EP250 - AI - Helpful or Headache in PPC

1
00:00:00,000 --> 00:00:02,800
Hello, and welcome to this
week's episode of the PPC chat

2
00:00:02,900 --> 00:00:06,200
Roundup in today's episode.
We have the awesome Julie bikini

3
00:00:06,200 --> 00:00:11,700
taking us on a topic of AI
helpful or headache in PPC.

4
00:00:12,100 --> 00:00:15,400
Now, I feel that this was not
quite pre-planned because over

5
00:00:15,400 --> 00:00:19,200
the weekend, search engine land
released an article saying that

6
00:00:19,500 --> 00:00:24,400
Google is now planning a new
search engine while working on

7
00:00:24,400 --> 00:00:27,000
new search features under the
project.

8
00:00:27,000 --> 00:00:31,100
Maggie's that M AG.
I so that Means like they are in

9
00:00:31,100 --> 00:00:34,700
the process of building an all
new search engine powered by its

10
00:00:34,700 --> 00:00:38,500
new AI technology.
And yeah, the way I actually

11
00:00:38,500 --> 00:00:42,200
heard about it was Julia
herself, writing a reaction

12
00:00:42,200 --> 00:00:45,000
articles.
It wasn't best pleased by this

13
00:00:45,000 --> 00:00:46,300
news.
Because yeah.

14
00:00:46,300 --> 00:00:48,600
And I even put a poll about it
today.

15
00:00:48,600 --> 00:00:52,600
Talking about is Google really
haven't their priorities, right?

16
00:00:52,600 --> 00:00:55,700
Is this where they really should
be focusing their resources on

17
00:00:55,700 --> 00:00:58,800
if you want to, you know, get
involved in that poll, please

18
00:00:58,800 --> 00:01:01,400
like let us know.
Let's just follow this, go on

19
00:01:01,400 --> 00:01:03,800
LinkedIn.
Look for the marketing, a new or

20
00:01:03,800 --> 00:01:06,500
an you addict bowl, and you will
find that pole.

21
00:01:06,700 --> 00:01:08,900
But yeah, I just found it
interesting that.

22
00:01:08,900 --> 00:01:11,100
Yeah, Google is coming out with
this.

23
00:01:11,100 --> 00:01:15,200
There's going to be an AI
powered search engine in a way,

24
00:01:15,200 --> 00:01:17,300
you know you just have to be
ready for change.

25
00:01:17,300 --> 00:01:20,200
It's one of those things that we
say many a time but like

26
00:01:20,200 --> 00:01:23,400
sometimes it's just a bit
worrisome as to whether they

27
00:01:23,400 --> 00:01:27,400
actually creating things to be
helpful or just create a

28
00:01:27,400 --> 00:01:30,800
headache for most advertisers
advertisers And paid search.

29
00:01:30,800 --> 00:01:34,300
And will they actually listened
to advertisers and paid search

30
00:01:34,300 --> 00:01:37,000
when they're releasing this,
when they're building it when

31
00:01:37,000 --> 00:01:39,700
they're creating it?
So that what they do create is

32
00:01:39,700 --> 00:01:42,100
something that can be effective
for us.

33
00:01:42,100 --> 00:01:44,500
Anyway, we chat about what we
think about it.

34
00:01:44,500 --> 00:01:47,300
What we think about this update
and yeah, well, we think about

35
00:01:47,300 --> 00:01:50,500
Ai and the future of it.
In PPC, hope you enjoy.

36
00:01:59,600 --> 00:02:02,600
Hello and a warm, welcome to
this episode of the PPC chat

37
00:02:02,600 --> 00:02:05,400
Roundup a podcast where I round
up the Twitter discussion that

38
00:02:05,400 --> 00:02:08,000
happens on Tuesdays at 5 p.m.
GMT.

39
00:02:08,400 --> 00:02:10,900
I'm your host, Andrew, and I'm
recording from my studio Hair

40
00:02:10,900 --> 00:02:14,000
London, UK, and I use this
platform to share, not just

41
00:02:14,000 --> 00:02:17,900
expert, but also my ideas and
considered best practices about

42
00:02:17,900 --> 00:02:20,800
paid media and the direction,
the digital industry is going

43
00:02:20,800 --> 00:02:22,400
in.
So if you want to keep up to

44
00:02:22,400 --> 00:02:25,500
date with my tips and Rick's in
the industry and get the latest

45
00:02:25,500 --> 00:02:28,500
on the podcast that has been
called your go-to if you don't

46
00:02:28,500 --> 00:02:31,500
make it to the PPC chat live
discussions and even if you do

47
00:02:31,500 --> 00:02:34,500
then like, follow share and
retweet on that, right?

48
00:02:34,500 --> 00:02:37,800
Twitter with my handle, the
marketing, a new or join us on

49
00:02:37,800 --> 00:02:39,900
our LinkedIn group.
PPC, chat Roundup.

50
00:02:39,900 --> 00:02:44,100
Or if you're an Instagram, check
us out at PPC, chat Round Up to

51
00:02:44,100 --> 00:02:47,500
get quotes, and Snippets from
our previous episodes.

52
00:02:47,700 --> 00:02:52,200
So, yeah, we have Julie taken us
on a very light timely topic

53
00:02:52,200 --> 00:02:55,800
about, you know, talking, In
about not just the latest on

54
00:02:55,800 --> 00:02:58,000
terms of the tips and tricks of
what's going on.

55
00:02:58,000 --> 00:03:03,100
But about how the future of paid
search how its evolving, we're

56
00:03:03,100 --> 00:03:06,400
really in the states of where
best practices are going to be

57
00:03:06,400 --> 00:03:09,800
really very much evolving and
changing over time.

58
00:03:09,800 --> 00:03:13,500
So I really hope we are keeping
up with that before we get into

59
00:03:13,500 --> 00:03:16,300
the questions and answers, I'm
going to get right into it

60
00:03:16,300 --> 00:03:18,400
today.
But I really like to mention

61
00:03:18,400 --> 00:03:21,700
this PSA saying that this chat
is intended for PPC,

62
00:03:21,700 --> 00:03:25,100
Professionals of all levels, if
you Something you want to share,

63
00:03:25,300 --> 00:03:26,900
please.
Do we are a friendly and

64
00:03:26,900 --> 00:03:29,600
supportive Community.
Please don't be afraid to

65
00:03:29,600 --> 00:03:32,300
participate.
Lurkers are of course always

66
00:03:32,300 --> 00:03:36,300
welcome to and lurkers and just
the listeners of the podcast.

