1
00:00:06,000 --> 00:00:10,700
Welcome to another edition of the Always Be Testing podcast with your 

2
00:00:10,700 --> 00:00:15,400
host, Ty De Grange. Get a guided tour of the world of growth, performance 

3
00:00:15,500 --> 00:00:20,000
marketing, customer acquisition, paid media, and affiliate marketing. 

4
00:00:20,400 --> 00:00:25,300
We talk with industry experts and discuss experiments and their learnings in growth, 

5
00:00:25,600 --> 00:00:30,500
marketing, and life. Time to nerd out, check your biases at the door, and 

6
00:00:30,500 --> 00:00:35,000
have some fun talking about data driven growth and lessons learned. 

7
00:00:36,100 --> 00:00:40,900
Hello, and welcome to the Always Be Testing podcast. I'm your host, Ty 

8
00:00:40,900 --> 00:00:45,800
DeGrange. With me today is Tom Salisbury Hunter from Commission 

9
00:00:45,800 --> 00:00:50,600
Junction. We're super excited to get going. Welcome, Tom. Thanks for having me. 

10
00:00:50,600 --> 00:00:55,300
Absolutely. We're excited to have you. Always be testing pod where we talk about 

11
00:00:56,300 --> 00:01:01,100
the growth performance marketing, testing experiments in business and in life. Super 

12
00:01:01,100 --> 00:01:06,100
excited about this episode to dive in with Tom Salisbury Hunter, the vice 

13
00:01:06,100 --> 00:01:11,100
president of client services at Commission Junction. Welcome again and, ready to dive 

14
00:01:11,100 --> 00:01:15,300
into it with you. Yeah. Me too. Been looking forward to this. Heck yeah. 

15
00:01:15,900 --> 00:01:20,800
So give a little bit of background on kinda just the basics of what you do high level 

16
00:01:20,800 --> 00:01:25,600
for the audience. We have a lot of performance marketers, growth people, affiliate people. 

17
00:01:25,600 --> 00:01:30,300
But for those who don't know, I'd love to get a sense of what you do. What's 

18
00:01:30,300 --> 00:01:34,400
commission junction. Give us a bit of a breakdown. Yeah. So CJ is, 

19
00:01:35,300 --> 00:01:40,100
one one of the world's largest affiliate networks. We work with a huge 

20
00:01:40,100 --> 00:01:44,200
multitude of brands, but, primarily, my team is working with 

21
00:01:44,900 --> 00:01:49,900
BlueJeep BlueJeep, enterprise brands to provide affiliate marketing strategy 

22
00:01:50,500 --> 00:01:55,100
and tracking publisher payment and all that good stuff for their for their affiliate and the 

23
00:01:55,100 --> 00:01:59,300
partnership programs. CJ also runs some CJ 

24
00:01:59,700 --> 00:02:04,200
leads lines. CJ also runs CJ influence. So we do, 

25
00:02:04,500 --> 00:02:09,300
influence. We do social. Those both sit in my room yet, but primarily is its core 

26
00:02:09,300 --> 00:02:14,200
affiliate. Awesome. Awesome. And what what kind of, like, like, brands are you 

27
00:02:14,200 --> 00:02:19,100
kind of working with maybe throughout your career or also throughout, you know, your management 

28
00:02:19,100 --> 00:02:24,100
of clients at Commission Junction? Yeah. So I've been with CJ for about ten years. I started 

29
00:02:24,100 --> 00:02:29,100
off as an account manager. I come from London, which is why I speak like this. In our 

30
00:02:29,100 --> 00:02:34,100
London office, I was working with Argos and TUI, who are a huge retailer and a 

31
00:02:34,100 --> 00:02:37,900
huge travel brand, out of the UK and Europe respectively. 

32
00:02:39,000 --> 00:02:43,700
But over time, I've developed. I have touched, 

33
00:02:44,300 --> 00:02:49,200
I would say, probably in excess of a hundred to a hundred and twenty different brands. Now being VP 

34
00:02:49,200 --> 00:02:54,100
knows my team's underneath me to look up the the the vast volume. But everybody from 

35
00:02:54,600 --> 00:02:59,500
Expedia, who've been a a long time client, Disney, who've been a very long time client, to 

36
00:02:59,800 --> 00:03:04,700
Nike, to, CIT Bank, USAA, Experian, TurboTax, 

37
00:03:05,000 --> 00:03:10,000
Intuit. The list kind of goes on and on. I love it. We tend to, again, 

38
00:03:10,000 --> 00:03:14,900
focus on enterprise brands, but I do also have a a large number of mid market and kind of mid 

39
00:03:14,900 --> 00:03:19,900
to large brands on the meet. That's awesome. What are some of the things you think brands 

40
00:03:20,200 --> 00:03:25,200
need to be thinking about in the affiliate marketing space? What do you think some of the things they might, 

41
00:03:25,700 --> 00:03:30,600
missing out on or or not thinking about? I'd love to learn more about some of your observations and 

42
00:03:30,600 --> 00:03:35,500
learnings working with all these great brands. Yeah. I can kinda meta answer that question. 

43
00:03:35,900 --> 00:03:40,800
There are a lot of different things that brands are missing out on, but one of the reasons that I was excited to be 

44
00:03:40,800 --> 00:03:45,600
featured on this podcast is I think the biggest thing that most brands miss out on is working out what they are 

45
00:03:45,600 --> 00:03:50,300
missing. It's instituting some kind of testing function, learning 

46
00:03:50,300 --> 00:03:55,200
function, something along those lines within their affiliate program so that they can start saying, okay. 

47
00:03:55,200 --> 00:03:59,900
We don't do x currently, and maybe we don't have the data that we 

48
00:03:59,900 --> 00:04:04,800
need or the, maybe, the expertise that we need on on their own side to forecast what 

49
00:04:04,800 --> 00:04:09,800
x would do from a ROI's perspective, but we are still gonna test that. We are still gonna work out whether 

50
00:04:09,800 --> 00:04:14,800
it works. Test and learn functionality is something that we try to instill with the majority of the 

51
00:04:14,800 --> 00:04:19,700
brands that we work with. Test and learn functionality, I think, is probably a thing I see 

52
00:04:19,700 --> 00:04:24,300
most regularly missed within the affiliate industry or misimplemented within the 

53
00:04:24,300 --> 00:04:29,200
affiliate industry. I love that. Obviously, it's the namesake of the pod. It's really 

54
00:04:29,200 --> 00:04:34,100
central to how my team and I think about and see so many 

55
00:04:34,100 --> 00:04:38,700
companies that tap into experimentation are the ones that are really 

56
00:04:39,200 --> 00:04:43,900
capturing more value, capturing more demand. They're on that 

57
00:04:44,300 --> 00:04:49,300
kind of more bleeding edge of the curve. So I think it's such an interesting topic. When you 

58
00:04:49,300 --> 00:04:54,100
think about an experiment or a test, maybe just explaining for the audience, like, what 

59
00:04:54,100 --> 00:04:59,000
what is sort of your definition of of a proper test that a brand that you 

60
00:04:59,000 --> 00:05:03,800
might manage should be thinking about running? And how do you kinda just define it? 

