1
00:00:00,000 --> 00:00:09,960
Welcome to Artificially Intelligent Marketing, a weekly podcast where we stay on top of the

2
00:00:09,960 --> 00:00:15,700
latest trends, tips, and tools in the world of marketing AI, helping you get the best

3
00:00:15,700 --> 00:00:18,520
results from your marketing efforts.

4
00:00:18,520 --> 00:00:22,360
Now let's join our hosts, Paul Avery and Martin Broadhurst.

5
00:00:22,360 --> 00:00:28,540
Hello, welcome to episode 24 of Artificially Intelligent Marketing, where we're looking

6
00:00:28,540 --> 00:00:33,640
forward to keeping all of you lovely marketers up to date on all the things in AI that you

7
00:00:33,640 --> 00:00:35,400
need to know about.

8
00:00:35,400 --> 00:00:41,640
Now not much news really in the world of marketing AI this week, but what there has been has

9
00:00:41,640 --> 00:00:44,220
been absolutely massive.

10
00:00:44,220 --> 00:00:48,440
So we've got three or four main topics that Martin and I are going to cover today, and

11
00:00:48,440 --> 00:00:53,320
then we're going to jump into our interview this week, which is with Jeff Coyle, co-founder

12
00:00:53,320 --> 00:00:58,920
at Market Muse, which is a leading software tool for AI powered content strategy.

13
00:00:58,920 --> 00:01:02,800
As usual, I'm going to be joined today by my co-host Martin Broadhurst.

14
00:01:02,800 --> 00:01:05,320
Martin, how are you this week?

15
00:01:05,320 --> 00:01:10,160
I am feeling good and I'm really enjoying the mix of articles that we've got coming

16
00:01:10,160 --> 00:01:11,480
up this week.

17
00:01:11,480 --> 00:01:14,200
Some big stories, as you say.

18
00:01:14,200 --> 00:01:19,440
I just want to spend a minute just looking at my experiences dealing with some clients

19
00:01:19,440 --> 00:01:24,080
this week, because I had an interesting one where I was doing a workshop, I was doing

20
00:01:24,080 --> 00:01:29,440
a HubSpot implementation with a client of mine, and as part of that process, I used

21
00:01:29,440 --> 00:01:33,680
ChatGPT in a way that I would typically use ChatGPT.

22
00:01:33,680 --> 00:01:35,840
I just asked it a simple question.

23
00:01:35,840 --> 00:01:41,920
I asked it for some ethnicity categories that you would typically find on UK surveys, and

24
00:01:41,920 --> 00:01:47,600
then I had a bit of a back and forth as I asked it to format the data for me.

25
00:01:47,600 --> 00:01:53,640
What was really surprising was the response of my clients as they sat there in awe of

26
00:01:53,640 --> 00:01:55,760
the way that I was using it.

27
00:01:55,760 --> 00:02:02,980
It just reminded me just how far we've still got to go until this adoption.

28
00:02:02,980 --> 00:02:04,840
These tools, I'm using them all day every day.

29
00:02:04,840 --> 00:02:08,640
I know that you're obviously using them regularly.

30
00:02:08,640 --> 00:02:14,940
People are still completely unaware of just how much these tools can improve efficiency

31
00:02:14,940 --> 00:02:16,640
and productivity in the workplace.

32
00:02:16,640 --> 00:02:17,640
I agree.

33
00:02:17,640 --> 00:02:22,080
I think it's very interesting you've had that experience this week, because we'll see from

34
00:02:22,080 --> 00:02:28,660
our news stories that we have to tell the listeners this week that the tipping point,

35
00:02:28,660 --> 00:02:31,800
you could argue it's coming, but I think this week it may be there.

36
00:02:31,800 --> 00:02:38,080
We'll get a bit deeper into that as we get through the stories for this week.

37
00:02:38,080 --> 00:02:44,360
Myself, I was on holiday last week in sunny Norfolk, so very little on the AI front for

38
00:02:44,360 --> 00:02:49,800
me, lots of trampoline bouncing instead with my nieces and nephews.

39
00:02:49,800 --> 00:02:54,400
But like you say, really excited to see some of the stories that we have this week.

40
00:02:54,400 --> 00:03:00,640
Let's get stuck in straight away with story one, which is that ChatGPT Enterprise has

41
00:03:00,640 --> 00:03:07,160
been unveiled by the team over at OpenAI, which they promise will be an enhanced version

42
00:03:07,160 --> 00:03:11,280
of ChatGPT designed specifically for business use.

43
00:03:11,280 --> 00:03:16,280
This is not a complete surprise, but some of the nuances of what you're going to be

44
00:03:16,280 --> 00:03:22,520
able to do with ChatGPT Enterprise is quite interesting.

45
00:03:22,520 --> 00:03:27,960
The first one is enterprise-grade security and privacy.

46
00:03:27,960 --> 00:03:31,480
Your prompts and your company data that you upload into the system are not going to be

47
00:03:31,480 --> 00:03:34,560
used for training OpenAI's models.

48
00:03:34,560 --> 00:03:39,440
Data is going to be encrypted pretty much all the time, and the system has been certified

49
00:03:39,440 --> 00:03:42,160
as SOC 2 compliant.

50
00:03:42,160 --> 00:03:47,400
So in essence, what this does is overcomes most of the objections that a lot of companies

51
00:03:47,400 --> 00:03:55,080
have had about letting their employees use ChatGPT around data security and privacy.

52
00:03:55,080 --> 00:04:00,600
Added to that, there's going to be a number of features that just make it easier for enterprises

53
00:04:00,600 --> 00:04:05,760
to roll ChatGPT out to their employees.

54
00:04:05,760 --> 00:04:09,840
The types of things that are honestly the mainstay in enterprise-level software that

55
00:04:09,840 --> 00:04:16,240
now ChatGPT will have, so there'll be an admin console for an admin to be able to bulk manage

56
00:04:16,240 --> 00:04:17,240
users.

57
00:04:17,240 --> 00:04:19,520
You're going to be able to use things like single sign-on.

58
00:04:19,520 --> 00:04:22,680
You're going to have domain verification, and there's even going to be an analytics

59
00:04:22,680 --> 00:04:27,520
dashboard for admins to be able to see who's using the tool, how much they're using it,

60
00:04:27,520 --> 00:04:30,280
and all of that good stuff.

61
00:04:30,280 --> 00:04:34,520
So again, the types of things that you'd expect to see really an enterprise-level software

62
00:04:34,520 --> 00:04:37,040
having that so far ChatGPT hasn't had.

63
00:04:37,040 --> 00:04:42,200
I think where it gets really interesting is that this is now promising to be the most

64
00:04:42,200 --> 00:04:47,080
powerful version of ChatGPT yet, because if you have enterprise, you're going to have

65
00:04:47,080 --> 00:04:48,880
unlimited access to GPT-4.

66
00:04:48,880 --> 00:04:50,920
So it's going to be no usage caps.

67
00:04:50,920 --> 00:04:56,280
It's not going to be like the 50 messages every three hours that most paid users currently

68
00:04:56,280 --> 00:04:57,280
face.

69
00:04:57,280 --> 00:05:02,300
It's also promising to be higher performance, up to two times faster.

70
00:05:02,300 --> 00:05:03,920
How are they achieving that?

71
00:05:03,920 --> 00:05:08,620
Those reasonable questions like, are they throwing more GPU at it, or have they actually

72
00:05:08,620 --> 00:05:11,760
changed the way the model works?

73
00:05:11,760 --> 00:05:13,360
Who knows?

74
00:05:13,360 --> 00:05:17,560
To be unveiled at some point in the future, one assumes.

75
00:05:17,560 --> 00:05:22,200
If you're an enterprise user, you're going to have unlimited access to what was known

76
00:05:22,200 --> 00:05:26,320
as code interpreter, but will now be called advanced data analysis.

77
00:05:26,320 --> 00:05:31,320
You're going to have 32k token context windows as standard, which of course most people have

78
00:05:31,320 --> 00:05:35,320
been waiting for both through the API and through the tool itself.

79
00:05:35,320 --> 00:05:38,600
But now it seems like that's going to be coming to enterprise first.

80
00:05:38,600 --> 00:05:42,560
You're going to have shareable chat templates for your company to use.