67
00:03:36,300 --> 00:03:40,200
Welcome to this episode.
Thank you for joining us today.

68
00:03:40,200 --> 00:03:43,800
So let's get on to question one
at so Julie goes yeah let's get

69
00:03:43,800 --> 00:03:45,500
to it.
I'm really curious went what

70
00:03:45,500 --> 00:03:49,300
everybody where everyone's
thinking is on today's topic?

71
00:03:49,500 --> 00:03:53,400
She asks, how do you define a i
when it comes to peace.

72
00:03:53,500 --> 00:03:56,700
See Travis starts off saying the
distinction between Ai and

73
00:03:56,708 --> 00:04:00,200
machine learning is important,
AI autonomous human.

74
00:04:00,200 --> 00:04:03,200
Mimicking decision making
machine learning performing,

75
00:04:03,200 --> 00:04:05,500
specific tasks with accurate
results.

76
00:04:05,500 --> 00:04:10,500
Based on inputs learning data
and pattern recognition to data.

77
00:04:10,500 --> 00:04:13,200
Google ads has not fully rolled
out a.

78
00:04:13,200 --> 00:04:16,000
I Gilda response.
A machine learning is now

79
00:04:16,000 --> 00:04:20,500
ubiquitous throughout Google ads
and analytics LL Em's,

80
00:04:20,500 --> 00:04:25,300
generating possible ad copy,
Matching bidding and lately,

81
00:04:25,300 --> 00:04:29,500
even the whole cookie-less
browser tracking models, at play

82
00:04:29,800 --> 00:04:33,400
Robert, Brady answers this.
And I think this was the perfect

83
00:04:33,500 --> 00:04:36,300
answer and I just was like this
will be my answer.

84
00:04:36,500 --> 00:04:38,100
Really, for this question
Herbert.

85
00:04:38,100 --> 00:04:41,600
Brady goes, I don't think we
have true AI yet in PPC.

86
00:04:41,600 --> 00:04:45,000
I believe we have machine
learning in lots of places, like

87
00:04:45,000 --> 00:04:48,700
bidding, keyword, matching
asset, creation, Etc, and many

88
00:04:48,700 --> 00:04:53,000
confuse that with a i Amanda
Folly goes 100% And I pretty

89
00:04:53,000 --> 00:04:54,500
much yeah.
Yeah, as I said, agree with

90
00:04:54,500 --> 00:04:56,600
that.
I was like there is probably

91
00:04:56,600 --> 00:04:59,800
some AI driving.
The machine learning fact.

92
00:04:59,800 --> 00:05:02,900
It's a, I driving the machine
learning that we are using, but

93
00:05:02,900 --> 00:05:06,300
we aren't using AI yet.
In my opinion, Julie goes.

94
00:05:06,300 --> 00:05:07,800
Yeah, echoing, Robert Brady as
well.

95
00:05:07,800 --> 00:05:10,600
A lot of things are being called
a, I right now and they are

96
00:05:10,600 --> 00:05:13,100
rapidly getting integrated into
platforms and tools.

97
00:05:13,100 --> 00:05:16,100
I don't think the level of the
tech is actually all that

98
00:05:16,100 --> 00:05:20,300
sophisticated at this time
either, despite all the hype

99
00:05:20,300 --> 00:05:24,700
stating otherwise Julia vices
while say, It's not actually a,

100
00:05:24,700 --> 00:05:27,300
I, at this point more like
strong machine learning.

101
00:05:27,600 --> 00:05:31,400
I said, creation optimization
search or interest matching.

102
00:05:31,400 --> 00:05:35,600
Basically, if your e-comm, it's
a very powerful work back from

103
00:05:35,600 --> 00:05:40,100
bottom of funnel to top.
Cool, but not what Asimov was

104
00:05:40,100 --> 00:05:43,900
thinking of whose Azimuth feel
like he's one of like the big

105
00:05:43,900 --> 00:05:47,200
thinkers of marketing, right?
And so Julie takes us on to

106
00:05:47,200 --> 00:05:50,800
question 2 saying, what aspect
of a I do think are helpful for

107
00:05:50,800 --> 00:05:54,700
PPC, starting off.
We have Travis as well saying so

108
00:05:54,700 --> 00:05:57,800
far I've leveraged chat DPT for
the following key word.

109
00:05:57,800 --> 00:06:01,700
Discovery keyword grouping into
themes at copy variations,

110
00:06:01,900 --> 00:06:07,300
directionally sqrl events
ranking GMC, title description,

111
00:06:07,300 --> 00:06:11,200
and augmenting plus description,
augmenting even custom audience,

112
00:06:11,200 --> 00:06:15,800
segments tearing landing page
content, ideation missile,

113
00:06:15,800 --> 00:06:17,600
respond saying, I enjoyed our
chat.

114
00:06:17,600 --> 00:06:21,400
Gbt supports me to write ads and
extensions when I'm trying to

115
00:06:21,400 --> 00:06:24,900
wear things differently.
I get some Buzz words and

116
00:06:24,900 --> 00:06:27,700
sometimes flip a sentence
around, which really helps.

117
00:06:27,700 --> 00:06:30,900
I definitely wouldn't say it can
write it yet.

118
00:06:30,900 --> 00:06:34,000
Mmm, Gill gonna respond saying
as digital marketers with shot

119
00:06:34,000 --> 00:06:36,800
ourselves in the foot, over the
past couple of decades, by over

120
00:06:36,800 --> 00:06:41,000
emphasis in trackability, I'm
pretty sure we'd be served.

121
00:06:41,000 --> 00:06:45,100
Well by taking a more heuristic
approach to data and there's a

122
00:06:45,100 --> 00:06:50,200
possibility that a, I could help
us make better decisions.

123
00:06:50,300 --> 00:06:53,600
Yeah, it could make help us make
Beth better decisions but Will

124
00:06:53,600 --> 00:06:55,500
it Ashton Clark response,
question once.

