money to make a podcast episode

70
00:03:54,680 --> 00:03:57,800
and to distribute it and to deliver those bites to somebody.

71
00:03:57,800 --> 00:04:00,000
We kind of want to know if it was worth it.

72
00:04:00,000 --> 00:04:04,560
So in a quick overview, what we're talking about with remote audio data is actually very,

73
00:04:04,560 --> 00:04:06,160
very simple.

74
00:04:06,160 --> 00:04:11,800
Inside those audio files, which are typically MP3s, although it can be MP4s or other formats,

75
00:04:11,800 --> 00:04:16,280
all we want to do is put a little extra metadata that tells where the important points in the

76
00:04:16,280 --> 00:04:17,280
file are.

77
00:04:17,280 --> 00:04:19,160
Where are the quartile markers?

78
00:04:19,160 --> 00:04:20,960
Where is the halfway point?

79
00:04:20,960 --> 00:04:24,600
For our sponsors, where was the sponsorship starting and ending?

80
00:04:24,600 --> 00:04:30,960
So that we can have the podcast playback platforms, when listening happens at one of those points,

81
00:04:30,960 --> 00:04:35,200
just send us back a little ping that says, listening happened at that point.

82
00:04:35,200 --> 00:04:37,000
That's all we're asking for.

83
00:04:37,000 --> 00:04:39,040
It's hardly any data at all.

84
00:04:39,040 --> 00:04:42,280
We're not asking for any user information or PII.

85
00:04:42,280 --> 00:04:47,240
We're simply, somebody listened at that point in that file and that's the basic metric we're

86
00:04:47,240 --> 00:04:48,800
looking to get back.

87
00:04:48,800 --> 00:04:54,720
So at the end of the day, at some point, will you be able to tell if there's a participating

88
00:04:54,720 --> 00:05:00,520
hosting company like Spreaker or Blueberry or Libsyn, will you be able to tell, okay,

89
00:05:00,520 --> 00:05:05,280
this podcast was listened to after they go through the download number, was listened

90
00:05:05,280 --> 00:05:13,280
to an actual number 58 times or this podcast was listened to for 30 minutes, 23 times.

91
00:05:13,280 --> 00:05:15,960
Is that how deep the numbers are going to get?

92
00:05:15,960 --> 00:05:16,960
That's pretty much it.

93
00:05:16,960 --> 00:05:18,560
And there's different variations of it.

94
00:05:18,560 --> 00:05:23,400
I think the current spec that's out there right now is very, very basic.

95
00:05:23,400 --> 00:05:29,360
It's just a step into this in a way that respects the user privacy so we can at least start

96
00:05:29,360 --> 00:05:31,320
getting something back.

97
00:05:31,320 --> 00:05:35,000
Not like minute by minute or second by second, but just something back.

98
00:05:35,000 --> 00:05:36,420
So that's exactly it.

99
00:05:36,420 --> 00:05:39,560
But it depends on kind of three things happening.

100
00:05:39,560 --> 00:05:44,800
One is publishers will need to encode the rad tags into their episodes before they send

101
00:05:44,800 --> 00:05:46,080
them out into the world.

102
00:05:46,080 --> 00:05:50,040
So that's sort of the basic starting block here.

103
00:05:50,040 --> 00:05:54,440
That requires effort from sort of the second group of people, which are the infrastructure

104
00:05:54,440 --> 00:05:58,560
companies, either the ad servers or the hosting companies who generally are the last touch

105
00:05:58,560 --> 00:05:59,560
of that file.

106
00:05:59,560 --> 00:06:03,200
And particularly if you're going to be doing dynamic ad insertion, which most publishers

107
00:06:03,200 --> 00:06:06,280
do, that changes where everything is in the file.

108
00:06:06,280 --> 00:06:10,640
So you'll need to kind of reinsert the rad tags just to make sure all the data is tight

109
00:06:10,640 --> 00:06:13,240
and this point is where you said it was.

110
00:06:13,240 --> 00:06:15,520
Those are the first two steps.

111
00:06:15,520 --> 00:06:20,160
And then the third step is all the places where podcasts are played will need to look

112
00:06:20,160 --> 00:06:21,540
for that data.

113
00:06:21,540 --> 00:06:25,120
And then when listening happens, securely send that ping directly back to the publisher

114
00:06:25,120 --> 00:06:28,960
so the publisher knows somebody listened to that file in that point.

115
00:06:28,960 --> 00:06:29,960
Gotcha.

116
00:06:29,960 --> 00:06:36,200
Do you think you're going to get all the major publishers or all the publishers to participate?