81
00:05:42,560 --> 00:05:47,160
There's been lots of tools emerged to try and help people have like, cropped databases

82
00:05:47,160 --> 00:05:48,280
and all this other stuff.

83
00:05:48,280 --> 00:05:51,800
So I think basically that's going to be built in.

84
00:05:51,800 --> 00:05:55,960
And you're going to get free credits to use the APIs if you need to extend the tool beyond

85
00:05:55,960 --> 00:05:58,340
just using it as a chat tool.

86
00:05:58,340 --> 00:06:03,720
So pretty exciting news, especially for those companies that have not been able to use chat

87
00:06:03,720 --> 00:06:12,560
GPT for a lot of those data security fears, but also some extra powerful aspects to chat

88
00:06:12,560 --> 00:06:15,000
GPT that we all can't access right now.

89
00:06:15,000 --> 00:06:19,640
And we'll be waiting to see how much this is going to cost because that hasn't been

90
00:06:19,640 --> 00:06:20,640
announced yet.

91
00:06:20,640 --> 00:06:23,680
What are your thoughts on this story, Mark?

92
00:06:23,680 --> 00:06:28,480
I'm very jealous of those users that get access to this early on with those two times faster

93
00:06:28,480 --> 00:06:32,480
speeds and 32k context window.

94
00:06:32,480 --> 00:06:35,680
But you can see that they've already got some big companies on board.

95
00:06:35,680 --> 00:06:39,800
So they've mentioned some other enterprise or early adopter enterprises, the likes of

96
00:06:39,800 --> 00:06:46,760
Canva and I think there was some Deloitte maybe and McKinsey, those clients are getting

97
00:06:46,760 --> 00:06:48,680
on board with it early.

98
00:06:48,680 --> 00:06:54,960
Yeah, it was the natural progression as we're going to discuss shortly.

99
00:06:54,960 --> 00:06:59,480
More AI tools are coming to the fore and getting more baked into the productivity suites that

100
00:06:59,480 --> 00:07:00,480
we all know and love.

101
00:07:00,480 --> 00:07:05,080
So OpenAI making this move this week makes logical sense.

102
00:07:05,080 --> 00:07:06,080
Absolutely.

103
00:07:06,080 --> 00:07:10,120
I mean, I think probably the bit that I'd be the most excited about is being able to

104
00:07:10,120 --> 00:07:17,760
use code interpreter, now advanced data analysis to analyze both client and biostrata data

105
00:07:17,760 --> 00:07:20,160
without worrying about what to upload.

106
00:07:20,160 --> 00:07:27,360
Because so far, I haven't had confidence enough in OpenAI to give them anything of any real

107
00:07:27,360 --> 00:07:32,680
meaning, if I'm honest, certainly nothing easily identifiable as data from biostrata

108
00:07:32,680 --> 00:07:33,680
or as clients.

109
00:07:33,680 --> 00:07:38,400
So to be able to actually lean into that and properly analyze data without that worry is

110
00:07:38,400 --> 00:07:46,040
a big thing for me, especially given the wide skill set that code interpreter has outside

111
00:07:46,040 --> 00:07:50,520
of just your standard data analysis and creating graphs and things that I know will not have

112
00:07:50,520 --> 00:07:52,160
the early use cases.

113
00:07:52,160 --> 00:07:54,920
You've been playing with code interpreter this week, haven't you, Mike?

114
00:07:54,920 --> 00:07:56,840
Yeah, just trying something a bit different.

115
00:07:56,840 --> 00:08:04,200
I was getting frustrated with some long text and going back to that 32k context window,

116
00:08:04,200 --> 00:08:05,320
that's what I really want.

117
00:08:05,320 --> 00:08:15,800
But in light of not having that, I tried uploading a TXT file with the text broken down into

118
00:08:15,800 --> 00:08:19,800
chapters that were very clearly marked in the file.

119
00:08:19,800 --> 00:08:30,760
And then I uploaded that file to code interpreter and basically helped me to write a blog post

120
00:08:30,760 --> 00:08:32,080
with that as the reference material.

121
00:08:32,080 --> 00:08:34,160
And it did a better job than Claude.

122
00:08:34,160 --> 00:08:39,120
Claude has that bigger context window, but I just feel that his writing isn't as sharp.

123
00:08:39,120 --> 00:08:46,960
So by saying to code interpreter, check the TXT file and go to this part of the text and

124
00:08:46,960 --> 00:08:55,160
then read that chapter, it was able to give it the, basically an artificially large context

125
00:08:55,160 --> 00:08:58,920
window through which to pull some source material down.

126
00:08:58,920 --> 00:09:02,720
And it runs all of that through its Python scripts and what have you.

127
00:09:02,720 --> 00:09:03,840
And it worked really well.

128
00:09:03,840 --> 00:09:06,280
I was very happy with the outputs.

129
00:09:06,280 --> 00:09:10,400
Not something that I thought to do in terms of taking a TXT file before, but yeah, I would

130
00:09:10,400 --> 00:09:15,640
definitely recommend people do that in order to interrogate some longer form text.

131
00:09:15,640 --> 00:09:17,320
Yeah, it's an interesting use case.

132
00:09:17,320 --> 00:09:22,560
And I think given that you did that to get around not having a very large context window

133
00:09:22,560 --> 00:09:29,320
so far, one has to assume that at some point we're all going to get 32k even if you're

134
00:09:29,320 --> 00:09:34,360
not on the enterprise plan, you can still use that approach to increase the context

135
00:09:34,360 --> 00:09:35,960
window even more.

136
00:09:35,960 --> 00:09:39,320
So yeah, very interesting stuff.

137
00:09:39,320 --> 00:09:45,280
And dear listeners, if you have any cool applications of code interpreter that you've been playing

138
00:09:45,280 --> 00:09:50,120
with, we'd love to hear about them because I think it's probably of all of the different

139
00:09:50,120 --> 00:09:54,640
things you can do with chat GPT and other, you know, Claude and other large language

140
00:09:54,640 --> 00:09:56,440
model bots.

141
00:09:56,440 --> 00:10:03,320
That I think is the one that has the largest scope to do really cool stuff beyond just

142
00:10:03,320 --> 00:10:11,680
having a content creation engine or having conversations with a AI bot.

143
00:10:11,680 --> 00:10:16,680
Right next story, it's going to be HubSpot's inbound conference this week.

144
00:10:16,680 --> 00:10:20,240
You'll be heading off there, Martin, what are your thoughts?

145
00:10:20,240 --> 00:10:23,360
There's going to be a maybe what are the announcements are going to be?

146
00:10:23,360 --> 00:10:26,560
What do we think is going to go on in Boston this week?

147
00:10:26,560 --> 00:10:30,520
Well, we know that AI is going to feature heavily.

148
00:10:30,520 --> 00:10:33,480
HubSpot have made some big moves in this.

149
00:10:33,480 --> 00:10:38,880
They've already integrated AI through the content assistant, which is kind of baked

150
00:10:38,880 --> 00:10:42,640
into the web tool and the email marketing tool.

151
00:10:42,640 --> 00:10:45,160
So it will write your subject line.

152
00:10:45,160 --> 00:10:47,640
It will actually write entire emails for you.

153
00:10:47,640 --> 00:10:54,360
If you use the HubSpot sales plugin on Chrome or Outlook, you can get the AI to write your

154
00:10:54,360 --> 00:10:59,360
emails and it will ask if it's an introduction email, follow up email or something like that.

155
00:10:59,360 --> 00:11:05,680
And you can use a bit of prompt engineering to get high quality outputs from it.

156
00:11:05,680 --> 00:11:09,760
But AI we know is going to be very heavily baked into the product going forward.

157
00:11:09,760 --> 00:11:11,000
And how do we know that?

158
00:11:11,000 --> 00:11:16,160
Well, HubSpot has already published an AI roadmap.

159
00:11:16,160 --> 00:11:20,880
So in the roadmap, this is freely available online.

160
00:11:20,880 --> 00:11:27,000
They talk about HubSpot AI in 2023 is really focusing on three main priorities.

161
00:11:27,000 --> 00:11:32,600
One is to increase AI powered efficiency across your company.