125
00:06:55,500 --> 00:06:58,000
And for starters, I think of a
is different from machine

126
00:06:58,000 --> 00:07:00,000
learning.
Ml is used by the platform

127
00:07:00,000 --> 00:07:03,700
algorithms to help serve ads
better where us a I can be used

128
00:07:03,700 --> 00:07:07,800
as a complementary tool to make
our jobs easier like drafting ad

129
00:07:07,800 --> 00:07:10,300
copy or automating a manual
process.

130
00:07:10,400 --> 00:07:11,900
I don't know, I wouldn't
describe it as that

131
00:07:11,900 --> 00:07:13,800
passionately.
Julie response question to say

132
00:07:13,800 --> 00:07:16,900
and I think the most helpful AI
capabilities at the moment are

133
00:07:16,900 --> 00:07:21,200
in idea generation and iteration
for keywords or ad copy,

134
00:07:21,200 --> 00:07:23,400
particularly it's interesting to
play around.

135
00:07:23,600 --> 00:07:29,200
With four competitors, to to see
how content differs do you have

136
00:07:29,200 --> 00:07:31,500
ice responds?
And gives two answers is, it

137
00:07:31,500 --> 00:07:35,400
could be great for new search
interest, Behavior, mapping.

138
00:07:35,400 --> 00:07:37,600
And of course, the forever,
elusive dream.

139
00:07:37,600 --> 00:07:40,600
Cross-channel attribution a girl
can dream, right?

140
00:07:40,600 --> 00:07:42,200
It's continuous while I'm
daydreaming.

141
00:07:42,200 --> 00:07:45,600
Imagine mapping out a new
feature based on reviews and

142
00:07:45,600 --> 00:07:48,300
correlating it with analytics
Behavior.

143
00:07:48,700 --> 00:07:51,300
Lord, imagine that.
That would be perfect Julie

144
00:07:51,300 --> 00:07:53,400
goes, I think about capabilities
like you're describing.

145
00:07:53,400 --> 00:07:56,700
Scribing and then I remember all
the retailers, who collect

146
00:07:56,700 --> 00:08:01,500
loyalty card data and do exactly
squat with it and come back down

147
00:08:01,500 --> 00:08:03,900
to reality, right?
And you never I says.

148
00:08:03,900 --> 00:08:05,900
Yeah, right.
But what if there was an

149
00:08:05,900 --> 00:08:08,900
affordable machine, a girl can
dream.

150
00:08:09,200 --> 00:08:12,500
She goes it also it can help
keep our eyes on the prize to I

151
00:08:12,500 --> 00:08:16,100
recall a load of work, put into
an app with a feature, I cared

152
00:08:16,100 --> 00:08:20,300
about and believed in our
audience, did not care humbling,

153
00:08:20,600 --> 00:08:23,100
but an important check on the
marketing.

154
00:08:23,600 --> 00:08:25,900
Yeah.
Sometimes we think it's the most

155
00:08:25,900 --> 00:08:29,800
important thing but then when we
actually put it to Market, it

156
00:08:29,800 --> 00:08:33,700
may not be Lauren's Chase
response question one saying,

157
00:08:33,700 --> 00:08:35,400
agree with what others have
said.

158
00:08:35,400 --> 00:08:37,200
We don't really have a i
marketing yet.

159
00:08:37,200 --> 00:08:39,799
We just have machine learning
which is very different.

160
00:08:39,900 --> 00:08:41,700
Robert, Brady goes to question
2.

161
00:08:41,700 --> 00:08:45,100
Saying, I really like how the
generative language models can

162
00:08:45,100 --> 00:08:47,100
shift their voice with a simple
prompt.

163
00:08:47,200 --> 00:08:51,400
That can give you more breadth
of ideas and tones Lawrence,

164
00:08:51,400 --> 00:08:53,100
chase down response to question
to say.

165
00:08:53,100 --> 00:08:57,000
And I think I think is where it
can get exciting, finding new

166
00:08:57,000 --> 00:09:00,600
markets ideas and anything we
have not thought of being able

167
00:09:00,600 --> 00:09:04,000
to look at data and find Trends
or opportunities.

168
00:09:04,000 --> 00:09:08,400
We may miss ideas for everything
from copy video, imagery

169
00:09:08,400 --> 00:09:12,700
excetera and you lavage response
to a question to what, more in a

170
00:09:12,700 --> 00:09:15,500
response to gild Nowhere.
She says gilgul, that makes a

171
00:09:15,508 --> 00:09:19,100
great point and reminds me fully
automated testing.

172
00:09:19,400 --> 00:09:22,100
Oh God, please enable this
immediately and still, let me

173
00:09:22,100 --> 00:09:25,600
see the findings of Solutely.
Yeah, if they can be automated

174
00:09:25,600 --> 00:09:29,100
testing.
There's actually reliable love

175
00:09:29,100 --> 00:09:32,200
that doing doing a/b testing and
all that kind of tests and see

176
00:09:32,200 --> 00:09:35,700
what works and trying to figure
out at what point in time that

177
00:09:35,700 --> 00:09:39,300
you've got significant data,
statistically significant data

178
00:09:39,300 --> 00:09:41,900
can be time.
Consuming Miller response to

179
00:09:41,900 --> 00:09:45,100
question 3.
Oh actually we go to question 3

180
00:09:45,100 --> 00:09:47,000
even I'm not sure if I've given
the question.

181
00:09:47,000 --> 00:09:50,600
First of all, Julie asked what
aspect of a I do you think are a

182
00:09:50,600 --> 00:09:53,900
headache or difficult for PPC
middle Response.

183
00:09:53,900 --> 00:09:56,300
And I think the biggest
challenge is that people think

184
00:09:56,300 --> 00:09:58,800
it's groundbreaking and going to
change everything.

185
00:09:59,000 --> 00:10:02,500
Make us job, less excetera were
a long way off that people

186
00:10:02,500 --> 00:10:06,000
sharing findings within our
community are great and really

187
00:10:06,000 --> 00:10:08,400
supports us all to learn and
grow.