117
00:06:36,200 --> 00:06:39,600
So the thing that makes me happiest about what we announced today is if you look at

118
00:06:39,600 --> 00:06:43,920
the list of companies that have committed to putting rad on their 2019 roadmap, it's

119
00:06:43,920 --> 00:06:48,640
basically every large major ad server and a lot of the big infrastructure hosting companies.

120
00:06:48,640 --> 00:06:54,360
And I think that's really important because that's kind of like laying the pipes, right?

121
00:06:54,360 --> 00:06:58,320
The plumbing, the infrastructure needed for this to work.

122
00:06:58,320 --> 00:07:02,760
After that, it's actually quite easy for a publisher to put the information in to encode

123
00:07:02,760 --> 00:07:04,300
those rad tags.

124
00:07:04,300 --> 00:07:09,300
And then NPR just announced we're going to deploy rad ping back in our apps.

125
00:07:09,300 --> 00:07:13,180
There are a number of companies that indicated interest that are pretty large podcast playback

126
00:07:13,180 --> 00:07:14,460
platforms.

127
00:07:14,460 --> 00:07:18,440
And I could tell you we're talking to a lot more than were listed in here.

128
00:07:18,440 --> 00:07:21,420
They haven't quite decided which way they're going to go yet.

129
00:07:21,420 --> 00:07:25,120
But I'm optimistic because I think we started in the right place.

130
00:07:25,120 --> 00:07:26,520
We've got a lot of publisher interest.

131
00:07:26,520 --> 00:07:28,520
We've got a lot of publishers ready to commit.

132
00:07:28,520 --> 00:07:34,280
We've got the infrastructure to get the tags in for the delivery systems that exist today.

133
00:07:34,280 --> 00:07:38,240
And now I think it's just a matter of the players looking at the SDKs and realizing

134
00:07:38,240 --> 00:07:42,560
that it's actually not that hard to implement and it totally respects user privacy.

135
00:07:42,560 --> 00:07:45,560
Do podcasters need to do anything?

136
00:07:45,560 --> 00:07:49,360
They'll need to work with a host that supports it or a vendor that supports it, I think is

137
00:07:49,360 --> 00:07:50,920
step one.

138
00:07:50,920 --> 00:07:56,160
And in some cases, that's all you might need to do because I imagine the ads wizards and

139
00:07:56,160 --> 00:07:59,680
tritons and panofleets of the world are going to try to make this as easy as possible for

140
00:07:59,680 --> 00:08:00,920
their customers.

141
00:08:00,920 --> 00:08:05,560
They'll have the encoding solution, the tag insertion, then they'll probably offer capture

142
00:08:05,560 --> 00:08:07,840
the analytics for their customers.

143
00:08:07,840 --> 00:08:11,160
So I think that's probably the easiest thing to do if you're a publisher is keep an eye

144
00:08:11,160 --> 00:08:13,400
on which companies are supporting it.

145
00:08:13,400 --> 00:08:17,680
And then encourage all the places where your audience listens to your podcast to go ahead

146
00:08:17,680 --> 00:08:21,720
and adopt the sending of the data back.

147
00:08:21,720 --> 00:08:28,160
Right now, if you're using a major hosting company, at least to my knowledge, it's somewhat

148
00:08:28,160 --> 00:08:33,160
easy to go to their site and check out your podcast and see how many downloads or listens

149
00:08:33,160 --> 00:08:34,160
you had.

150
00:08:34,160 --> 00:08:40,440
Do you have a vision for how it'll work eventually when there's enough data to read the more

151
00:08:40,440 --> 00:08:43,680
specific data for a podcaster?

152
00:08:43,680 --> 00:08:44,680
Is it going to be the same thing?

153
00:08:44,680 --> 00:08:47,500
You're just going to go back to your hosting company and say, okay, now I know how many

154
00:08:47,500 --> 00:08:52,720
people listen for 30 minutes or how many people fast forwarded past the ad.

155
00:08:52,720 --> 00:08:56,440
Is it going to be the same process or will there have to be a different way of looking

156
00:08:56,440 --> 00:08:57,440
up all that data?

157
00:08:57,440 --> 00:09:00,760
I think you could go either way or maybe even both ways.

158
00:09:00,760 --> 00:09:05,480
I think most of those companies that kind of provide those analytics to you now, be

159
00:09:05,480 --> 00:09:11,960
a PodTrack or your ad server slash distribution company, an ad whether a Triton or a Panoply,

160
00:09:11,960 --> 00:09:15,840
they'll probably deliver a second metric alongside that, that is listen.

161
00:09:15,840 --> 00:09:19,840
You'll be able to look at a show in our case like a Planet Money, see how many downloads

162
00:09:19,840 --> 00:09:25,160
there were with whatever kind of metric definition we use, which for us is the current IABV2

163
00:09:25,160 --> 00:09:26,160
standard.