162
00:11:32,600 --> 00:11:38,720
And the idea is that they want to help companies to eliminate repetitive tasks that take time,

163
00:11:38,720 --> 00:11:42,680
money, deep expertise, and they want AI to do the heavy lifting there.

164
00:11:42,680 --> 00:11:48,120
Number two, they want to drive effectiveness using AI for better business outcomes.

165
00:11:48,120 --> 00:11:56,680
And that's more about using AI to unlock intelligent insights and basically giving you actionable

166
00:11:56,680 --> 00:12:01,920
insight on what way to maybe use your data.

167
00:12:01,920 --> 00:12:03,560
And that's going to be driving that.

168
00:12:03,560 --> 00:12:08,040
And then number three is connect with your customers better than ever before, which is

169
00:12:08,040 --> 00:12:14,140
going to be more about recommendation engines, personalization, individualized engagement

170
00:12:14,140 --> 00:12:16,880
strategies is something that they talk about in the roadmap.

171
00:12:16,880 --> 00:12:21,960
Now this is going to be rolled out across all of their main hubs.

172
00:12:21,960 --> 00:12:26,520
So they've got marketing hub, sales hub, service hub, and CMS hub.

173
00:12:26,520 --> 00:12:33,200
And I'm very, very excited to see what particular products they launch across these hubs.

174
00:12:33,200 --> 00:12:38,040
Now in marketing hub as it exists today, they've already got content assistant, which I've

175
00:12:38,040 --> 00:12:43,240
already mentioned for landing pages, social media, et cetera.

176
00:12:43,240 --> 00:12:46,400
But they talk about image generation.

177
00:12:46,400 --> 00:12:52,320
This is something that I haven't seen baked into the platform as of yet, but presumably

178
00:12:52,320 --> 00:12:58,160
it is on the horizon for social media posts and you can well imagine how if you're designing

179
00:12:58,160 --> 00:13:06,920
landing pages or call to actions, having a stable diffusion type model baked into HubSpot

180
00:13:06,920 --> 00:13:12,120
will make finding the right image just that little bit easier.

181
00:13:12,120 --> 00:13:16,920
Now on the sales hub side of things, this is where I think the intelligence really comes

182
00:13:16,920 --> 00:13:17,920
into it.

183
00:13:17,920 --> 00:13:24,480
They're talking about AI powered sales forecasts, predictive deal health scores, as well as

184
00:13:24,480 --> 00:13:32,120
content assistant for Gmail, Outlook, and their mobile app, which is also brilliant.

185
00:13:32,120 --> 00:13:34,840
I think that's going to be huge.

186
00:13:34,840 --> 00:13:43,080
If they can get the AI powered sales forecasting right and things like AI powered lead scoring,

187
00:13:43,080 --> 00:13:49,120
and they've had smart leads going for some time and it's worked very well for organizations

188
00:13:49,120 --> 00:13:54,620
that I know have deployed it, but if they can do that with even more data and in even

189
00:13:54,620 --> 00:14:04,400
more areas of the sales process, I think this is going to help them stand out against other

190
00:14:04,400 --> 00:14:07,120
CRM systems.

191
00:14:07,120 --> 00:14:08,380
There's one piece in the roadmap.

192
00:14:08,380 --> 00:14:10,800
There's a bunch of other elements that are in here.

193
00:14:10,800 --> 00:14:14,000
I'm not going to go through them all line by line, but there's one thing that really

194
00:14:14,000 --> 00:14:18,920
stood out to me in the roadmap and we haven't seen any sign of it in HubSpot as of yet,

195
00:14:18,920 --> 00:14:26,520
but other companies are bringing it to the fore, and that is an AI website builder.

196
00:14:26,520 --> 00:14:29,160
So HubSpot has CMS hub.

197
00:14:29,160 --> 00:14:31,640
You can build your entire website on there.

198
00:14:31,640 --> 00:14:34,400
It's a very good content management system.

199
00:14:34,400 --> 00:14:38,080
We saw Wix announce their AI powered website builder.

200
00:14:38,080 --> 00:14:43,120
We saw the demo, the trailer looks brilliant, looks very slick.

201
00:14:43,120 --> 00:14:44,640
Is that what HubSpot's going for?

202
00:14:44,640 --> 00:14:49,800
Are they going for that same prompt driven website builder or is it going to be something

203
00:14:49,800 --> 00:14:50,800
entirely different?

204
00:14:50,800 --> 00:14:54,080
I don't know, and I'm excited to find out.

205
00:14:54,080 --> 00:14:58,520
Hopefully they're going to make some announcements about that next week, but yeah, I wait with

206
00:14:58,520 --> 00:14:59,520
baited breath.

207
00:14:59,520 --> 00:15:02,760
Yeah, I think it's going to be really interesting.

208
00:15:02,760 --> 00:15:08,120
Since HubSpot have been much more transparent or maybe even just proactive is a better word

209
00:15:08,120 --> 00:15:10,820
when it comes to their roadmap.

210
00:15:10,820 --> 00:15:17,000
Some of the excitement has perhaps gone out of HubSpot product updates because they have

211
00:15:17,000 --> 00:15:19,360
this roadmap and so they're kind of telling us what's coming.

212
00:15:19,360 --> 00:15:25,160
But one of the fun things about going to inbound was always what exciting new thing are they

213
00:15:25,160 --> 00:15:30,000
going to launch within the tool set that we don't know about.

214
00:15:30,000 --> 00:15:35,720
It would be really great if there are a few aspects there that are not so overtly obvious

215
00:15:35,720 --> 00:15:40,400
on their roadmap that just will get us all excited.

216
00:15:40,400 --> 00:15:48,120
For me, I'm hoping that they bait AI into the content creation process in a more effective

217
00:15:48,120 --> 00:15:49,120
way.

218
00:15:49,120 --> 00:15:53,120
We've got an interview later with Jeff at Market Muse.

219
00:15:53,120 --> 00:16:01,400
There's so much scope for AI to touch and influence every part of the HubSpot stack.

220
00:16:01,400 --> 00:16:04,880
And I feel like they've made really good progress with things like content assistant, but I

221
00:16:04,880 --> 00:16:09,560
want to be able to get a full blog draft that I edit a little bit like some of the workflows

222
00:16:09,560 --> 00:16:17,080
that BioStratasone Marketing leverages with ChatGPT, for example, rather than the morbidity

223
00:16:17,080 --> 00:16:19,040
process that it is right now.

224
00:16:19,040 --> 00:16:23,360
It should also be baked into the SEO aspects of HubSpot, right, by definition.

225
00:16:23,360 --> 00:16:27,960
And as you've talked about having images that are automatically generated, at least for

226
00:16:27,960 --> 00:16:34,320
the header for the blog, should be relatively easy to do and is drifting towards table stakes,

227
00:16:34,320 --> 00:16:39,960
I think, at this point in terms of how hard is it to connect a content creation tool with

228
00:16:39,960 --> 00:16:45,640
an image generation tool where the content in the blog post is, in essence, the prompt

229
00:16:45,640 --> 00:16:46,880
that drives the image generation.

230
00:16:46,880 --> 00:16:47,880
Right.

231
00:16:47,880 --> 00:16:48,880
That shouldn't be that hard.

232
00:16:48,880 --> 00:16:52,240
I'm really open to see some of that cool stuff.

233
00:16:52,240 --> 00:16:56,800
Again, they've been really proactive with those betas.

234
00:16:56,800 --> 00:17:00,960
I just got access this week to the subject line beta, so I'm going to be giving that

235
00:17:00,960 --> 00:17:02,280
a try this week.

236
00:17:02,280 --> 00:17:06,600
For those of you that don't know, it's using generative AI to come up with different subject

237
00:17:06,600 --> 00:17:10,740
lines for your emailers, but doing it automatically and allowing you to quickly and easily run

238
00:17:10,740 --> 00:17:13,080
things like A-B tests and all that good stuff.

239
00:17:13,080 --> 00:17:15,400
So again, I really want those things.

240
00:17:15,400 --> 00:17:19,880
I really want some cool announcements, but my worry is all the cool stuff now you can

241
00:17:19,880 --> 00:17:24,960
get on restricted betas, especially if you're a partner agency, and you've got the roadmap.