188
00:10:08,600 --> 00:10:11,300
Julie goes on till it knows the
difference reliably between

189
00:10:11,300 --> 00:10:12,900
chocolate milk and milk
chocolate.

190
00:10:12,900 --> 00:10:16,100
I am not terribly worried.
Yeah, that was the first first

191
00:10:16,100 --> 00:10:19,500
things that I mentioned that we
mentioned about when close

192
00:10:19,500 --> 00:10:23,000
variance when exact match was no
longer exact and close variants

193
00:10:23,000 --> 00:10:26,500
with Was brought in because
basically chocolate milk and

194
00:10:26,500 --> 00:10:28,700
milk chocolate will be seen as
the same thing.

195
00:10:28,900 --> 00:10:31,500
Anyway, we were all worried
about that middle then response

196
00:10:31,500 --> 00:10:34,000
to what to do.
Lee said says I mean we can go

197
00:10:34,000 --> 00:10:36,800
into Google's machine learning
for broad match but I don't have

198
00:10:36,800 --> 00:10:40,000
the brain capacity to even
mention that without getting

199
00:10:40,200 --> 00:10:42,600
rage Lord, let's leave that
alone.

200
00:10:42,600 --> 00:10:45,200
Then, Travis response to
question 3 seeing a common theme

201
00:10:45,200 --> 00:10:47,200
is, is this going to take my
job?

202
00:10:47,200 --> 00:10:51,400
The short answer is not yet.
So far those tools save time and

203
00:10:51,400 --> 00:10:53,300
mental bandwidth.
If understood.

204
00:10:53,400 --> 00:10:55,700
The headache is learning how to
get them to work.

205
00:10:55,700 --> 00:10:59,500
Well for you, that's a unique
skill, set plus learning curve

206
00:10:59,500 --> 00:11:02,100
for all, Gilman response to
question 3.

207
00:11:02,100 --> 00:11:06,300
Saying it's not perfect, yet, in
many cases generated content

208
00:11:06,300 --> 00:11:09,600
isn't something that's just plug
and play, it needs to be

209
00:11:09,600 --> 00:11:12,300
shepherded.
John Keegan joins us response to

210
00:11:12,300 --> 00:11:15,300
question one saying, as of right
now of fun buzzword to get

211
00:11:15,300 --> 00:11:17,200
everyone excited about
something.

212
00:11:17,200 --> 00:11:19,600
That's probably going to cause
you more of a headache than

213
00:11:19,600 --> 00:11:22,800
anything else in the question to
question, two, who responds

214
00:11:22,800 --> 00:11:26,100
saying that Used for suggestions
for ad copy or keywords and

215
00:11:26,100 --> 00:11:29,300
certain bit strategies, Julia,
response to question 3.

216
00:11:29,300 --> 00:11:32,900
Saying, the way it's applied,
I'm not interested in a, take

217
00:11:32,900 --> 00:11:35,300
your word for it approached,
show me the stuff.

218
00:11:35,300 --> 00:11:38,800
Be on the team with me deploy
this powerful tool in a cooler

219
00:11:38,800 --> 00:11:40,500
way.
Be cool, I guess.

220
00:11:40,500 --> 00:11:42,800
Is the ask be called.
Lawrence Chase response to

221
00:11:42,808 --> 00:11:45,900
question, 3 saying honestly
whoever programs the AI is going

222
00:11:45,900 --> 00:11:49,100
to have an influence on what or
how it does, what it does.

223
00:11:49,100 --> 00:11:53,200
Absolutely this is both good and
bad companies have biases and

224
00:11:53,400 --> 00:11:57,400
This may sway what results are
presented to Searchers versus

225
00:11:57,400 --> 00:11:59,900
what is best for the Searcher
at.

226
00:11:59,900 --> 00:12:04,000
This is such a fair point
because people who are very

227
00:12:04,000 --> 00:12:07,900
imperfect people that might
have, you know, bias has two

228
00:12:07,900 --> 00:12:10,900
different, you know, skin
colors, to different

229
00:12:10,900 --> 00:12:15,200
sexualities, two different
genders, two different things

230
00:12:15,200 --> 00:12:18,800
you have got to be careful about
how you show those biases and

231
00:12:18,800 --> 00:12:23,300
how you, you know, assume tone
of voice is in.

232
00:12:23,400 --> 00:12:28,300
Of a man or woman or non-binary
person, you know, we've got to

233
00:12:28,300 --> 00:12:30,800
be really careful.
Ashton cock response to question

234
00:12:30,800 --> 00:12:34,300
to say and I think that
generative a tie tools will

235
00:12:34,300 --> 00:12:36,200
continue to be the most useful
for PPC.

236
00:12:36,200 --> 00:12:40,200
Especially for time-consuming
tasks like ad copy, keyword

237
00:12:40,200 --> 00:12:42,600
audience, research and
competitor analysis.

238
00:12:42,900 --> 00:12:47,600
Other helpful use cases, maybe
spreadsheet or data cleanups and

239
00:12:47,600 --> 00:12:51,500
automating internal processes.
Julia herself response.

240
00:12:51,500 --> 00:12:55,000
Question 3, saying does hearing
about it, Breathlessly reported

241
00:12:55,000 --> 00:12:59,800
Ad nauseam count more seriously
hearing about it being deployed

242
00:12:59,800 --> 00:13:03,600
after the fact is a pain like
Google is doing right now with

243
00:13:03,600 --> 00:13:05,700
all their it would totally doing
it.

244
00:13:05,700 --> 00:13:08,200
I to announcement every other
year.

245
00:13:08,200 --> 00:13:10,900
It's almost like everybody else
is doing a and they're like, oh

246
00:13:10,900 --> 00:13:12,600
yeah, don't worry.
We're still doing it as well.

247
00:13:12,800 --> 00:13:15,500
And universe girls, you had the
comes on this Dogpile are an

248
00:13:15,500 --> 00:13:17,700
absolute mess.
Julie goes, I did write a very

249
00:13:17,700 --> 00:13:22,000
salty blog post, so if you go to
Neptune Moon dotnet you'll find

250
00:13:22,000 --> 00:13:26,600
that post about Her opinion as
to how they announce this Robert

251
00:13:26,600 --> 00:13:29,200
Brady their response to question
3 saying the biggest difficulty

252
00:13:29,200 --> 00:13:32,600
will be managing expectations.
People believe AI is a magic

253
00:13:32,600 --> 00:13:35,800
wand it can do anything and do
it better than people.