61
00:05:04,200 --> 00:05:09,100
Yeah. From a methodology standpoint, it's typically pretty simple, actually. The majority of 

62
00:05:09,100 --> 00:05:14,100
the time, the reason that I see the need for a test is we don't have a good 

63
00:05:14,100 --> 00:05:19,000
enough amount of visibility into whether this will definitely return, you know, the CPA, the ROAS 

64
00:05:19,000 --> 00:05:23,700
that we need, the effective CPA or the ROAS that we need on a large scale. And 

65
00:05:23,700 --> 00:05:28,300
so I would advocate for putting aside a smaller amount of budget that's not gonna affect your 

66
00:05:28,300 --> 00:05:33,300
overall ROAS, your overall program, just to flat out test whether it is 

67
00:05:33,300 --> 00:05:38,300
gonna return a ACPA for you. Now the actual output is gonna depend on the client. 

68
00:05:38,500 --> 00:05:43,300
It may be that you just want straight up last click CPA through your affiliate 

69
00:05:43,300 --> 00:05:48,300
network, through, you know, chosen attribution provider or wherever else. It may be that the 

70
00:05:48,300 --> 00:05:53,100
the action that you're looking for in, you know, in CPA is a lead. It may be that it's a 

71
00:05:53,100 --> 00:05:58,100
volume of impressions. But, really, it's it's putting money down. It's working out the amount 

72
00:05:58,100 --> 00:06:02,700
of money that you're willing to spend on x to try to work out whether it will work for 

73
00:06:02,700 --> 00:06:07,700
you. It's agreeing on what the KPIs that you're looking for as an output from that would be. 

74
00:06:07,800 --> 00:06:12,600
And then it's running a test in good faith, which I think pretty much everybody that ends up doing 

75
00:06:12,600 --> 00:06:17,300
this does, where you are making the best possible effort to allow that 

76
00:06:17,300 --> 00:06:22,200
publisher to allow that strategy to allow that partner, to achieve whatever that that 

77
00:06:22,200 --> 00:06:26,200
CPA looks like. I love that. And what do you find to be some of the pitfalls 

78
00:06:27,000 --> 00:06:31,900
for clients and brands that are kind of going through that test process? Like where do 

79
00:06:31,900 --> 00:06:36,800
they kind of falter? Where do they go wrong? How do you kind of coach them on that? Yeah. So the 

80
00:06:36,800 --> 00:06:41,500
pitfalls themselves, I see, I guess, three different pitfalls that are that are pretty 

81
00:06:41,500 --> 00:06:46,400
regular. One of them would be going into the test with a bias, going into 

82
00:06:46,400 --> 00:06:50,900
the test assuming this is how this is gonna act. And so 

83
00:06:51,400 --> 00:06:56,200
not to the point that I made a second ago, kind of not throwing everything that you can at it, not giving the 

84
00:06:56,200 --> 00:07:01,100
partner the best possible chance for success. If you go into somebody and you say, I'm only gonna invest 

85
00:07:01,200 --> 00:07:06,100
ten bucks in this, and I need you to return absolutely everything that I that I've asked for, there's 

86
00:07:06,100 --> 00:07:11,000
no way that partner's gonna do that. You need to go in. You need to say I'm willing to fund this to the extent that you 

87
00:07:11,000 --> 00:07:16,000
tell me you need to to drive results. After that, you can work out what scaling 

88
00:07:16,000 --> 00:07:21,000
back looks like. You can't go in and say, I'm only gonna give you, you know, a little bit of money. I'm only gonna give you 

89
00:07:21,000 --> 00:07:25,900
a little bit of investment or time or assets or whatever else it requires for success. And then be surprised 

90
00:07:25,900 --> 00:07:30,600
if they come back to you and say, well, I didn't have enough to drive, you know, what, what my full 

91
00:07:30,600 --> 00:07:35,500
potential was. So the the kind of bias or the caution, I guess, as well 

92
00:07:35,500 --> 00:07:40,400
is problematic. On the other side of that, I have seen problems where people haven't had enough caution 

93
00:07:40,700 --> 00:07:45,700
with the overall structure of their test and learn programs, and they've been willing to just kind of throw money at 

94
00:07:45,700 --> 00:07:50,700
anything that comes across their plate. It's a really interesting thing to manage as as, a 

95
00:07:50,700 --> 00:07:55,400
strategist and as an agency slash network, because we're effectively going to people and 

96
00:07:55,400 --> 00:08:00,100
saying stop doing stuff. So you have to have parameters in place that ensure 

97
00:08:00,400 --> 00:08:04,600
stop spending money. Right. Exactly. Stop making like, stop paying me money, stop paying everybody money. 

98
00:08:05,200 --> 00:08:10,000
Affiliate network structures. And I'm sure agency structures as well have changed a little bit, so that it's 

99
00:08:10,000 --> 00:08:15,000
not always dependent on the amount of money that comes through us, but it still feels odd to say 

100
00:08:15,000 --> 00:08:19,800
to people don't do as much as you are at the moment, but it is super 

101
00:08:19,800 --> 00:08:24,300
important to say, to discuss as part of the setup of the test and learn program 

102
00:08:24,700 --> 00:08:29,500
whether the parameters that you have, whether the spend that you have assigned are at the right level, they're not gonna be 

103
00:08:29,500 --> 00:08:34,500
detrimental to your wider program. While I will advocate all day for testing, all 

104
00:08:34,500 --> 00:08:39,300
day for testing, I don't want somebody to run a budget or to risk a 

105
00:08:39,300 --> 00:08:44,100
budget that is gonna overall affect the veracity of the channel, the integrity of the channel, 

106
00:08:44,400 --> 00:08:49,300
their their ability to have good conversations with their boss. So one would be make sure you aren't too cautious. 

107
00:08:49,900 --> 00:08:54,800
Two would be make sure you're cautious enough. And then three would be ensuring that your 

108
00:08:54,800 --> 00:08:59,600
test and learn budget is rolling rather than something that you put aside. So I've seen some 

109
00:08:59,600 --> 00:09:04,100
brands say, okay, we have a twenty thousand dollar test and learn budget for this month. 