242
00:17:24,960 --> 00:17:30,640
So I want to see between the lines of the roadmap something cool and interesting that

243
00:17:30,640 --> 00:17:32,120
I'm not expecting.

244
00:17:32,120 --> 00:17:34,000
Yeah, agreed.

245
00:17:34,000 --> 00:17:40,320
Certainly on the SEO front, I think the topic cluster tool that they have within HubSpot

246
00:17:40,320 --> 00:17:48,400
at the moment is good for helping you literally join the dots on your content strategy, but

247
00:17:48,400 --> 00:17:56,360
it isn't doing the AI-powered semantic SEO user intent NLP analysis.

248
00:17:56,360 --> 00:17:59,080
And that's where, maybe they're just saying, look, that's not us.

249
00:17:59,080 --> 00:18:03,800
And that is for the likes of MarketMuse, that is for the likes of SEMrush and what have

250
00:18:03,800 --> 00:18:04,800
you to do.

251
00:18:04,800 --> 00:18:06,920
We're not stepping into that territory.

252
00:18:06,920 --> 00:18:12,720
But I do feel like that is where there's big gains to be made in the HubSpot stack.

253
00:18:12,720 --> 00:18:13,920
Yeah, I agree.

254
00:18:13,920 --> 00:18:21,420
And by default, having all of our blog posts on HubSpot CMS should give us a leg up in

255
00:18:21,420 --> 00:18:26,140
terms of being able to train the AI to produce content that's written more like we write

256
00:18:26,140 --> 00:18:28,680
because it has access to all of our content.

257
00:18:28,680 --> 00:18:34,280
And I know they gave it some capabilities to train it on your content, but they're quite

258
00:18:34,280 --> 00:18:38,200
limited so far, which I'm sure is context window limited.

259
00:18:38,200 --> 00:18:43,280
So when some of the underlying tools, their context window increases, like we just talked

260
00:18:43,280 --> 00:18:45,660
about with ChatGBT, perhaps that will get better.

261
00:18:45,660 --> 00:18:51,480
But that's another thing I would love to see announced this week.

262
00:18:51,480 --> 00:18:57,800
Now there were some large announcements the week just gone, but in this case from Google.

263
00:18:57,800 --> 00:19:05,000
So in our next story, Google Workspace is now rolling out Google Duet to all Workspace

264
00:19:05,000 --> 00:19:11,000
users, which is a real time AI system that is aiming to simplify pretty much all your

265
00:19:11,000 --> 00:19:12,000
work tasks.

266
00:19:12,000 --> 00:19:13,000
Why is this a big deal?

267
00:19:13,000 --> 00:19:17,400
Well, many of you that are listening will probably be using Google Workspace to power

268
00:19:17,400 --> 00:19:23,680
your work within your businesses, whether that's Gmail, Google Docs, etc, etc.

269
00:19:23,680 --> 00:19:31,880
And Google Workspace already has 3 billion users and 10 million paying businesses customers.

270
00:19:31,880 --> 00:19:38,280
So you can imagine that rolling out this type of AI powered assistant to that many businesses

271
00:19:38,280 --> 00:19:43,120
is that inflection point that Mark and I were talking about the start of the show in terms

272
00:19:43,120 --> 00:19:50,640
of really handing AI powered tools on a silver platter, making it super easy for people to

273
00:19:50,640 --> 00:19:51,840
use them.

274
00:19:51,840 --> 00:19:57,520
To give you a bit of an idea of what this is going to look like, do AI promises to be

275
00:19:57,520 --> 00:20:02,400
able to compile a complete presentation for you based on data in your Google Drive and

276
00:20:02,400 --> 00:20:07,480
Gmail big fat asterisk on that on to how well that actually works.

277
00:20:07,480 --> 00:20:11,960
But turning something that might have taken all day in something you can get done before

278
00:20:11,960 --> 00:20:16,760
lunchtime or maybe even before your morning copy.

279
00:20:16,760 --> 00:20:20,920
Perhaps just as excitingly, at least for me is the fact that do AI is also going to be

280
00:20:20,920 --> 00:20:23,440
integrated into Google Meet.

281
00:20:23,440 --> 00:20:28,480
So it's going to have features like auto translated captions in 18 languages, improved audio and

282
00:20:28,480 --> 00:20:29,880
video quality.

283
00:20:29,880 --> 00:20:33,960
And fundamentally, the coolest thing is if you miss a meeting, or you're on a meeting

284
00:20:33,960 --> 00:20:39,040
and you don't want to manually take notes, the AI will attend for you take notes and

285
00:20:39,040 --> 00:20:43,320
then send out summaries with actions to keep everybody in the loop.

286
00:20:43,320 --> 00:20:44,320
So that's pretty cool.

287
00:20:44,320 --> 00:20:46,360
And I think pretty powerful.

288
00:20:46,360 --> 00:20:51,440
We've been waiting for something like this to be easily accessible in Teams and Zoom

289
00:20:51,440 --> 00:20:56,920
and it looks like Google may actually gives up the more than a little bit in terms of

290
00:20:56,920 --> 00:21:00,600
bringing that power to people more widely.

291
00:21:00,600 --> 00:21:04,760
And of course, there are lots of users of workspace.

292
00:21:04,760 --> 00:21:07,640
There is going to be some improvements to Google Chat as well.

293
00:21:07,640 --> 00:21:11,280
I have to admit we're a Google Workspace business, but we don't really use Google Chat.

294
00:21:11,280 --> 00:21:18,520
So it'll be interesting to see if this makes it a true competitor to Slack and Teams and

295
00:21:18,520 --> 00:21:24,120
other chat tools that people use, where it's going to be able to enable you to do things

296
00:21:24,120 --> 00:21:30,040
like ask the do AI questions directly in chat and also keep up with what's going on in your

297
00:21:30,040 --> 00:21:36,720
team and across your business by integrating and having conversations with the Google do

298
00:21:36,720 --> 00:21:39,320
AI model within the chat interface.

299
00:21:39,320 --> 00:21:43,520
And then the last part as a callback to what we were talking about earlier with chat GPT

300
00:21:43,520 --> 00:21:49,440
for enterprise is that do AI is being designed with privacy and security in mind, ensuring

301
00:21:49,440 --> 00:21:52,260
that your data remains your own.

302
00:21:52,260 --> 00:21:55,720
So that's obviously going to be really important for most businesses that want to hopped into

303
00:21:55,720 --> 00:21:56,720
this.

304
00:21:56,720 --> 00:22:03,440
The relevance here for marketers is be prepared to upskill quickly in what you can use do

305
00:22:03,440 --> 00:22:06,980
AI for within your business.

306
00:22:06,980 --> 00:22:13,560
If it gets activated at the moment, it's being rolled out as a 14 day free trial limited

307
00:22:13,560 --> 00:22:14,560
to 10 users.

308
00:22:14,560 --> 00:22:16,240
Martin and I have got access to it.

309
00:22:16,240 --> 00:22:18,300
We'll go into that in a bit.

310
00:22:18,300 --> 00:22:25,280
But once you are through your free trial, it will be here in the UK, 30 pounds plus

311
00:22:25,280 --> 00:22:27,240
a user per month.

312
00:22:27,240 --> 00:22:31,980
Now Biostrat I think pays about six pound a user a month as it stands.

313
00:22:31,980 --> 00:22:37,000
So certainly for SMEs and potentially even mid sized businesses, this is going to be

314
00:22:37,000 --> 00:22:43,080
a big jump in how much you're paying Google for software and is in line with the reported

315
00:22:43,080 --> 00:22:49,220
costs of Microsoft Copilot when that comes out, which I'm sure they will be scrambling

316
00:22:49,220 --> 00:22:54,680
to do in the next week or two now that Google do it has been released.

317
00:22:54,680 --> 00:22:56,360
What are your thoughts on this one, Martin?

318
00:22:56,360 --> 00:22:58,400
I think you've been having a play, haven't you?

319
00:22:58,400 --> 00:23:02,240
I have, but before we get into that, I just want to touch on the price.

320
00:23:02,240 --> 00:23:09,720
So the 30 pounds per month is on top of any existing subscription that you've got.

321
00:23:09,720 --> 00:23:11,160
So it's not a replacement for that.

322
00:23:11,160 --> 00:23:17,320
So if you're paying seven pound per user per month, that's now 37 pound per user per month.