254
00:13:35,800 --> 00:13:41,200
The reality is much more mundane
and that leads to disappointment

255
00:13:41,200 --> 00:13:44,600
and the rubber Brady also agrees
with Lawrence's points as I did,

256
00:13:44,600 --> 00:13:47,700
where Lawrence had said honestly
whoever programs the AI is going

257
00:13:47,700 --> 00:13:50,900
to have an influence on water or
how it does what it does.

258
00:13:51,200 --> 00:13:52,500
And Robert goes great Point
here.

259
00:13:52,500 --> 00:13:55,800
If you can't see See, I would
how its put together or at least

260
00:13:55,800 --> 00:13:58,500
see, more of How It's coming to
the results.

261
00:13:58,500 --> 00:14:02,200
Then there is a lot of potential
for manipulation whether

262
00:14:02,200 --> 00:14:06,100
intentional or not.
So it might be just as I say

263
00:14:06,100 --> 00:14:09,900
unconscious bias.
But yeah, those would need to be

264
00:14:09,900 --> 00:14:13,100
dealt with John Kagan response
to question, 3 saying the

265
00:14:13,100 --> 00:14:17,600
misconception that one a I can
replace humans unassisted and to

266
00:14:17,800 --> 00:14:20,300
that AI isn't just a giant
headache.

267
00:14:20,300 --> 00:14:24,700
So yeah, giant headache is from
from Taken, I think lock

268
00:14:24,700 --> 00:14:25,600
response to question.
Three.

269
00:14:25,600 --> 00:14:30,200
Same one, biases in the data
that the AI is trained with to

270
00:14:30,200 --> 00:14:33,800
plagiarism and copyright issues.
Especially for a i generated

271
00:14:33,800 --> 00:14:38,000
creative three, combating
misinformation and false claims

272
00:14:38,000 --> 00:14:42,200
for blander and less interesting
content for second, I forgot

273
00:14:42,200 --> 00:14:46,400
what the word blander meant.
So Bland as and more boring,

274
00:14:46,500 --> 00:14:49,400
Julie then goes on to question
for saying, what do you think is

275
00:14:49,400 --> 00:14:53,700
most exciting and or potentially
the most useful aspect of a AI

276
00:14:53,700 --> 00:14:57,000
for PPC.
So, so what do you think, is

277
00:14:57,000 --> 00:15:01,800
most exciting and or potentially
the most useful aspect of AI for

278
00:15:01,800 --> 00:15:04,500
PPC missile response saying, I'm
a freelancer.

279
00:15:04,500 --> 00:15:06,900
So I rarely have people to
bounce ideas off.

280
00:15:07,300 --> 00:15:10,900
I use chat GPT to do that, and
obviously, I'm loving the add

281
00:15:10,900 --> 00:15:15,100
ideas support from it to
overall, a great little helper

282
00:15:15,100 --> 00:15:18,100
doesn't beat as PPC brain at
times though.

283
00:15:18,600 --> 00:15:21,200
Travis response thing for me the
prospect of automating the

284
00:15:21,200 --> 00:15:24,900
mundane or time-consuming task.
It's quite exciting this holds

285
00:15:24,900 --> 00:15:28,500
up the potential for better
deeper thinking more time in my

286
00:15:28,500 --> 00:15:32,700
day or shorter working days.
All Stellar mental health boost.

287
00:15:32,700 --> 00:15:33,800
Absolutely.
I agree.

288
00:15:33,800 --> 00:15:37,700
So even if it's literally less
working time and spending my

289
00:15:37,700 --> 00:15:42,200
brain doing thinking of more fun
intrapreneurial ideas, I love

290
00:15:42,200 --> 00:15:45,400
that girl gonna respond saying
by far LMS.

291
00:15:45,600 --> 00:15:49,200
We bought a premium seat of chat
GPT for the agency and it's been

292
00:15:49,200 --> 00:15:51,500
really useful.
And even though I was very

293
00:15:51,500 --> 00:15:55,000
disappointed in bards release.
It turns out bad was

294
00:15:55,000 --> 00:15:58,500
intentionally, pretty limited.
I have hopes that they will

295
00:15:58,500 --> 00:16:02,400
release a newer version soon.
So they released something, but

296
00:16:02,400 --> 00:16:05,900
it wasn't that very much
competitive.

297
00:16:05,900 --> 00:16:08,000
Hmm, interesting.
Julia herself response to

298
00:16:08,000 --> 00:16:10,400
question, 14, the platform can
get AI.

299
00:16:10,400 --> 00:16:13,100
Or machine learning to marry up
interests, and queries and

300
00:16:13,100 --> 00:16:17,300
produce High converting ad copy.
I'd be quite excited about that,

301
00:16:17,300 --> 00:16:21,400
but it would also do it for
anyone who asks, for the danger

302
00:16:21,400 --> 00:16:24,000
of all ads being the same way.
Be real.

303
00:16:24,200 --> 00:16:25,800
Yep.
Jellyfish respondents said

304
00:16:25,800 --> 00:16:30,900
automated testing automated data
Vis in reporting major potential

305
00:16:30,900 --> 00:16:34,500
for automated landing page
creation, audience curation and

306
00:16:34,500 --> 00:16:39,400
you know, a partridge in a pear
tree, love Julia.

307
00:16:39,600 --> 00:16:43,100
She just comes out with some
phrases that you like they're

308
00:16:43,100 --> 00:16:46,700
like, wait why?
Ah, and go response to that same

309
00:16:46,700 --> 00:16:49,200
partridge in a pear.
Tree must be one of those

310
00:16:49,200 --> 00:16:53,100
bizarrely Canadian terms like
Weevil in my syrup.

311
00:16:53,300 --> 00:16:57,300
No, we use that phrase in the UK
as well, but it's more during

312
00:16:57,300 --> 00:16:58,900
Christmas time.
Yeah, exactly.

313
00:16:58,900 --> 00:17:01,100
Deliver ice goes yet an old
Christmas Carol.