110
00:09:04,600 --> 00:09:09,400
They will find five thousand dollars of things that work, they will find another five thousand dollars the next 

111
00:09:09,400 --> 00:09:13,600
month, do that for four months and say, right, our entire test and learn budget is, 

112
00:09:14,400 --> 00:09:19,400
is working, but it's taken up. You kinda need to graduate people 

113
00:09:19,400 --> 00:09:24,200
that are in a a testing function that you have into your main program when they start to 

114
00:09:24,200 --> 00:09:29,200
work. Otherwise, you know, in that example, all you're gonna do is find four partners. Yep. And 

115
00:09:29,200 --> 00:09:33,300
so, like, without naming a brand, what's kind of like the best 

116
00:09:33,800 --> 00:09:38,100
structured test you've seen? Or what's like a textbook example where, 

117
00:09:38,600 --> 00:09:43,500
man, this this brand really set it up well. They had the right expectations going in, 

118
00:09:43,500 --> 00:09:48,500
kind of played it perfectly. Like, what what would be an example in your mind, without saying the 

119
00:09:48,500 --> 00:09:53,400
brand's name? I'd love to hear kind of, like, what happened. What was the result? Yeah. So about 

120
00:09:53,600 --> 00:09:58,400
about nine months ago now, actually, because it was just before q four. Don't test in q four, folks. 

121
00:09:59,500 --> 00:10:04,500
Or at least don't don't make big bets in q four. Fair. About nine months ago, we had 

122
00:10:04,500 --> 00:10:09,500
a brand that put aside a pretty significant budget, in kind 

123
00:10:09,500 --> 00:10:14,400
of objective terms. There are bigger brands that are able to afford it to test 

124
00:10:14,700 --> 00:10:19,600
a completely new publisher model. It was something that they and their competitors had 

125
00:10:19,600 --> 00:10:24,500
not really made any headway into. The conversations around the 

126
00:10:24,500 --> 00:10:29,300
misalignment of the potential for that publisher vertical and the achievement of that that 

127
00:10:29,300 --> 00:10:34,100
client and, and their competitors, was rife. We had an 

128
00:10:34,100 --> 00:10:38,900
enormous conversation about, you know, how can this be a six, seven, eight figure 

129
00:10:39,300 --> 00:10:44,200
potential publisher, and yet we are only making x amount of money. And the publisher 

130
00:10:44,200 --> 00:10:49,000
came to us and said, look. The reason for it is there is effectively a startup cost to this. It's not an 

131
00:10:49,000 --> 00:10:53,900
integration cost. It is you know, you need to be spending x amount of money in order to engage with our 

132
00:10:53,900 --> 00:10:58,800
audience. The temptation at the beginning of that conversation for that brand and actually for a 

133
00:10:58,800 --> 00:11:03,700
couple of their competitors that I happen to know about was to say, okay. We will we will put aside 

134
00:11:03,700 --> 00:11:08,300
this significant budget. We will give this to you. We are going to dictate to you 

135
00:11:08,300 --> 00:11:13,100
exactly how you should spend it. And the really it was a 

136
00:11:13,100 --> 00:11:17,900
really cool thing. It's kind of sad to say, but the really cool thing I saw throughout the process was the 

137
00:11:17,900 --> 00:11:22,700
brand's realization that is not the best way to set a publisher up for success. 

138
00:11:22,800 --> 00:11:27,600
That if we're going to judge whether a publisher is capable of making money, what we need to do is 

139
00:11:27,600 --> 00:11:32,600
say to them, here's the money. You tell me how you're gonna spend it. I just need you to tell 

140
00:11:32,600 --> 00:11:37,600
me at the end of it, I put my best foot forward. Can't be any conversation at the end where you say you hampered us, you 

141
00:11:37,600 --> 00:11:42,400
changed, you know, you you changed how we would have liked or worked. We have to know what you can 

142
00:11:42,400 --> 00:11:47,300
do if if the gloves come off. Obviously, there have to be branding restrictions. Obviously, there have to be, 

143
00:11:47,300 --> 00:11:52,000
you know, fraud and compliance and everything else restrictions, and there's zero 

144
00:11:52,000 --> 00:11:56,700
question about that. But, really, the control was handed over. It was here is the amount of money that we 

145
00:11:56,700 --> 00:12:01,600
have. Give me the most amount of money back within, you know, these these few 

146
00:12:01,600 --> 00:12:06,600
parameters that we have. And it was a wild success. I believe that brand is now something 

147
00:12:06,600 --> 00:12:11,600
like six percent of their total program. That publisher is now about six percent of their total program. It 

148
00:12:11,600 --> 00:12:16,300
did graduate. What type of publisher it sounds like there was a lot of conversation around, 

149
00:12:16,300 --> 00:12:21,200
like, controversy, for lack of better term, about partnering with this this 

150
00:12:21,200 --> 00:12:26,200
this affiliate, this partner. Can you share more about, like, the type of partner it was or, like, maybe some of 

151
00:12:26,200 --> 00:12:31,000
the why behind that analysis going into the test? Yeah. I don't wanna name 

152
00:12:31,000 --> 00:12:36,000
exactly what kind of partner it is because it's gonna betray the exact partner. There are very few people in the space, 

153
00:12:36,000 --> 00:12:40,900
but it was a a really emerging publisher model to the extent that there are probably only two 

154
00:12:40,900 --> 00:12:45,600
or three publishers that are in that space. The reason for the hesitation, I think, is really 

155
00:12:45,600 --> 00:12:50,400
the the reason that we need testing, testing our budgets within, the affiliate 

156
00:12:50,400 --> 00:12:54,800
industry at large. It was that there was zero data there was zero data to 

157
00:12:55,100 --> 00:12:59,700
forecast the CPA from. There was a lot of data that indicated that there was audience 

158
00:12:59,700 --> 00:13:04,000
alignment. There was a hell of a lot of data that indicated that there was not much, 

159
00:13:04,400 --> 00:13:09,300
actual customer overlap. So you had perfect audience. We 

160
00:13:09,300 --> 00:13:14,300
haven't really spoken to them. We know that that audience engages with brands that are different to us, 

161
00:13:14,300 --> 00:13:19,200
but shop in the same way, you know, the same kind of demographic, same kind of income levels. But there was 

162
00:13:19,200 --> 00:13:23,900
nothing to say, okay. If we try to execute this, this is what this is going to cost. The 

163
00:13:23,900 --> 00:13:28,500
controversy came purely from the fact that affiliates are CPA channel, and I think our 

164
00:13:28,500 --> 00:13:33,300
temptation as affiliate managers, affiliate marketers, clients, agencies, everybody 

165
00:13:33,800 --> 00:13:38,800
is to always be able to say, this is the amount of money that we're gonna make from this test, or at 

166
00:13:38,800 --> 00:13:43,800
least this is with a high confidence the range that we're gonna we're gonna make. Because of 

167
00:13:43,800 --> 00:13:48,500
the CPA structure of affiliate? Because of the CPA structure of affiliate and because it's the 

168
00:13:48,600 --> 00:13:53,600
paradigm within which we engage on a day to day basis. Right? When we're talking about established 