323
00:23:17,320 --> 00:23:22,200
If you pay annually rather than monthly, it is 25 pounds per month.

324
00:23:22,200 --> 00:23:27,840
So there is a significant jump, sorry, discount for the annual commitment, which I think is

325
00:23:27,840 --> 00:23:28,920
reasonable.

326
00:23:28,920 --> 00:23:32,480
The other thing I did appreciate when I was looking at the pricing is that you don't have

327
00:23:32,480 --> 00:23:35,560
to assign the licenses company wide.

328
00:23:35,560 --> 00:23:41,520
So unlike with Google workspace, where if the workspace is on say the business plan

329
00:23:41,520 --> 00:23:46,080
or the enterprise plan, everyone in the organization has to have that with this, that isn't the

330
00:23:46,080 --> 00:23:47,080
case.

331
00:23:47,080 --> 00:23:53,320
And that was a real consideration for me thinking about the rollout and adoption of this, where

332
00:23:53,320 --> 00:23:57,440
some organizations are going to have people that desperately want to get their hands on

333
00:23:57,440 --> 00:24:02,600
these tools and other organizations are going to have people who quite simply are never

334
00:24:02,600 --> 00:24:03,840
going to use these tools.

335
00:24:03,840 --> 00:24:10,080
They're going to stay in their lane and just use the tools that they know work for them.

336
00:24:10,080 --> 00:24:13,960
If they had to upgrade because one person wanted it and everyone suddenly has to pay

337
00:24:13,960 --> 00:24:18,480
$30 per month, that felt like it was going to be a step too far.

338
00:24:18,480 --> 00:24:22,500
Fortunately, the licensing on this doesn't work like that.

339
00:24:22,500 --> 00:24:29,720
In terms of initial thoughts, my immediate impression is boy, oh boy, are we buying the

340
00:24:29,720 --> 00:24:31,920
roadmap here?

341
00:24:31,920 --> 00:24:37,820
This is an unfinished product.

342
00:24:37,820 --> 00:24:40,640
This is early.

343
00:24:40,640 --> 00:24:45,640
And is it particularly capable at this point?

344
00:24:45,640 --> 00:24:48,920
No, not in my limited trials.

345
00:24:48,920 --> 00:24:50,680
Admittedly, it's very limited.

346
00:24:50,680 --> 00:24:58,680
But a couple of examples, they have the write an email for me tool in Gmail.

347
00:24:58,680 --> 00:25:01,120
Fine.

348
00:25:01,120 --> 00:25:02,280
It's okay.

349
00:25:02,280 --> 00:25:04,080
It's the same as using Bard.

350
00:25:04,080 --> 00:25:09,200
The output, the quality of the outputs is exactly the same as using Bard.

351
00:25:09,200 --> 00:25:13,120
There didn't seem to be any additional functionality that I came across in early usage.

352
00:25:13,120 --> 00:25:16,920
Now they do say that more features are coming, but that's why I say it's like buying the

353
00:25:16,920 --> 00:25:17,920
roadmap.

354
00:25:17,920 --> 00:25:23,400
You're buying the future of what's coming, not the reality of where it is today.

355
00:25:23,400 --> 00:25:29,240
The notion that it can create slide decks for you, as far as I can see, that's not a

356
00:25:29,240 --> 00:25:30,240
reality yet.

357
00:25:30,240 --> 00:25:36,840
What you get if you go into slides is an image generator.

358
00:25:36,840 --> 00:25:41,280
You can put in a prompt like you would in mid-journey or stable diffusion, and it will

359
00:25:41,280 --> 00:25:45,040
create an image for you and you can add that into your deck.

360
00:25:45,040 --> 00:25:50,520
In Sheets, you can write a prompt and ask it to do a thing for you.

361
00:25:50,520 --> 00:25:59,940
Maybe create a project plan for maybe a web development project or something like that.

362
00:25:59,940 --> 00:26:07,560
My testing of that was that it was kind of okay, but I would probably do a better job

363
00:26:07,560 --> 00:26:08,560
myself.

364
00:26:08,560 --> 00:26:15,640
I asked it to do a project plan for a web development project and it created the table,

365
00:26:15,640 --> 00:26:19,880
assigned some timelines and some tasks.

366
00:26:19,880 --> 00:26:30,320
If you use a Sana or another project management tool that has AI baked into it, the project

367
00:26:30,320 --> 00:26:34,560
plans that you get from those AI-generated project plans are better, more detailed, more

368
00:26:34,560 --> 00:26:36,080
comprehensive.

369
00:26:36,080 --> 00:26:43,480
The one thing that I thought it did do, which was pretty interesting, was brought in some

370
00:26:43,480 --> 00:26:47,720
elements of Google Sheets that I know are there, but I never use.

371
00:26:47,720 --> 00:26:53,960
For example, the tabs or the ability to, like the buttons that you can add to cells.

372
00:26:53,960 --> 00:26:55,480
They announced it a couple of years ago.

373
00:26:55,480 --> 00:27:01,200
I don't know if you actively use them in your Sheets poll, but I certainly don't.

374
00:27:01,200 --> 00:27:04,240
I knew they were there, but they're great for project management.

375
00:27:04,240 --> 00:27:07,880
They're just not a feature that I'd ever incorporated into my workflow.

376
00:27:07,880 --> 00:27:11,800
It brought them into the project plan.

377
00:27:11,800 --> 00:27:13,320
I thought, okay, that's quite neat.

378
00:27:13,320 --> 00:27:19,680
It's making me use more of the functionality that it has available to it.

379
00:27:19,680 --> 00:27:26,960
It isn't creating particularly, I asked it to create a Gantt chart, didn't do that at

380
00:27:26,960 --> 00:27:29,360
all, so that was a complete fail.

381
00:27:29,360 --> 00:27:33,320
It can't create a slide deck right now.

382
00:27:33,320 --> 00:27:38,840
It can do image generation for $30 or £30 a month.

383
00:27:38,840 --> 00:27:43,720
I'm not sure I would upgrade to it.

384
00:27:43,720 --> 00:27:51,360
The moment I'm on day two of my trial, so who knows, speak to me in two weeks' time

385
00:27:51,360 --> 00:27:58,080
and maybe I'm more excited and more enthusiastic, but right now, it's somewhat lacking.

386
00:27:58,080 --> 00:27:59,080
I'd agree.

387
00:27:59,080 --> 00:28:03,080
I think that 14-day trial is going to turn out to be a real risk for Google because I'm

388
00:28:03,080 --> 00:28:08,760
not sure businesses are going to be able to tap into enough power to make them to want

389
00:28:08,760 --> 00:28:10,760
to commit to paying for it.

390
00:28:10,760 --> 00:28:15,640
When I think about how best to roll this out within an organization, me and you are haley

391
00:28:15,640 --> 00:28:20,880
nerds, so we are in a really strong position to start playing with it and try and get it

392
00:28:20,880 --> 00:28:21,880
to do things.

393
00:28:21,880 --> 00:28:28,600
A big, big gap for most businesses is going to be training and empowering people to leverage

394
00:28:28,600 --> 00:28:31,720
the tools in their daily workflows.

395
00:28:31,720 --> 00:28:35,840
All of that is a massive barrier and I'm not sure how much progress a business can make

396
00:28:35,840 --> 00:28:38,800
in 14 days to go, all right, let's commit to this.

397
00:28:38,800 --> 00:28:43,480
The best case scenario for Google, but to a certain extent for businesses as well, is

398
00:28:43,480 --> 00:28:50,640
going to be to create some sort of AI working group within a business that's a small collection

399
00:28:50,640 --> 00:28:55,780
of users that are well positioned to get benefits through the tools, that have interests in

400
00:28:55,780 --> 00:29:01,120
trying to want to get to grips with the tools, that basically pilot this on behalf of the

401
00:29:01,120 --> 00:29:06,160
rest of the business to figure out if you can maximize the use of it to actually get

402
00:29:06,160 --> 00:29:10,480
value out of that 30 pounds per user per month.

403
00:29:10,480 --> 00:29:12,600
So yeah, we can have a play.

404
00:29:12,600 --> 00:29:17,600
I think we'll probably test it reasonably well in that 14 day period, but most businesses

405
00:29:17,600 --> 00:29:20,280
are not going to have to, so they are not going to be able to.