314
00:17:01,300 --> 00:17:04,500
I know if it's not shopping
related, you Americans.

315
00:17:04,500 --> 00:17:07,900
Can't understand where it's got
a good Rhythm.

316
00:17:09,400 --> 00:17:13,000
Also and Go-Go's we use the same
tune but our lyrics changed to

317
00:17:13,000 --> 00:17:17,800
pickups in a liquor store locked
pick up in a liquor store lot.

318
00:17:17,800 --> 00:17:19,599
Like let's just move on from
that.

319
00:17:19,599 --> 00:17:23,200
I think that Cai iced in
response to a question.

320
00:17:23,300 --> 00:17:26,599
Same kind of mentioned this in a
two as well, but I think that

321
00:17:26,599 --> 00:17:29,800
generative AI tools.
Like, chat GPT will continue to

322
00:17:29,800 --> 00:17:33,400
get more powerful and produce
more useful outputs.

323
00:17:33,400 --> 00:17:36,400
This will help speed up more,
tedious tasks, so that we can

324
00:17:36,400 --> 00:17:40,200
focus more on strategy analysis,
Etc, John Kagan response to

325
00:17:40,200 --> 00:17:43,600
question for saying one that can
pull useful insights after

326
00:17:43,600 --> 00:17:47,300
watching my most frequent
reports and then Julie takes us

327
00:17:47,300 --> 00:17:50,000
on to question five asking, what
do you think is the most

328
00:17:50,000 --> 00:17:53,100
frustrating Troublesome or
negative aspect?

329
00:17:53,400 --> 00:17:57,200
Of AI for PPC.
So Travis respond saying there

330
00:17:57,200 --> 00:18:01,800
is an inherent risk of human
operators, taking lazy and

331
00:18:01,800 --> 00:18:05,300
brackets shortcuts.
The tools are no substitute for

332
00:18:05,300 --> 00:18:08,400
great strategy, quality
copywriting or data analysis

333
00:18:08,400 --> 00:18:12,300
insights.
That's actually a i I'd hate to

334
00:18:12,300 --> 00:18:16,700
see a door side content quality
degrade over time from Tool

335
00:18:16,700 --> 00:18:19,500
abuse.
That is such a fair point and

336
00:18:19,500 --> 00:18:22,600
shaba the ioc's agreed, I think
we'll see.

337
00:18:23,400 --> 00:18:27,700
Morris Q reaction across all
media from art to PC copy.

338
00:18:27,700 --> 00:18:31,200
Where made or written by hand is
a badge of honor.

339
00:18:31,200 --> 00:18:33,300
Absolutely.
Those kind of things of like,

340
00:18:33,300 --> 00:18:35,700
you know, I did this as my human
hand.

341
00:18:35,700 --> 00:18:39,700
This was not a, I supported with
one of those badge of honor like

342
00:18:39,700 --> 00:18:42,200
comments.
I agree with that Gilda response

343
00:18:42,200 --> 00:18:44,600
to question fights, and
ultimately machine learning uses

344
00:18:44,600 --> 00:18:48,000
data that's been generated by
real people, quality output

345
00:18:48,000 --> 00:18:52,000
requires quality input.
So the whole world can't totally

346
00:18:52,000 --> 00:18:56,000
shift to a He still needs to be
some human creativity.

347
00:18:56,000 --> 00:18:58,100
Agree.
And Julia voice response to

348
00:18:58,100 --> 00:19:01,800
question five saying it's a tie,
the comes the Chicken Little,

349
00:19:01,800 --> 00:19:06,300
the it will take over the P.
Max the trust, the machine, the

350
00:19:06,300 --> 00:19:09,400
Panic.
The immediate trust the boat

351
00:19:09,400 --> 00:19:12,800
battle, the antitrust, the mess
Lord.

352
00:19:12,800 --> 00:19:14,400
There's there's a lot to go
through here.

353
00:19:14,600 --> 00:19:17,300
There again, let me repeat what
she said.

354
00:19:17,400 --> 00:19:21,500
It's a tie, the cons, the
Chicken Little, the, it will

355
00:19:21,500 --> 00:19:25,000
take over the Pima.
The trust the machine, the

356
00:19:25,000 --> 00:19:27,900
Panic.
The immediate trust the bot

357
00:19:27,900 --> 00:19:31,200
battle.
The antitrust the mess.

358
00:19:31,200 --> 00:19:34,000
So yes, a lot to think about
them and then see if you can

359
00:19:34,100 --> 00:19:37,700
think that the tie between all
of those Julie respond saying

360
00:19:37,700 --> 00:19:41,200
I'm concerned about too much.
Only geniusness, if everyone is

361
00:19:41,200 --> 00:19:44,100
using AI for keyword ad, copy
and landing page ideas.

362
00:19:44,100 --> 00:19:47,300
Although it also creates space
for an advantage.

363
00:19:47,300 --> 00:19:51,600
For those who don't just use the
AI for all that training data

364
00:19:51,600 --> 00:19:54,900
bias is a real issue.
Issue to Travis goes.

365
00:19:54,900 --> 00:19:57,300
Great point about training days,
biases.

366
00:19:57,600 --> 00:19:59,800
That's not addressed nearly
often enough.

367
00:19:59,900 --> 00:20:03,900
There's a risk of diminishing
marginal return on total utility

368
00:20:03,900 --> 00:20:07,400
funny enough because of human
involvement, right?

369
00:20:07,500 --> 00:20:09,100
And Julie continue saying that.
Yeah.

370
00:20:09,300 --> 00:20:12,800
And then there's this, it has
the potential to really flatten

371
00:20:12,800 --> 00:20:16,300
research right now.
You have to either use a tool of

372
00:20:16,300 --> 00:20:19,100
some kind or manually, compile
your research.

373
00:20:19,100 --> 00:20:23,100
If AI is doing it all for you,
everyone will have the same.

374
00:20:23,600 --> 00:20:27,000
Again room to be better than a.
I do you live.

375
00:20:27,000 --> 00:20:30,500
I said this exact thing is going
to push propriety tools forward.