169
00:13:53,800 --> 00:13:58,500
partners, we're always talking about how much they made, what their CPA was, what we think their 

170
00:13:58,500 --> 00:14:03,500
CPA is gonna be next month, what we think the ROAS of this newsletter will be, you know, if we run it 

171
00:14:03,500 --> 00:14:08,200
on this day rather than that date. We always have that backstop of 

172
00:14:08,200 --> 00:14:13,000
data that allows us to reach a, hopefully, very accurate ROAS 

173
00:14:13,000 --> 00:14:17,800
range. And when we lack that data, I think the affiliate channel's kind of unique in 

174
00:14:17,800 --> 00:14:22,600
that it causes caution because nobody wants to put money understandably, nobody wants to put money 

175
00:14:22,600 --> 00:14:27,500
up upfront for a risk when, you know, there are twenty other conversations in the wings about publishers that were 

176
00:14:27,500 --> 00:14:32,200
already established with how you could spend that money with some kind of guaranteed range. 

177
00:14:32,600 --> 00:14:37,400
What interesting concepts. I I've what I'm hearing you saying is, like, brands need to come into 

178
00:14:37,700 --> 00:14:42,500
affiliate marketing with a very test and learn mentality. Right? Brands need to come 

179
00:14:42,500 --> 00:14:47,300
into affiliate marketing with a kind of check their biases at the door as much as they 

180
00:14:47,300 --> 00:14:52,100
can recognize them. Right? And you're saying, hey. Let's have a a small 

181
00:14:52,100 --> 00:14:57,100
percentage of a of a budget allocation to go towards willingness to 

182
00:14:57,800 --> 00:15:02,700
spend to learn, not necessarily spend to get a return. And then, hey. Once that signal is 

183
00:15:02,700 --> 00:15:07,700
met, we're gonna pause that test and move that success or that 

184
00:15:07,700 --> 00:15:12,700
winning test into the greater budget allocation that you said it's just evergreen and running 

185
00:15:12,700 --> 00:15:17,700
and per expected to perform. Yeah. Then you've got this other testing budget that's 

186
00:15:17,700 --> 00:15:21,700
smaller that can keep firing, keep going after those new emerging 

187
00:15:22,500 --> 00:15:27,500
counterintuitive things. Right? Is that kind of what your sounds like that's kind of your your point there, 

188
00:15:27,500 --> 00:15:32,400
which I think is really, really spot on. Yeah. That's exactly it. I think there was a lot of talk 

189
00:15:32,800 --> 00:15:37,500
in tech and marketing in general five five years ago, maybe even ten years ago, about 

190
00:15:37,500 --> 00:15:42,200
failing fast. Yeah. And I don't think you wanna try and fail fast with your entire 

191
00:15:42,200 --> 00:15:47,100
program. I don't think you wanna risk all of your budget. I think ring fencing a small amount of budget 

192
00:15:47,400 --> 00:15:52,400
to learn how to fail fast, to learn what successes look like, also to learn, you know, how we 

193
00:15:52,400 --> 00:15:56,900
run this and maybe we need to adjust x y z and run it again is 

194
00:15:57,500 --> 00:16:02,400
deeply needed. Otherwise, what you end up relying on is kind of just organic growth. It's, you know, 

195
00:16:02,400 --> 00:16:07,000
how can the established players in the game, your retail units, your Ebates, 

196
00:16:07,800 --> 00:16:12,500
your Wirecutters even, how can they drive me growth and 

197
00:16:12,500 --> 00:16:17,300
expecting that you're gonna see something different from those partners that you've been working with 05:10, significantly 

198
00:16:17,700 --> 00:16:22,100
different from those partners that you've been working with for 05:10, fifteen, twenty years 

199
00:16:22,600 --> 00:16:27,200
without, again, without testing and learning with them, without trying new strategies with 

200
00:16:27,300 --> 00:16:32,200
them. I I don't think it makes any sense to do that. There is no brand that all of a sudden is gonna drive 

201
00:16:32,200 --> 00:16:37,100
you 10:20, thirty, forty percent growth without changing how you work with 

202
00:16:37,100 --> 00:16:42,100
them. And there are very few brands who you can change who you, how you work with without putting some kind 

203
00:16:42,100 --> 00:16:47,000
of investment in upfront. Yeah. For sure. You you kind of touched on something interesting before 

204
00:16:47,000 --> 00:16:51,700
about setting aside that experimentation budget and counseling clients 

205
00:16:51,700 --> 00:16:56,300
and brands to say, it's okay if this loses 

206
00:16:56,400 --> 00:17:01,400
money. And I I think that was kind of like like share more about that. Like, how 

207
00:17:01,400 --> 00:17:06,300
do you how do people react when you tell them that initially, and how is how does that 

208
00:17:06,300 --> 00:17:11,300
conversation go? Yeah. I get a lot of wide eyes when I talk about it. The 

209
00:17:11,600 --> 00:17:16,600
the concise way that I will put that usually is if your test and 

210
00:17:16,600 --> 00:17:21,500
learn budget is consistently delivering you a positive ROAS, you are doing 

211
00:17:21,500 --> 00:17:26,200
it. Fundamentally, that will mean or you're psychic. And if you're 

212
00:17:26,200 --> 00:17:31,200
psychic, great, I wanna work with you. But if your test and learn budget is consistently raw as 

213
00:17:31,200 --> 00:17:35,800
positive or at least meeting your program raw as targets, it probably means that you're not taking enough 

214
00:17:35,800 --> 00:17:40,700
risks and that that should have been spent that just sat within your evergreen program. Or it 

215
00:17:40,700 --> 00:17:45,500
means that you're coming up against what I spoke about earlier, which is, you know, you've set a test and learn 

216
00:17:45,500 --> 00:17:50,400
budget, but you found four things that work, and you're continuing to call them tests for far too 

217
00:17:50,400 --> 00:17:55,200
long. You should graduate things that work into your core program. 

218
00:17:55,300 --> 00:18:00,200
And so by definition, that experimentation budget, that experimentation that you have 

219
00:18:00,200 --> 00:18:04,700
ring fenced should always be riskier below program 

220
00:18:05,100 --> 00:18:10,000
ROAS below program CPA in order to to source those those 

221
00:18:10,000 --> 00:18:14,900
positive rise new opportunities. Very cool. I love that, Tom. The 

222
00:18:14,900 --> 00:18:19,900
brands you've worked with are insanely awesome. You've worked you've seen a lot, you 

223
00:18:19,900 --> 00:18:24,900
know, hundreds of complex, challenging topics and issues 

224
00:18:24,900 --> 00:18:29,900
in affiliate marketing. What what do you think is kind of the some of the 

225
00:18:29,900 --> 00:18:34,600
emblems or signals of a really healthy, fantastic program? We've talked about 

226
00:18:34,600 --> 00:18:39,500
testing, But kind of when you, like, look under the hood and see what's available from a 

227
00:18:39,500 --> 00:18:44,500
strategy tactics mix perspective, I'm excited to just just get your perspective on what you've 

228
00:18:44,500 --> 00:18:49,300
seen work really well for for some of the best brands. Yeah. I think that's a 

229
00:18:49,300 --> 00:18:54,100
relatively easy one because I definitely have a favorite way to to make up a program. 