406
00:29:20,280 --> 00:29:27,000
So they're either going to have to create that working group and suck up a bit of costs

407
00:29:27,000 --> 00:29:28,000
or leave it.

408
00:29:28,000 --> 00:29:31,040
So I think it would be really interesting to see what the uptake of this actually ends

409
00:29:31,040 --> 00:29:32,040
up being.

410
00:29:32,040 --> 00:29:36,560
Yeah, it's going to be a big commitment for a lot of businesses if they don't see value

411
00:29:36,560 --> 00:29:38,960
from it quickly.

412
00:29:38,960 --> 00:29:47,840
It's also interesting, and it comes a week after rumors started circulating about the

413
00:29:47,840 --> 00:29:52,040
expected price of paid plans for Claude.

414
00:29:52,040 --> 00:30:02,880
So it was rumored this week that Claude.ai will cost users $50 per month, which compared

415
00:30:02,880 --> 00:30:13,440
to all of these other tools is quite steep, $20 per month for ChatGPT+, and $50 is a big

416
00:30:13,440 --> 00:30:16,840
jump, copilot rumored to be 30.

417
00:30:16,840 --> 00:30:22,760
But strangely, I would pay $50 per month for Claude without batting an eyelid.

418
00:30:22,760 --> 00:30:25,760
We had this discussion, didn't we, over WhatsApp?

419
00:30:25,760 --> 00:30:28,640
I honestly don't agree.

420
00:30:28,640 --> 00:30:34,640
I'm really looking for these tools to be a little bit lower cost than $50 a month, because

421
00:30:34,640 --> 00:30:40,440
I think one of the biggest challenges is going to be shifting user behavior to have them

422
00:30:40,440 --> 00:30:42,480
really leverage these tools.

423
00:30:42,480 --> 00:30:48,280
I think $50 per month in your workflow and the way that you work is probably good value

424
00:30:48,280 --> 00:30:53,280
for you, but I think for most people it's not going to be.

425
00:30:53,280 --> 00:30:58,720
If they try and charge too much for this too early, I think they're going to blunt adoption,

426
00:30:58,720 --> 00:31:02,960
because a lot of small and mid-sized businesses are not going to be able to take the punt

427
00:31:02,960 --> 00:31:08,140
on those types of costs without having some idea of how they're going to get and measure

428
00:31:08,140 --> 00:31:09,240
return on investment.

429
00:31:09,240 --> 00:31:14,640
I think the return on investment will be there, but only if you, as we said, train people

430
00:31:14,640 --> 00:31:19,160
and empower them to find effective use cases for using the tools.

431
00:31:19,160 --> 00:31:20,720
That's a massive barrier.

432
00:31:20,720 --> 00:31:24,080
I think a lot of software companies going, right, we want to charge this.

433
00:31:24,080 --> 00:31:28,960
Here you go, here's the tool, expecting that somehow the businesses are going to figure

434
00:31:28,960 --> 00:31:34,160
out how to leverage it effectively across a wider user base across their employees.

435
00:31:34,160 --> 00:31:36,680
I don't think it's going to work out like that.

436
00:31:36,680 --> 00:31:41,440
A little plug, Martin and I do offer AI training and consultancy for businesses that want to

437
00:31:41,440 --> 00:31:47,560
unlock the potential of these systems to massively boost their productivity and efficiency.

438
00:31:47,560 --> 00:31:50,280
But I think that's going to be a critical piece, right, because without it, you're going

439
00:31:50,280 --> 00:31:52,520
to just be throwing money away because AI is calling.

440
00:31:52,520 --> 00:31:56,080
You thought you should buy some, right?

441
00:31:56,080 --> 00:32:01,920
Let's move on to our last story today, which is a good example of AI being pretty cool,

442
00:32:01,920 --> 00:32:04,000
but not quite as cool as it sounds.

443
00:32:04,000 --> 00:32:06,920
Martin, tell us about, is it ideogram?

444
00:32:06,920 --> 00:32:07,920
Is that how it's pronounced?

445
00:32:07,920 --> 00:32:08,920
I don't know.

446
00:32:08,920 --> 00:32:10,920
Ideogram, ideogram, tell us about it.

447
00:32:10,920 --> 00:32:14,800
Ideogram, ideogram, take your pit.

448
00:32:14,800 --> 00:32:22,520
It's a new startup, which is the brainchild of some ex-Google brain researchers, and they've

449
00:32:22,520 --> 00:32:33,480
managed to secure $16.5 million in seed funding from A16Z and Index Ventures.

450
00:32:33,480 --> 00:32:35,880
And this is an AI image generator.

451
00:32:35,880 --> 00:32:45,600
So it's a new tool, new model that has a specific unique feature.

452
00:32:45,600 --> 00:32:54,520
So unlike other AI image generators, ideogram is generating reliable text.

453
00:32:54,520 --> 00:32:58,080
I say reliable in air quotes.

454
00:32:58,080 --> 00:33:00,160
That's its special trick.

455
00:33:00,160 --> 00:33:07,200
That's its party trick, is it can do text in images where other models suffer horribly.

456
00:33:07,200 --> 00:33:15,600
So it's available in beta via a web app, and it has some preset image generation styles,

457
00:33:15,600 --> 00:33:18,180
one of which is typography.

458
00:33:18,180 --> 00:33:23,640
And this feature allows users to produce lettering in different colors, fonts, sizes, styles

459
00:33:23,640 --> 00:33:27,240
using prompts.

460
00:33:27,240 --> 00:33:33,120
You can also ask it to just create images of somebody holding a sign with some text

461
00:33:33,120 --> 00:33:37,120
on it, and it will do exactly that.

462
00:33:37,120 --> 00:33:39,320
And there's some pretty cool examples.

463
00:33:39,320 --> 00:33:45,440
If you log in, you can see almost like a real-time feed of favorited images, ones that people

464
00:33:45,440 --> 00:33:47,180
think are particularly good.

465
00:33:47,180 --> 00:33:53,720
One example that we've got, and we'll put this onto the website show notes, is an image

466
00:33:53,720 --> 00:34:01,360
of Star Wars Mandalorian holding a big tub of protein powder.

467
00:34:01,360 --> 00:34:04,880
And across the front it says, this is the way.

468
00:34:04,880 --> 00:34:05,880
Very clever.

469
00:34:05,880 --> 00:34:08,720
So way, W-H-E-Y, right?

470
00:34:08,720 --> 00:34:11,120
Yeah, whey protein.

471
00:34:11,120 --> 00:34:12,800
Ha ha ha.

472
00:34:12,800 --> 00:34:17,440
And then another one, which is of a big wall.

473
00:34:17,440 --> 00:34:25,080
And the prompt just said, your prompt are public, you weirdos, Banksy Graffiti.

474
00:34:25,080 --> 00:34:30,280
And then on the wall in spray paint, it says, your prompt are public, you weirdos.

475
00:34:30,280 --> 00:34:32,100
And it's perfectly legible.

476
00:34:32,100 --> 00:34:35,280
So in those two examples, it's done a very good job.

477
00:34:35,280 --> 00:34:43,620
In trying it for the podcast, I added the prompt, cheering crowds waving signs saying

478
00:34:43,620 --> 00:34:46,460
artificially intelligent marketing.

479
00:34:46,460 --> 00:34:51,720
And when you generate this, it creates four images, kind of very standard across these

480
00:34:51,720 --> 00:34:53,400
kind of models.

481
00:34:53,400 --> 00:34:56,860
The first one nails the brief.

482
00:34:56,860 --> 00:35:02,040
There is a crowd of people and in the foreground there is a woman holding a sign that says

483
00:35:02,040 --> 00:35:05,400
artificially intelligent marketing.

484
00:35:05,400 --> 00:35:11,080
But from there on, it, well, it misses the mark substantially.

485
00:35:11,080 --> 00:35:14,640
The next one says artificially long intelligent.

486
00:35:14,640 --> 00:35:17,440
And then the next one says something that's absolute nonsense.

487
00:35:17,440 --> 00:35:25,260
And the final one says there is no planet B. So I'm not 100% convinced by its reliability

488
00:35:25,260 --> 00:35:26,340
just yet.