164
00:09:26,160 --> 00:09:29,400
We'll be able to say, here's how many valid downloads happened.

165
00:09:29,400 --> 00:09:33,040
Then hopefully we get wide enough adoption from the playback platforms.

166
00:09:33,040 --> 00:09:36,600
We'll see and here's how many listens we recorded.

167
00:09:36,600 --> 00:09:39,280
Should podcasters expect to see their numbers go down?

168
00:09:39,280 --> 00:09:41,440
Some really good question.

169
00:09:41,440 --> 00:09:45,480
I think yes is the answer because I think anybody who looks at their own behavior realizes

170
00:09:45,480 --> 00:09:48,400
they probably aren't listening to everything they download.

171
00:09:48,400 --> 00:09:52,520
When we look at our research and our audience panels and you look at the Chervere and the

172
00:09:52,520 --> 00:09:54,520
Edison data, people say they listen to most.

173
00:09:54,520 --> 00:09:57,880
When I look at my own behavior, it's pretty much the same.

174
00:09:57,880 --> 00:09:59,880
But the problem is we just don't know.

175
00:09:59,880 --> 00:10:05,520
There's no way to know because you don't have a holistic way to look at it across all the

176
00:10:05,520 --> 00:10:06,520
different places.

177
00:10:06,520 --> 00:10:11,120
Just to illustrate the problem, for NPR, the biggest podcaster in the world, Apple is the

178
00:10:11,120 --> 00:10:13,600
biggest place people listen to our podcasts.

179
00:10:13,600 --> 00:10:17,760
Just I think a little under 60% of our podcast downloads happen in Apple.

180
00:10:17,760 --> 00:10:25,040
The next biggest chunk is 5% and then 4% and then 3% and then 2% and then 2% and then 1%

181
00:10:25,040 --> 00:10:26,080
and then 1%.

182
00:10:26,080 --> 00:10:31,760
There are so many places where people listen to our podcasts that it needs to be something

183
00:10:31,760 --> 00:10:37,080
that can work everywhere, not just one platform or one app solution to it.

184
00:10:37,080 --> 00:10:40,320
Now is Apple on board for this?

185
00:10:40,320 --> 00:10:44,040
I think we've said in the past, and I'll say again, we've been talking to Apple about this

186
00:10:44,040 --> 00:10:45,040
since day one.

187
00:10:45,040 --> 00:10:49,440
They've been a very interested party in hearing about it.

188
00:10:49,440 --> 00:10:50,960
They're not in the announcement that you said.

189
00:10:50,960 --> 00:10:54,640
They're still kind of thinking it through, but they've been an active participant as

190
00:10:54,640 --> 00:10:57,160
we've developed this over the past two years.

191
00:10:57,160 --> 00:11:01,200
Are you optimistic that they'll be on board at some point?

192
00:11:01,200 --> 00:11:04,320
I'm optimistic that I think everybody will because it just makes sense.

193
00:11:04,320 --> 00:11:08,160
I think one of the biggest issues that we've heard over the past two years as we work with

194
00:11:08,160 --> 00:11:12,320
dozens of companies around this is, will Apple do it?

195
00:11:12,320 --> 00:11:15,680
At the end of the day, I think Apple is going to do what's good for the ecosystem.

196
00:11:15,680 --> 00:11:20,080
I think this is going to be good for the ecosystem and the industry because it's in nobody's

197
00:11:20,080 --> 00:11:23,760
interest to have downloads that aren't played.

198
00:11:23,760 --> 00:11:25,840
That's frankly a waste of resources.

199
00:11:25,840 --> 00:11:27,680
It's a waste of space for users.

200
00:11:27,680 --> 00:11:31,280
It's a waste of bandwidth for publishers.

201
00:11:31,280 --> 00:11:33,480
It's not how most other mediums work.

202
00:11:33,480 --> 00:11:36,280
I think it's kind of inevitable we'll move there.

203
00:11:36,280 --> 00:11:37,920
Will RAD be the solution?

204
00:11:37,920 --> 00:11:38,920
I don't know.

205
00:11:38,920 --> 00:11:43,440
I think we've always taken the position that RAD is the step that we think is possible

206
00:11:43,440 --> 00:11:47,280
because we've tried it and tested it and we've gotten a lot of broad support that it could

207
00:11:47,280 --> 00:11:48,360
be.

208
00:11:48,360 --> 00:11:49,880
If it's something else, that's great.

209
00:11:49,880 --> 00:11:52,000
If it's RAD, that's great.

210
00:11:52,000 --> 00:11:57,520
As a publisher, I kind of need to understand how our content is playing out there.

211
00:11:57,520 --> 00:12:01,260
What do you think this means for advertisers?