230
00:18:54,600 --> 00:18:59,600
So unanimously, the brands that I see 

231
00:18:59,600 --> 00:19:04,400
who are most exciting to work within the industry, not just for me, 

232
00:19:04,400 --> 00:19:09,300
but, you know, where I see publishers being enthused, I see teams being enthused, I see even the client 

233
00:19:09,300 --> 00:19:12,600
contacts themselves being enthused. A client where 

234
00:19:13,900 --> 00:19:18,600
they understand that there are different measures of value for 

235
00:19:18,600 --> 00:19:23,500
different partners in the space. Linking back to something I said earlier on, I think we can 

236
00:19:23,500 --> 00:19:28,500
get really focused on last click CPA. And on the last click CPA of 

237
00:19:28,500 --> 00:19:33,300
an individual publisher in an individual journey, what I see with really 

238
00:19:33,300 --> 00:19:38,300
exciting programs, partners, clients, even measurement tactics, where they start to understand, 

239
00:19:38,400 --> 00:19:43,300
okay, what is the full holistic downstream value of each element of 

240
00:19:43,300 --> 00:19:48,200
my plan? Is it somebody that drives the last click? Is it somebody that 

241
00:19:48,200 --> 00:19:53,100
grabs somebody who's nearly there and gets them to be there at a better rate than I can do 

242
00:19:53,100 --> 00:19:58,100
that myself? And if so, you know, understands the value of that because I definitely hear some 

243
00:19:58,100 --> 00:20:03,100
people that argue with the value of that and will say if they're nearly there, I don't want don't wanna interact with them. The other end 

244
00:20:03,100 --> 00:20:08,100
of the spectrum, though, it's understanding, okay, there are a load of affiliate partners out there. There is 

245
00:20:08,100 --> 00:20:13,100
a load of media out there that is only available through the affiliate channel and the partnerships channel. Mhmm. That 

246
00:20:13,100 --> 00:20:18,100
is not necessarily gonna convert one hundred percent on the last click, but I understand is gonna 

247
00:20:18,100 --> 00:20:23,100
contribute maybe just their partnership's ecosystem or maybe to their overall site ecosystem. 

248
00:20:24,300 --> 00:20:29,100
Again, not just understanding that that is true, but truly understanding, 

249
00:20:29,200 --> 00:20:33,700
measuring, and analyzing what that looks like and being able to have an open conversation 

250
00:20:33,800 --> 00:20:38,800
internally about why you are running a strategy with that if you're CPA based or on the 

251
00:20:38,800 --> 00:20:43,600
other end of the program, why you're running, you know, effectively bottom of 

252
00:20:43,600 --> 00:20:48,500
funnel strategies that help conversion. Those open conversations, that ability to measure, that 

253
00:20:48,500 --> 00:20:53,500
ability to understand every element of the funnel or the messy middle 

254
00:20:53,500 --> 00:20:58,500
or whatever terminology we're using this year tends to create, one, a healthy program, because 

255
00:20:58,500 --> 00:21:03,100
you have great mix of publishers. Two, healthy growth because you are 

256
00:21:03,300 --> 00:21:07,700
dipping into every possible pool that you can for growth, and you are leveraging 

257
00:21:08,300 --> 00:21:12,800
you're leveraging that you are pulling the levers for growth at both ends of the spectrum. 

258
00:21:13,300 --> 00:21:18,300
Like, when when we're talking about awesome, I kind of wanna include that as well. It allows you to play a little 

259
00:21:18,300 --> 00:21:23,300
bit of jazz. It allows you because you have that measurement, because you have that understanding, it allows 

260
00:21:23,300 --> 00:21:28,100
you to play outside of the parameters of just, hey, if we press this button, this number goes 

261
00:21:28,100 --> 00:21:33,100
up and start to say, okay, what happens if I mix these ingredients? What happens if I take this thing 

262
00:21:33,100 --> 00:21:37,900
and this thing that you would never would have expected to have worked together and bring them together to drive success? 

263
00:21:38,100 --> 00:21:42,400
I love that. Play a little jazz. That's, such a great, mantra for, I think, 

264
00:21:42,900 --> 00:21:47,900
many things. You kinda touched on briefly, like, last click. Can you 

265
00:21:47,900 --> 00:21:52,100
kinda describe, you know, some of the attribution models maybe that you've 

266
00:21:53,000 --> 00:21:58,000
seen work well? Obviously, to your point earlier, it's very unique to the brand, but what's 

267
00:21:58,000 --> 00:22:03,000
your take on attribution generally speaking? Where do people kind of not get it right? Where do you 

268
00:22:03,000 --> 00:22:07,900
like we're going back to the core question of what's that best in class affiliate program 

269
00:22:07,900 --> 00:22:12,900
look like, what what do you see often when it comes to attribution? Yeah. I would kick 

270
00:22:12,900 --> 00:22:17,700
off by saying for most clients maybe or at least for a large 

271
00:22:17,700 --> 00:22:22,500
number of of business types, LastClick isn't an attribution model at all. LastClick is a 

272
00:22:22,500 --> 00:22:27,300
payment model. Another one of my my mantras, outside of 

273
00:22:27,600 --> 00:22:32,500
playable jazz, which is actually one of my director's, mantras, which I love and and whatever else I set 

274
00:22:32,500 --> 00:22:37,200
up the stream, is it doesn't matter whether you pay your partners via 

275
00:22:37,200 --> 00:22:41,700
carrier pigeon. It doesn't. What matters is the amount that you pay them 

276
00:22:42,200 --> 00:22:47,000
and whether that aligns to what you're getting from them. Whether you're paying last click, 

277
00:22:47,200 --> 00:22:52,000
first click, fractional, carrier pigeon, placement, whatever 

278
00:22:52,000 --> 00:22:56,700
else. At the end of the day, partners are looking at what they're getting in their bank account, and they are 

279
00:22:56,700 --> 00:23:01,600
prioritizing your business based on that And not necessarily against your competitors either, just, you 

280
00:23:01,600 --> 00:23:06,600
know, in the wider context of their own business and who they're making money from. But equally for 

281
00:23:06,700 --> 00:23:11,600
clients, for advertisers, for networks as well, the fact that you pay on the last click doesn't 

282
00:23:11,600 --> 00:23:16,600
mean that your attribution has to be on the last click CPA. It is just a 

283
00:23:16,600 --> 00:23:21,400
payment model, and it is, I think, a pretty effective payment model. If you have, you 

284
00:23:21,400 --> 00:23:26,200
know, somebody looking at your program who has, any form of experience in the industry, they're gonna be 

285
00:23:26,200 --> 00:23:31,200
able to adjust the downstream impacts of that payment model to make sure that each partner is being paid the right amount 

286
00:23:31,500 --> 00:23:36,500
even if you're measuring on the world's most complex attribution system. On actual attribution, 

287
00:23:37,400 --> 00:23:42,300
the cool stuff that I see tends to be focused around, I'm a massive data nerd. 