489
00:35:26,340 --> 00:35:29,000
It is somewhat hit and miss.

490
00:35:29,000 --> 00:35:34,800
And another observation that I made in testing was, does suffer from the hands and the face

491
00:35:34,800 --> 00:35:39,640
ghoulishness that we've seen on other AI image generator models.

492
00:35:39,640 --> 00:35:41,520
Yeah, that's a great summary.

493
00:35:41,520 --> 00:35:42,520
Thanks, Myan Earth.

494
00:35:42,520 --> 00:35:43,740
Haven't played with it a bit.

495
00:35:43,740 --> 00:35:46,840
My experiences have been the same as yours.

496
00:35:46,840 --> 00:35:51,600
And looking, even just looking at the images that you generated, I get the sense that they've

497
00:35:51,600 --> 00:35:57,320
deliberately tried to train this model on signage and writing.

498
00:35:57,320 --> 00:36:01,520
And I think that's one of the reasons it's actually quite bad at people and hands and

499
00:36:01,520 --> 00:36:02,520
all the other things.

500
00:36:02,520 --> 00:36:09,180
Like it's, it's like image generators from like a year ago, you know, almost in terms

501
00:36:09,180 --> 00:36:12,000
of its general image generation capabilities.

502
00:36:12,000 --> 00:36:15,400
But it's really good at text, probably because it was trained on text.

503
00:36:15,400 --> 00:36:21,560
It's interesting that even the best artificially intelligent marketing placard that was generated

504
00:36:21,560 --> 00:36:24,360
spell artificially wrong, right?

505
00:36:24,360 --> 00:36:28,640
So that's still pretty much a big issue, even though it's quite clear in the prompt because

506
00:36:28,640 --> 00:36:31,200
you spelled it right in the prompt.

507
00:36:31,200 --> 00:36:37,480
And even in some of the like outputs that you got, there is no planet B sign that you

508
00:36:37,480 --> 00:36:38,480
got.

509
00:36:38,480 --> 00:36:43,440
I think that's a real one from the training data and then the image next to it is clearly

510
00:36:43,440 --> 00:36:47,960
based on that, but is just some weird amalgamation.

511
00:36:47,960 --> 00:36:51,840
So as a user, you probably don't need to care too much about that.

512
00:36:51,840 --> 00:36:58,720
But in essence, the summary here is it's much better at text than any of the other tools.

513
00:36:58,720 --> 00:37:05,680
You can, with some brute force, get a good and legible output, including some of the

514
00:37:05,680 --> 00:37:09,600
sort of interesting and funny examples of mine gave us at the beginning.

515
00:37:09,600 --> 00:37:15,000
But if you think you can go boot this up and start producing production ready signage and

516
00:37:15,000 --> 00:37:19,240
logos, I would seriously question your ability to do that.

517
00:37:19,240 --> 00:37:24,960
I think there's much better, but you're going to have to prompt a lot in order to get what

518
00:37:24,960 --> 00:37:26,260
you want.

519
00:37:26,260 --> 00:37:32,780
And it's going to mess up quite a bit of stuff and worth keeping in mind that us as a group

520
00:37:32,780 --> 00:37:36,520
of weirdos have to keep in mind that our prompts are public because whilst this is a free tool

521
00:37:36,520 --> 00:37:43,200
right now, whatever prompts you put in, they do end up on the sort of recent examples board

522
00:37:43,200 --> 00:37:45,280
within the Ediegram tool.

523
00:37:45,280 --> 00:37:50,640
So when you're having a play with this, do keep in mind people can see what you're generating

524
00:37:50,640 --> 00:37:52,600
and that's probably why it's free.

525
00:37:52,600 --> 00:37:58,240
And when you want to be able to generate private images, be expecting to pay for that when

526
00:37:58,240 --> 00:38:00,360
the paid version is rolled out.

527
00:38:00,360 --> 00:38:04,360
I think one of the things I'm excited with here is whether or not this will nudge the

528
00:38:04,360 --> 00:38:10,920
power tools out there that kick drops the mid journeys to try and improve their text

529
00:38:10,920 --> 00:38:15,840
generation or whether or not we'll end up with different image generation AI tools for

530
00:38:15,840 --> 00:38:17,320
different use cases.

531
00:38:17,320 --> 00:38:22,000
But one thing I really want to see is better text because it does limit your ability to

532
00:38:22,000 --> 00:38:26,660
produce a whole suite of images that would be interesting, certainly as a marketer when

533
00:38:26,660 --> 00:38:29,640
it comes to using AI driven image generation tools.

534
00:38:29,640 --> 00:38:32,120
Well, it's the next frontier, isn't it?

535
00:38:32,120 --> 00:38:34,360
Text in AI images.

536
00:38:34,360 --> 00:38:36,660
It's the next logical progression.

537
00:38:36,660 --> 00:38:38,480
We know that it is capable.

538
00:38:38,480 --> 00:38:44,800
Google about a year ago, maybe over a year ago, published their research paper on their

539
00:38:44,800 --> 00:38:48,720
image generation model party, PARTI.

540
00:38:48,720 --> 00:38:57,440
And that has a 20 billion parameter model, which was able to create images in text and

541
00:38:57,440 --> 00:39:00,240
on signs, things like that, very well.

542
00:39:00,240 --> 00:39:05,080
Now, obviously in the paper, they were probably selecting some of the better examples and

543
00:39:05,080 --> 00:39:07,880
on the blog post talking about it.

544
00:39:07,880 --> 00:39:13,560
We know it's possible, but 20 billion parameters is a particularly large model for image generation.

545
00:39:13,560 --> 00:39:14,560
Yeah, absolutely.

546
00:39:14,560 --> 00:39:18,560
I think we're going to see a lot here over the next six months.

547
00:39:18,560 --> 00:39:23,000
And hopefully this will be one of the things that helps kickstart some of those improvements.

548
00:39:23,000 --> 00:39:24,000
Right.

549
00:39:24,000 --> 00:39:28,800
So that's all of the stories that we're covering this week, really focusing in on the big ones

550
00:39:28,800 --> 00:39:30,560
you mentioned at the beginning.

551
00:39:30,560 --> 00:39:33,560
From here, we're going to move into our interview.

552
00:39:33,560 --> 00:39:36,400
Martin, tell us a bit more about your interview this week.

553
00:39:36,400 --> 00:39:39,920
Yeah, a good interview with Jeff Coyle.

554
00:39:39,920 --> 00:39:45,680
Now, Jeff has forgotten more about SEO than I'll ever know.

555
00:39:45,680 --> 00:39:52,720
He's the co-founder of MarketMuse, a computer scientist by trade.

556
00:39:52,720 --> 00:40:00,720
And we had a great conversation about how AI is changing search engine optimization and

557
00:40:00,720 --> 00:40:02,160
content marketing.

558
00:40:02,160 --> 00:40:08,120
But I started the conversation by asking him to tell us a little bit about what MarketMuse

559
00:40:08,120 --> 00:40:10,720
does.

560
00:40:10,720 --> 00:40:15,200
So MarketMuse is a content intelligence and content strategy software platform.

561
00:40:15,200 --> 00:40:20,400
So it's focused on giving insights as to what people should be writing, why they should

562
00:40:20,400 --> 00:40:21,400
be writing it.

563
00:40:21,400 --> 00:40:22,880
What should they need to write?

564
00:40:22,880 --> 00:40:28,700
And then taking them as far down that journey of creation or updating or optimization as

565
00:40:28,700 --> 00:40:29,880
they want.

566
00:40:29,880 --> 00:40:34,400
So we've been around for as long as nine and a half years.

567
00:40:34,400 --> 00:40:38,840
I joined the company when we took it to market and went full time about eight and a half

568
00:40:38,840 --> 00:40:39,840
years ago.

569
00:40:39,840 --> 00:40:46,440
The challenge that I saw consistently was that teams didn't have direction on what to

570
00:40:46,440 --> 00:40:51,440
create and why they should be creating it or what to update and what would be the highest

571
00:40:51,440 --> 00:40:53,840
priority for updating.

572
00:40:53,840 --> 00:41:00,040
And those were really where a lot of the challenges that search engine optimization teams, content

573
00:41:00,040 --> 00:41:06,360
marketing teams, demand gen, editorial had getting along effectively.