376
00:20:30,500 --> 00:20:34,800
Get ready for the political
chatter about polling quality to

377
00:20:34,800 --> 00:20:39,600
directly affect our industry and
Ashton clock response to

378
00:20:39,600 --> 00:20:41,100
question.
Five saying this isn't

379
00:20:41,100 --> 00:20:45,100
necessarily specific to PC by
tend to agree with the recent

380
00:20:45,100 --> 00:20:48,400
letters and statements about
slowing down AI production in

381
00:20:48,400 --> 00:20:52,400
general, this Tech is clearly
very powerful but can also be

382
00:20:52,400 --> 00:20:55,400
extremely I'm Philippa
development moves too quickly

383
00:20:55,600 --> 00:20:59,500
and we aren't careful and really
takes us on to the last question

384
00:20:59,500 --> 00:21:01,700
of the day and does another one
of those?

385
00:21:01,700 --> 00:21:04,100
Yeah, let's look include the
crystal ball.

386
00:21:04,400 --> 00:21:07,900
If we were to wish for anything.
And let's, let's for example,

387
00:21:08,100 --> 00:21:12,400
you know, and dream and wish
that actually our answers are

388
00:21:12,400 --> 00:21:18,000
queries, shall be listen to what
is on your AI wish list for PPC.

389
00:21:18,000 --> 00:21:21,400
He asks and trap is goes,
there's plenty of room for

390
00:21:21,400 --> 00:21:23,100
machine learning to address
poor.

391
00:21:23,200 --> 00:21:26,800
In search query, data example,
if my learning data is negative

392
00:21:26,800 --> 00:21:31,100
keyword, lists learn from that
address pluralization infer and

393
00:21:31,100 --> 00:21:34,700
avoid similar queries
advertisers provide powerful

394
00:21:34,700 --> 00:21:38,400
signals to Google to Casey stump
response saying, so all of the

395
00:21:38,400 --> 00:21:42,000
campaign building on its own
with just instructions from me,

396
00:21:42,200 --> 00:21:45,600
I still want to give its
settings and guidelines but if I

397
00:21:45,600 --> 00:21:48,200
didn't have to spend time
dealing with ADD managers

398
00:21:48,200 --> 00:21:51,700
anymore, I wouldn't be sad.
Yeah do you do Humane like

399
00:21:51,700 --> 00:21:53,500
Google ad managers because I
Agree.

400
00:21:53,600 --> 00:21:56,800
Julie advice gear comes in with
her wish list saying for now

401
00:21:56,800 --> 00:22:00,600
testing data Vis, and
high-volume AD creation, if it

402
00:22:00,600 --> 00:22:04,100
can understand specific
parameters, it's a tool

403
00:22:04,100 --> 00:22:08,600
ultimately probably more on my
list as things developed but

404
00:22:08,600 --> 00:22:12,000
that's on the short term for me.
We can decide if or how to

405
00:22:12,000 --> 00:22:15,500
deploy, then she continues.
Oh you know what find and manage

406
00:22:15,500 --> 00:22:18,500
and Miss information, that would
actually be great.

407
00:22:18,500 --> 00:22:21,400
Thank you.
And yeah, literally mr.

408
00:22:21,400 --> 00:22:22,000
Goes.
Yeah.

409
00:22:22,100 --> 00:22:23,000
Oh my God.
Yes.

410
00:22:23,200 --> 00:22:24,500
To use to this.
So yeah.

411
00:22:24,500 --> 00:22:26,500
Misinformation we need to start
with that.

412
00:22:26,900 --> 00:22:30,400
Then, Julia's World answers
question 16 where to start I

413
00:22:30,400 --> 00:22:34,200
would love a report Creator
where I told it what I wanted

414
00:22:34,500 --> 00:22:37,500
and it put all the data together
without me having to select

415
00:22:37,500 --> 00:22:41,200
every data point AI alerts.
Based on what I tell the

416
00:22:41,200 --> 00:22:45,400
platform is important to me
better recommendations, she

417
00:22:45,400 --> 00:22:48,100
continues, more advanced
semantic, understanding of

418
00:22:48,100 --> 00:22:52,200
words, how about some Basics?
Like not showing ads in

419
00:22:52,200 --> 00:22:56,300
untargeted Geographies
understanding query words that

420
00:22:56,300 --> 00:22:59,700
conflict with Geo targeting
settings and not showing the ADD

421
00:23:00,100 --> 00:23:04,900
no more disapprovals for single
words, words, cuz trademarks

422
00:23:04,900 --> 00:23:08,000
yeah, some some of them are
totally ridiculous and it takes

423
00:23:08,300 --> 00:23:11,700
a long time to.
Yeah, get over that trademark

424
00:23:11,700 --> 00:23:15,200
and to be showing for your ad
again, she continues suggesting

425
00:23:15,200 --> 00:23:18,400
possible negative keywords based
on the ones already added to the

426
00:23:18,408 --> 00:23:22,100
campaign or account highlighting
suggested placement, exclusions

427
00:23:22,100 --> 00:23:25,100
based on previously.
Excluded placements ad copy

428
00:23:25,100 --> 00:23:28,900
suggestions based on the ads
from advertisers in auction,

429
00:23:28,900 --> 00:23:31,700
insights as you continues, I
just can't stop.

430
00:23:31,700 --> 00:23:34,900
Now that I've started how about
a setting to Auto compared

431
00:23:34,900 --> 00:23:38,500
against a timeframe and
highlighting, what's changed and

432
00:23:38,500 --> 00:23:43,300
a why from the a I like cpcs are
up for this time because

433
00:23:43,300 --> 00:23:47,500
Advertiser X is spending a
buttload on that term now.

434
00:23:47,500 --> 00:23:52,400
So, yeah, more in-depth insight
as to why your performance or

435
00:23:52,400 --> 00:23:55,900
different.
Efficiency metrics has gone awry

436
00:23:56,300 --> 00:23:58,500
Angela versus well gives another
wish saying.

437
00:23:58,500 --> 00:24:00,800
I don't hate the idea of machine
learning or simple.