288
00:23:42,300 --> 00:23:47,300
I'm a massive data structure nerd. And it tends to be focused around maybe some game theory 

289
00:23:47,300 --> 00:23:52,100
attribution. I see some great stuff with econometrics and models that take that into account. 

290
00:23:52,700 --> 00:23:57,700
It tend to ask tends to ask questions about lifetime value or at least long term value rather than, 

291
00:23:57,700 --> 00:24:02,600
you know, this individual initial transaction, which I think can be a really mild review 

292
00:24:03,100 --> 00:24:07,900
of of customers. I run I run a load of brands who, effectively do their 

293
00:24:07,900 --> 00:24:12,800
customer acquisition through the channel because they're a subscription brand, and then they get a long term 

294
00:24:12,800 --> 00:24:17,800
value. And I think every retailer, every travel brand, every finance brand, especially 

295
00:24:17,900 --> 00:24:22,600
finance brands, could do really well looking at things in that way. In what way? Can you can you 

296
00:24:22,600 --> 00:24:27,500
elaborate on that more? Yeah. Starting to think, okay. I'm gonna make ten bucks 

297
00:24:27,500 --> 00:24:32,400
on this person right now, and they might cost me five bucks to acquire. But I know 

298
00:24:32,400 --> 00:24:37,000
that this is the kind of customer that's gonna engage with my brand for the next five years, ten years, twenty 

299
00:24:37,000 --> 00:24:41,000
years, and accepting that that is a very different type of customer 

300
00:24:42,100 --> 00:24:47,000
to the other type of customer that legitimately does come in through the affiliate channel, which is, you know, the kind of dust that 

301
00:24:47,000 --> 00:24:51,900
comes in once and then never buys again. That latter form, those are the kind of customers that you wanna 

302
00:24:51,900 --> 00:24:56,700
say, right, is my immediate ROAS on this one sale 

303
00:24:57,300 --> 00:25:01,900
If it's not, it makes zero sense to make the sale. But if you're bringing in a customer with a slightly higher cost to entry that is gonna continue to 

304
00:25:01,900 --> 00:25:06,900
buy from you, maybe from within the affiliate channel, maybe from within the partner ecosystem, or maybe from within the partner ecosystem, or maybe from within the partner ecosystem, or maybe from within the maybe from within the affiliate 

305
00:25:06,900 --> 00:25:11,200
channel, maybe from within the partner ecosystem, or maybe, you know, direct in the future, 

306
00:25:11,900 --> 00:25:16,200
I see a load of value in in attribution models or measurement models, 

307
00:25:16,700 --> 00:25:21,600
that take that into consideration and allow you to chase down the things that the best for your 

308
00:25:21,600 --> 00:25:26,400
business long term. Yep. Okay. Hot hot topic. Is 

309
00:25:26,400 --> 00:25:30,000
affiliate marketing a channel? No. Yes. 

310
00:25:31,500 --> 00:25:36,500
What's your view? What do you think? I like that you're answering it that way. I I probably won't stop calling 

311
00:25:36,500 --> 00:25:40,700
in the channel because it people are kind of they want an easy way to mentally 

312
00:25:41,600 --> 00:25:46,400
conceptualize it vis a vis paid search, paid social, organic. And I think 

313
00:25:46,400 --> 00:25:51,400
that because it can touch all of those things and involves all of those things, it it 

314
00:25:51,400 --> 00:25:56,300
it shouldn't be it's not technically a channel. It sounds like you're in agreement with that. Yeah. I'm 

315
00:25:56,300 --> 00:25:59,600
in full agreement. Should I just leave it a no and we can move on? No. 

316
00:26:01,500 --> 00:26:06,100
I have, Want a debate. I'm about to go on the record being super praiseful 

317
00:26:06,100 --> 00:26:11,100
of my mentor and our head of strategy. And if anybody knows me, 

318
00:26:11,100 --> 00:26:15,900
that's listening, they're gonna know that I'm, I'm pretty British and dry. And I 

319
00:26:15,900 --> 00:26:20,000
don't often go all out and say, yeah, I'm behind the company line on this. But 

320
00:26:20,600 --> 00:26:25,400
my my mentor, certainly since I moved to the US, Summer Arrears is our, head of 

321
00:26:25,400 --> 00:26:30,400
strategy for CJ now. And I think she's absolutely nailed it with the description of what affiliate 

322
00:26:30,400 --> 00:26:34,900
is. She calls it the channel of channels, and you'll probably see in any CJ marketing 

323
00:26:35,400 --> 00:26:40,400
we've done over the past few months and something we're doing going forwards. The channel of channels or the channel 

324
00:26:40,400 --> 00:26:45,400
four channels is is featured within that. And what she means by that is it 

325
00:26:45,400 --> 00:26:50,400
is a structure. Affiliate is a structure which tends to be focused around a 

326
00:26:50,400 --> 00:26:55,400
performance marketing payout structure, but, really, it spans all of 

327
00:26:55,400 --> 00:27:00,300
your channels. And that's not just to say, you know, affiliate can do paid search or affiliate can do 

328
00:27:00,300 --> 00:27:05,100
display. It can. It absolutely can. The reason I said channel of channels and the 

329
00:27:05,100 --> 00:27:10,000
reason she says channel of channels or channel four channels is it enables 

330
00:27:10,100 --> 00:27:14,900
those channels to amplify their own success. If you have a great affiliate program, 

331
00:27:15,500 --> 00:27:20,500
you are able to use that to amplify the results from your paid search channel, to extend the 

332
00:27:20,500 --> 00:27:25,500
results from your paid search channel, extend the reach of it. Same for display. The same for social 

333
00:27:25,500 --> 00:27:30,400
and influencer. Very, very, very much with social influencer. It gives you 

334
00:27:30,400 --> 00:27:35,400
the ability to reach a much wider sort of keywords in paid search because you 

335
00:27:35,400 --> 00:27:40,300
can do that in a guaranteed CPA, a much wider audience of influencers because you don't need to get a one 

336
00:27:40,300 --> 00:27:45,100
on one agreement in place with them in your social channels and have a payment system upfront. You you 