212
00:12:01,260 --> 00:12:05,560
We've talked to a lot of them and I think they're very excited about it.

213
00:12:05,560 --> 00:12:10,040
I would argue pretty happy with the definition of the download as outlined by the IAB.

214
00:12:10,040 --> 00:12:14,160
We're all certainly conducting great business around that metric and everyone's kind of

215
00:12:14,160 --> 00:12:16,800
built their ROIs and know that it works.

216
00:12:16,800 --> 00:12:20,800
The repeat endemic sort of direct response advertisers just keep coming back because

217
00:12:20,800 --> 00:12:24,000
it works and we're getting a lot of new brand advertisers in.

218
00:12:24,000 --> 00:12:28,600
I think they're finding success through research and actual performance.

219
00:12:28,600 --> 00:12:33,600
I think at the end of the day, everybody would love to actually have a tighter loop so they

220
00:12:33,600 --> 00:12:38,680
understand not just that the download made it to somebody, but we're talking about people

221
00:12:38,680 --> 00:12:41,520
who heard it and then comparing that to the actions they received.

222
00:12:41,520 --> 00:12:44,880
I think at the end of the day, what you're going to find if we could get there is that

223
00:12:44,880 --> 00:12:49,320
podcasting will be one of the most transparent and effective mediums that you could purchase

224
00:12:49,320 --> 00:12:50,500
as an advertiser.

225
00:12:50,500 --> 00:12:53,680
How long do you think it'll be before we have some usable data and numbers?

226
00:12:53,680 --> 00:12:57,040
Will it be six months or nine months or a year or less than that?

227
00:12:57,040 --> 00:12:59,560
It's going to come in tiny slices.

228
00:12:59,560 --> 00:13:03,320
The NPR One Android app is rolling out RAD support today.

229
00:13:03,320 --> 00:13:09,000
If you're able to encode your files as a publisher and receive the data, you can start seeing

230
00:13:09,000 --> 00:13:14,120
at least from NPR One how listening behavior looks for your content.

231
00:13:14,120 --> 00:13:17,800
I think as more platforms turn on, it's just a nice thing.

232
00:13:17,800 --> 00:13:21,760
The firehose will just get bigger and bigger of the data coming back at you.

233
00:13:21,760 --> 00:13:29,140
My hope, my aspiration is that by the end of next year, 2019, 80% or so of the listening

234
00:13:29,140 --> 00:13:31,040
to our podcast is captured via RAD.

235
00:13:31,040 --> 00:13:32,040
I think that would be great.

236
00:13:32,040 --> 00:13:34,760
I think we would be able to be a much better publisher.

237
00:13:34,760 --> 00:13:37,120
I think the business models will be stronger.

238
00:13:37,120 --> 00:13:42,360
Finally, Ryan, how big a day is today for the podcasting space?

239
00:13:42,360 --> 00:13:43,360
I think that's to be determined.

240
00:13:43,360 --> 00:13:44,760
I mean, me, I'm very happy.

241
00:13:44,760 --> 00:13:46,720
We've been working on this for a very long time.

242
00:13:46,720 --> 00:13:52,720
I think I am grateful that we've had so many companies involved in this, eagerly involved,

243
00:13:52,720 --> 00:13:56,800
pushing it through in different working groups, thinking about all the different problems,

244
00:13:56,800 --> 00:14:00,800
trying to come up with a solution that we think should be able to work for everybody.

245
00:14:00,800 --> 00:14:02,480
I think that's great.

246
00:14:02,480 --> 00:14:07,120
There's been a lot of different ways to attack this, but I think from what I've seen, this

247
00:14:07,120 --> 00:14:11,340
is probably the most holistic and probably the simplest way to attack it.

248
00:14:11,340 --> 00:14:12,340
That excites me.

249
00:14:12,340 --> 00:14:15,960
Again, it's going to depend on how people adopt it from here.

250
00:14:15,960 --> 00:14:18,720
I think the other thing is how many people get involved in the discussion.

251
00:14:18,720 --> 00:14:21,920
Well, we've had 30 companies involved to this point.

252
00:14:21,920 --> 00:14:23,480
We need 300.

253
00:14:23,480 --> 00:14:28,000
If you're a podcast app, if you're a publisher, if you're a platform that hasn't been involved,

254
00:14:28,000 --> 00:14:29,920
go to the URL, get involved.

255
00:14:29,920 --> 00:14:30,920
Let us know what you think.

256
00:14:30,920 --> 00:14:31,920
That's the point.

257
00:14:31,920 --> 00:14:33,200
We need to know what the roadblocks are.

258
00:14:33,200 --> 00:14:34,480
Brian, thanks for your time.

259
00:14:34,480 --> 00:15:01,480
Hey, no problem.