438
00:24:00,800 --> 00:24:04,100
AI media planning.
Hey plan a pal emphasize local

439
00:24:04,100 --> 00:24:09,000
media in my next by or tell me
what mood people are in on the

440
00:24:09,000 --> 00:24:13,000
various platforms.
I use so unfortunately, because

441
00:24:13,000 --> 00:24:16,800
I was actually in between trying
to multitask and when halfway

442
00:24:16,800 --> 00:24:20,400
through the chat, I have a
weekly internal meeting work

443
00:24:20,400 --> 00:24:22,800
meeting.
So I was trying to catch up with

444
00:24:22,800 --> 00:24:24,300
some of the Questions.
And I did want to answer

445
00:24:24,300 --> 00:24:27,900
question 5 and I said, I think
for me, it's the whole worry

446
00:24:28,000 --> 00:24:29,900
that it will take away their
jobs.

447
00:24:29,900 --> 00:24:32,500
So, literally in terms of what
you think is the most

448
00:24:32,500 --> 00:24:35,700
frustrating thing Troublesome,
one negative aspect of AI.

449
00:24:35,700 --> 00:24:38,100
So the whole it's going to take
away all our jobs AI, still

450
00:24:38,100 --> 00:24:41,000
needs a lot of guidance and it
only as good as the people

451
00:24:41,000 --> 00:24:44,400
creating it, having said that I
don't want Google forcing us to

452
00:24:44,400 --> 00:24:47,300
use it before.
It's ready, they'll be for the

453
00:24:47,300 --> 00:24:51,600
people creating it have actually
done, you know, a good job a

454
00:24:51,600 --> 00:24:54,800
thorough job.
Killer Bass response to that to

455
00:24:54,800 --> 00:24:57,700
my own so saying I have
fantastic news, Amazon and

456
00:24:57,700 --> 00:25:01,700
Facebook have already done this
Google, just told us about their

457
00:25:01,700 --> 00:25:04,900
plans.
So yeah, so Amazon and Facebook

458
00:25:04,900 --> 00:25:07,900
just went ahead without actually
telling us there, right?

459
00:25:07,900 --> 00:25:09,900
Oh, well, well, that brings us
to the end of that.

460
00:25:09,900 --> 00:25:15,200
Riverton chat about Ai and PPC.
And yeah, what we think about

461
00:25:15,200 --> 00:25:18,300
it, is it helpful, is it a bit
confusing?

462
00:25:18,300 --> 00:25:21,800
Should Google really be more
worried about the announcements

463
00:25:21,800 --> 00:25:24,400
and how they do announcing
Anyway, what do you think?

464
00:25:24,400 --> 00:25:27,800
And if you want to join us and
you have opinions about this

465
00:25:27,800 --> 00:25:31,400
topic and you'd like to, you
know, get on it and the more

466
00:25:31,400 --> 00:25:35,200
vocal way in a voice note, kind
of way, why don't you join us on

467
00:25:35,200 --> 00:25:38,900
Thursday at 5:00 p.m.
GMT. 12 p.m. eastern time, where

468
00:25:38,900 --> 00:25:42,300
we'll be having the Twitter
spaces chat about this, very

469
00:25:42,300 --> 00:25:46,400
topic for any feedback about the
podcast or maybe even some

470
00:25:46,400 --> 00:25:48,500
correction on anything that has
been shared.

471
00:25:48,800 --> 00:25:51,800
Do get in touch with me on
Twitter, my DMs are open on them

472
00:25:51,800 --> 00:25:54,700
handle at the mall.
Getting a new or find me on

473
00:25:54,700 --> 00:25:56,600
LinkedIn.
We also have the PPC chat

474
00:25:56,800 --> 00:25:59,600
Roundup podcast group so join us
on there.

475
00:25:59,800 --> 00:26:02,300
They'll also be great for
long-standing members and also

476
00:26:02,300 --> 00:26:05,600
brand new listeners to give the
podcast a rating or review.

477
00:26:05,600 --> 00:26:11,200
If you just go to link tree.com,
/p P c-- chat on the score round

478
00:26:11,200 --> 00:26:13,500
up, you Liam then be taken to
the different links.

479
00:26:13,500 --> 00:26:16,200
For that Spotify, I pull
excetera where they where you

480
00:26:16,200 --> 00:26:19,400
can, you know, take you seconds
to just leave a rating and it

481
00:26:19,400 --> 00:26:22,600
just gives an indication to
people that support cast worth

482
00:26:22,600 --> 00:26:25,600
listening to It's a great
motivator for me and means means

483
00:26:25,600 --> 00:26:29,000
that I'm motivated to keep doing
this every week.

484
00:26:29,000 --> 00:26:31,000
Even the night before I'm about
to travel.

485
00:26:31,300 --> 00:26:35,300
Yeah, we got a trip to her but
I'm loving doing this.

486
00:26:35,600 --> 00:26:41,500
Yeah, you'll also be great if
you check out our PPC, live UK

487
00:26:41,700 --> 00:26:45,500
is next event and that's going
to be on the 25th of May, if you

488
00:26:45,500 --> 00:26:49,800
know anyone who wants to
sponsor, please email, hello at

489
00:26:49,800 --> 00:26:55,000
PPC live UK.com to let us know
the packages start from as low

490
00:26:55,000 --> 00:26:59,300
as 500 pounds to a thousand five
and it really just helps us make

491
00:26:59,300 --> 00:27:02,200
this event.
A lovely, you know, boisterous

492
00:27:02,200 --> 00:27:05,900
networking event and allows us
to provide food allows us to

493
00:27:05,900 --> 00:27:08,500
record the talk.
So they're they're available on

494
00:27:08,500 --> 00:27:12,600
YouTube and just allows us to
learn more about what's going on

495
00:27:12,900 --> 00:27:17,300
in this industry in a very nice
and comfortable environment and

496
00:27:17,300 --> 00:27:19,500
face-to-face.
I don't know about some of you.

497
00:27:19,600 --> 00:27:22,900
I'm an extrovert I love meat
around with people so yeah.

498
00:27:23,100 --> 00:27:24,100
The law.
It's always great.

499
00:27:24,100 --> 00:27:27,200
Having these kind of events,
finally remember for your

500
00:27:27,200 --> 00:27:29,100
campaigns and businesses to
Glide smoothly.

501
00:27:29,100 --> 00:27:31,800
There's a lot of hard work
needed beneath the surface, so

502
00:27:31,800 --> 00:27:34,700
keep his hands kicking by and
speak to you next week.