337
00:27:45,100 --> 00:27:50,000
can say, look. This is a CPA structure. This is what we're all used to talk about. Really, when you think about even 

338
00:27:50,000 --> 00:27:54,900
your, you know, your site, your site amplification, your site, conversion, Really, when you think about 

339
00:27:54,900 --> 00:27:59,800
any part of your business, there is a partner within the 

340
00:27:59,800 --> 00:28:04,800
affiliate space that can help to work with them to amplify their own results or replicate 

341
00:28:04,800 --> 00:28:09,800
their results. Love it. Good segue. What are some emerging affiliate 

342
00:28:09,800 --> 00:28:14,600
partners that you're super excited about? I am all in 

343
00:28:14,700 --> 00:28:19,500
on a few different publisher models from a philosophical standpoint I'm seeing coming out at the 

344
00:28:19,500 --> 00:28:24,500
moment. I have no idea whether these are gonna be, you know, the biggest publishing models in 

345
00:28:24,500 --> 00:28:29,500
the world or whether they're not. And if I did, I'd be making a lot of money off of off of that 

346
00:28:29,500 --> 00:28:34,400
ability to predict. But interestingly, as I was thinking about this podcast, these were the people that 

347
00:28:34,400 --> 00:28:39,300
I that I was thinking about that I was thinking about when I was considering 

348
00:28:39,400 --> 00:28:43,200
how beneficial tests and loan budgets and experimentation budgets are in establishing. 

349
00:28:44,100 --> 00:28:49,000
So a few different models I really like at the moment, or a few different 

350
00:28:49,000 --> 00:28:53,600
emerging models that I really like at the moment. It's then a video media or connected 

351
00:28:53,600 --> 00:28:58,400
TV. I'm starting to see a couple of brands within the affiliate 

352
00:28:58,400 --> 00:29:03,300
space who are able to kind of do what I was just describing with paid search, where they say, look. We 

353
00:29:03,300 --> 00:29:08,200
understand that you have your own assets. We understand that you have your own channel. We understand that you're 

354
00:29:08,200 --> 00:29:12,400
not covering every possible customer that you can because you have to do some risk management. 

355
00:29:13,000 --> 00:29:18,000
So let us take those assets or maybe even let us work with you to create assets 

356
00:29:18,300 --> 00:29:23,200
so that we can cover those gaps on the CPA. Love that. Bullish on connected TV 

357
00:29:23,200 --> 00:29:28,100
as well. So exciting because it's worked so well for brands that I've seen 

358
00:29:28,100 --> 00:29:33,100
do it. You know, take the proper proper precautions, but execute it within paid search. 

359
00:29:33,200 --> 00:29:38,000
And I've seen it to some extent with retargeting as well, although it's a little bit less in vogue. I'm 

360
00:29:38,000 --> 00:29:42,900
also seeing some really interesting stuff happening with, browser extensions. I 

361
00:29:42,900 --> 00:29:47,900
know, people can get cagey at that term, but, one, I don't think they 

362
00:29:47,900 --> 00:29:52,500
should be. You know, when we're talking about testing, I think it's, super important to test your hypotheses. 

363
00:29:53,400 --> 00:29:58,200
We can talk about this later on, but I have some, some interesting data on existing browser 

364
00:29:58,200 --> 00:30:03,100
extensions. But what I'm seeing is happening with browser extensions at the moment is we're getting these 

365
00:30:03,100 --> 00:30:08,100
emerging partners who recognize that the benefit of a browser extension is really 

366
00:30:08,100 --> 00:30:12,900
to increase consumer convenience or to serve an immediate consumer need. And they're starting to 

367
00:30:12,900 --> 00:30:16,900
ask the question, okay. What consumer needs exist outside of 

368
00:30:17,600 --> 00:30:22,300
coupons, price comparison, cashback, which really tends to be the browser extension kind of 

369
00:30:22,300 --> 00:30:27,100
mantra right now. So Mhmm. We can name check these publishers. There's a there's a 

370
00:30:27,200 --> 00:30:32,200
an extension called Benny that I'm a huge fan of the publisher model for. And what 

371
00:30:32,200 --> 00:30:37,200
they do is they help they help customers discover pre loved secondhand 

372
00:30:37,700 --> 00:30:42,500
refurbished, items. It's particularly focused within, the 

373
00:30:42,600 --> 00:30:47,500
fashion space, at least for now, as they're browsing a customs 

374
00:30:47,800 --> 00:30:52,500
sorry, a client site. So I know that they're working with Patagonia at the moment, which is wearing a 

375
00:30:52,500 --> 00:30:57,500
Patagonia vest. So, you know, you're looking at this Patagonia vest. Repping the brand. And they flag 

376
00:30:57,500 --> 00:31:02,400
up, hey. Patagonia have their own pre loved program. You could buy this pre loved for 

377
00:31:02,400 --> 00:31:07,300
x amount of money. And that engagement in circular economy, but also that engagement in 

378
00:31:07,300 --> 00:31:12,300
recognizing that consumers have some consumers have needs to save money, but that that need doesn't always need 

379
00:31:12,300 --> 00:31:17,300
to be in the form of a straight up discount on a product, I think, is awesome. Absolutely. 

380
00:31:17,600 --> 00:31:22,500
Very cool. Very interesting stuff. Alright. You're from the UK, came over 

381
00:31:22,500 --> 00:31:27,000
when? Twenty seventeen, twenty eighteen? Twenty eighteen, I think. 

382
00:31:27,100 --> 00:31:32,100
K. What's better, football or rugby? I went to college in 

383
00:31:32,100 --> 00:31:37,100
Wales, so I have to say rugby. Plus there's more, there's more blood, which I think is always 

384
00:31:37,100 --> 00:31:42,100
that's what you're looking for in a sport. Right? Yeah. That's the idea with the whole that whole genre. 

385
00:31:42,100 --> 00:31:46,400
How about, baseball or cricket? What what's better? Come on. Oh, 

386
00:31:47,200 --> 00:31:51,900
If you asked me before I went to an NBA baseball game, I would've said, sorry, 

387
00:31:51,900 --> 00:31:56,100
NBA. Okay. That's my British. 

388
00:31:57,500 --> 00:32:02,300
I would've said cricket. I've seen four baseball games in my life and I've 

389
00:32:02,300 --> 00:32:07,300
loved every one of them. And I've seen five cricket games live in my life, and I've hated every minute 

390
00:32:07,300 --> 00:32:12,200
of those experiences. Maybe we need to cut that bit out. So I think baseball might 

391
00:32:12,200 --> 00:32:17,200
be winning me over. It might be the best. That's the most insightful. That's very 

392
00:32:17,300 --> 00:32:22,100
that's a hey. You know, a little something for the British audience, a little something for the American 

393
00:32:22,100 --> 00:32:27,100
audience. Everybody's happy. You nailed it. We talked about emerging partners. You're talking about 

394
00:32:27,100 --> 00:32:32,000
testing, talking about best in class programs. You have a little sneak 

395
00:32:32,000 --> 00:32:37,000
peek of what's coming. Behind the curtain, you're working on some really interesting data around, you 

396
00:32:37,000 --> 00:32:42,000
know, how incremental are some of these browser extensions and plug ins. So, Tom, I'm I'm super 

397
00:32:42,000 --> 00:32:46,200
excited about all the stuff you you're thinking about around browser extensions, 

398
00:32:47,000 --> 00:32:51,200
toolbars coming from that place of testing, learning, 

399
00:32:51,700 --> 00:32:56,700
objectivity. There's obviously a lot of controversy, and and people kind of rile 

400
00:32:56,700 --> 00:33:01,600
at the the thought of them. How are you thinking about them? How are you approaching 

401
00:33:01,600 --> 00:33:06,600
it? What are some of the learnings that you're working on on that topic? Yeah. That definitely controversial. 

402
00:33:06,800 --> 00:33:11,700
I came into the industry 10:12 years ago, which were all you you 

403
00:33:11,700 --> 00:33:16,700
considered my entry, and I think it was right when browser extensions were just starting to enter the 

404
00:33:16,700 --> 00:33:21,700
market. And so I've been there from the beginning when I was and still 

405
00:33:21,700 --> 00:33:26,600
remain to some extent a massive skeptic of of anything, that comes into the market that was 

406
00:33:26,600 --> 00:33:31,300
to be engaging with the customer that's already deep in the sales process. And 

407
00:33:31,500 --> 00:33:35,800
for at least seven years, I've been having, 

408
00:33:36,900 --> 00:33:41,800
head on conversations with clients about how we test that, about how we measure 

409
00:33:41,800 --> 00:33:46,800
that. I spoke earlier on about kind of overcoming biases, and I 

410
00:33:46,800 --> 00:33:51,700
would I have been challenging clients to 

411
00:33:51,700 --> 00:33:56,700
test their biases, to test their theories about browser extensions. I hear a lot of people using a 

412
00:33:56,700 --> 00:34:01,700
common sense standpoint where they'll say this person's already on my site, and so they're gonna buy anyway. And 

413
00:34:01,700 --> 00:34:06,500
I've been desperate to test that over time. I legitimately think we've just hit on a 

414
00:34:06,500 --> 00:34:11,500
methodology that allows us to do that. So we're this isn't public public 

415
00:34:11,500 --> 00:34:16,500
yet, but we can talk about some preliminary results on the podcast. We identified a tech 

416
00:34:16,500 --> 00:34:21,100
partnership with somebody who sits on a large number of client sites. 

417
00:34:21,400 --> 00:34:25,900
We have, I think, about seventy million journeys that we've looked at, who can 

418
00:34:26,500 --> 00:34:31,500
see when a browser extension fires. They can see 

419
00:34:31,500 --> 00:34:36,500
if somebody has a browser extension installed, and they can 

420
00:34:36,500 --> 00:34:41,500
determine the difference in behavior between people who have an 

421
00:34:41,500 --> 00:34:46,400
extension installed and did not see a message and people who have an extension install 

422
00:34:46,500 --> 00:34:51,400
installed and did see a message. We might need to cut this bit out or follow-up on this afterwards, 

423
00:34:51,400 --> 00:34:56,200
but, that is while also determining whether 

424
00:34:56,200 --> 00:35:01,100
that message was proactive or reactive. So they're able to say, look. This person would not have 

425
00:35:01,100 --> 00:35:05,400
received a browser extension message unless the extension chose to fire. 

426
00:35:05,700 --> 00:35:10,400
And when we look at that en masse, when we look at that across seventy million journeys, we see 

427
00:35:10,400 --> 00:35:15,100
a significant uplift in revenue per session from customers who received 

428
00:35:15,400 --> 00:35:20,000
a message from a browser extension versus those that didn't. So Interesting. 

429
00:35:20,300 --> 00:35:25,200
Yeah. When we when we're saying, I think this person was on my site, so they were gonna buy anyway, 

430
00:35:25,400 --> 00:35:30,400
my my knee jerk response to that was always site conversion rates are about three to four percent on a great day, 

431
00:35:30,400 --> 00:35:35,300
so ninety six percent of them weren't gonna convert anyway. And what we're seeing from this data is the 

432
00:35:35,300 --> 00:35:40,000
extensions is the people who interact with your customers who are already deep in journey are able 

433
00:35:40,000 --> 00:35:44,400
to significantly uplift revenue for the for the customers that they interact with. 

434
00:35:45,000 --> 00:35:49,600
And really interestingly, on the flip side, we were able to see how Amazon Assistant 

435
00:35:49,600 --> 00:35:54,200
affects people, which purely exist to drag people away from your site into Amazon. 

436
00:35:54,700 --> 00:35:59,700
And it shows the inverse in the way that makes the most possible sense. If Amazon's 

437
00:35:59,700 --> 00:36:04,600
assistant says, hey. You shouldn't buy from these people. You should buy elsewhere. We saw a massive 

438
00:36:04,700 --> 00:36:09,400
decline in revenue per journey because we're seeing the exact inverse happen. The overall 

439
00:36:09,400 --> 00:36:14,200
indication was consumers seem to listen and respond to messages from browser 

440
00:36:14,200 --> 00:36:18,800
extensions in a way that genuinely, genuinely surprised me. Like, I went into this 

441
00:36:19,200 --> 00:36:23,800
without bias. I went into this wanting to test hypotheses. Obviously, from a network 

442
00:36:23,800 --> 00:36:28,600
perspective, I hope that it gave us the, you know, the answer that, yeah, what we've been doing for 

443
00:36:28,600 --> 00:36:33,500
05:10 years isn't isn't completely without worth. But I think the extent to which 

444
00:36:33,500 --> 00:36:38,500
we saw we saw an uplift, and we'll have a white paper coming out at some point in the near future that 

445
00:36:38,500 --> 00:36:43,200
really delves into this was surprising to absolutely everybody involved. I love it. 

446
00:36:43,200 --> 00:36:47,800
Excited to read it. Excited to dig into it more. I think it might be, time for 

447
00:36:47,800 --> 00:36:52,700
a a mini part three just to delve into that topic on itself even who 

448
00:36:52,700 --> 00:36:57,700
knows? Maybe with white paper in hand. Yeah. But it's been a pleasure. Loved loved chatting 

449
00:36:57,700 --> 00:37:02,700
about all these topics with you, Tom, and really appreciate your time today. It's been awesome. Yep. It's 

450
00:37:02,700 --> 00:37:05,300
been fantastic. I love doing it. Thanks, man.