574
00:41:06,360 --> 00:41:09,500
We think you should write this page and they're like, we don't want to write that page.

575
00:41:09,500 --> 00:41:10,760
Why would we write it?

576
00:41:10,760 --> 00:41:12,520
And it's like, because keywords?

577
00:41:12,520 --> 00:41:15,200
And then they're like, that's not enough of a good reason.

578
00:41:15,200 --> 00:41:22,160
And my goal kind of as having lived through that and worked on manual processes was what

579
00:41:22,160 --> 00:41:24,640
of these things can be optimized?

580
00:41:24,640 --> 00:41:27,040
What of these things can be automated?

581
00:41:27,040 --> 00:41:28,880
And that was the journey of Market Muse.

582
00:41:28,880 --> 00:41:33,720
And so we're doing a lot of insight in things that are automating people's day to day, but

583
00:41:33,720 --> 00:41:38,740
also giving them ideas that they wouldn't have gotten elsewhere.

584
00:41:38,740 --> 00:41:43,640
So you talk about the giving the insights and the intelligence.

585
00:41:43,640 --> 00:41:48,800
Just break that down for us in terms of, if I was to sign up for an account today, what

586
00:41:48,800 --> 00:41:51,840
am I going to get as a content marketer?

587
00:41:51,840 --> 00:41:54,860
So I like to think about it as site level and page level.

588
00:41:54,860 --> 00:42:00,320
That's the easiest way to think of content strategy versus more, I'm touching a page,

589
00:42:00,320 --> 00:42:03,400
I'm updating a page, I'm making a page beautiful.

590
00:42:03,400 --> 00:42:07,180
So we have solutions at both of those kind of zones.

591
00:42:07,180 --> 00:42:12,620
So in the page level, you may already have a draft, no matter where you got it from,

592
00:42:12,620 --> 00:42:14,840
it may be a page that you invested in years ago.

593
00:42:14,840 --> 00:42:18,200
It may be something you're working on or recently published.

594
00:42:18,200 --> 00:42:20,240
What insights do you want from that?

595
00:42:20,240 --> 00:42:21,240
How do I make that better?

596
00:42:21,240 --> 00:42:23,320
How do I make it more comprehensive?

597
00:42:23,320 --> 00:42:26,340
How do I make it more aligned with today's intent?

598
00:42:26,340 --> 00:42:32,240
Or how do I ensure that it is equal to or better than all of my competitors every time?

599
00:42:32,240 --> 00:42:37,960
So that's where we provide insights, but not just by copying your competitor, by understanding

600
00:42:37,960 --> 00:42:40,640
what it means to be an expert.

601
00:42:40,640 --> 00:42:43,680
And then at the site level, you're looking at what you currently have.

602
00:42:43,680 --> 00:42:49,320
So you want to know what breadth and depth of coverage you have, the quality, if you

603
00:42:49,320 --> 00:42:53,800
have any momentum, and are there off-page factors you should consider.

604
00:42:53,800 --> 00:43:00,920
So we do effectively a on-demand content inventory to say, if I were to focus on this topic,

605
00:43:00,920 --> 00:43:06,280
we can give you an idea of what coverage you have, how much momentum you have, and we are

606
00:43:06,280 --> 00:43:09,080
the only people that personalize a difficulty score.

607
00:43:09,080 --> 00:43:12,640
So if you're in the SEO world, everybody's got their own scores, everybody's got their

608
00:43:12,640 --> 00:43:14,600
own difficulty.

609
00:43:14,600 --> 00:43:19,000
We personalize that so you can say, how hard will this be for us?

610
00:43:19,000 --> 00:43:22,680
Is this impossible or is this a quick win?

611
00:43:22,680 --> 00:43:27,760
And what we found is by looking at the quality of the content and the coverage levels, we're

612
00:43:27,760 --> 00:43:35,080
able to very effectively predict outcomes and say, hey, you need a seven to 10 page

613
00:43:35,080 --> 00:43:38,600
cluster on this to even start putting points on the board.

614
00:43:38,600 --> 00:43:43,760
It really changes the way organizations work because so many teams, they think that writing

615
00:43:43,760 --> 00:43:48,400
content and publishing it is somewhat of a dice roll.

616
00:43:48,400 --> 00:43:51,400
They don't know, is it going to do well?

617
00:43:51,400 --> 00:43:54,200
Maybe if we look at it next year, we'll know.

618
00:43:54,200 --> 00:44:00,880
And being able to get to the point where they're asking for budget with effective predictions

619
00:44:00,880 --> 00:44:02,640
really changes the way teams work.

620
00:44:02,640 --> 00:44:07,740
How is AI baked into market views?

621
00:44:07,740 --> 00:44:12,520
So it has always been an AI first platform and that is 10 years ago.

622
00:44:12,520 --> 00:44:19,920
And then you have things like topic modeling, graph analysis, and entity analysis.

623
00:44:19,920 --> 00:44:25,480
And now we've innovated over the past six, seven, eight years in the field of generation

624
00:44:25,480 --> 00:44:31,520
and how can we use generative technology to empower not just writing pages.

625
00:44:31,520 --> 00:44:37,440
People often mistake that text generation means going to draft.

626
00:44:37,440 --> 00:44:43,080
There's a lot of use cases for natural language processing, natural language generation, that

627
00:44:43,080 --> 00:44:50,920
output, even giving a writer a prompt, building briefs, building outlines, making it so that

628
00:44:50,920 --> 00:44:56,520
their job staying in the loop and exhibiting their expertise isn't stunted.

629
00:44:56,520 --> 00:45:03,080
And so when you take technology and data and now it's being used as part of conversations

630
00:45:03,080 --> 00:45:07,600
where it never would have been in the past, that's where I see market news fitting in

631
00:45:07,600 --> 00:45:14,200
is to say that editor who's the most brilliant subject matter expert on this topic wants

632
00:45:14,200 --> 00:45:17,640
data to be part of their discussion.

633
00:45:17,640 --> 00:45:21,160
And this is really the last couple of years has been the first time that that's ever been

634
00:45:21,160 --> 00:45:23,440
a reality.

635
00:45:23,440 --> 00:45:29,960
And I think so much of that requirement to demonstrate expertise is becoming increasingly

636
00:45:29,960 --> 00:45:30,960
important.

637
00:45:30,960 --> 00:45:38,600
And the Google guidelines with the eat and the experience expertise, all of that is becoming

638
00:45:38,600 --> 00:45:39,960
increasingly important.

639
00:45:39,960 --> 00:45:46,400
And I think marketers are just, people have been playing SEO and doing SEO hardcore.

640
00:45:46,400 --> 00:45:50,640
They know this, they're staying on top of the Google guidelines, but lots of copyrights

641
00:45:50,640 --> 00:45:58,000
don't understand the importance of going deep on expertise in particular.

642
00:45:58,000 --> 00:46:00,920
Where do you think this is taking us?

643
00:46:00,920 --> 00:46:08,280
If we're looking 18, 24 months out from today, where do you think this is taking us from

644
00:46:08,280 --> 00:46:10,080
a search perspective?

645
00:46:10,080 --> 00:46:16,360
The biggest change I'll highlight would be this is going to be the time period where

646
00:46:16,360 --> 00:46:26,560
the rise of the expert happens and content quality is no longer an option.

647
00:46:26,560 --> 00:46:32,720
And if you were doing things to get away with mediocre or heaven forbid worse than mediocre

648
00:46:32,720 --> 00:46:38,000
content, that is going to be devastating for you.

649
00:46:38,000 --> 00:46:43,760
It could potentially make you ignorable, but it may actually negatively impact your brand

650
00:46:43,760 --> 00:46:46,880
at a level that's going to be difficult to remediate.

651
00:46:46,880 --> 00:46:53,420
And the unique hilarious part about this is it's getting back to early days of editorial,

652
00:46:53,420 --> 00:47:01,480
early days of journalism, early days of search, where you were thinking critically about user

653
00:47:01,480 --> 00:47:06,120
intents and you have to bring your A game every time.

654
00:47:06,120 --> 00:47:12,840
And how can we do that while also being fast and while also automating?

655
00:47:12,840 --> 00:47:15,360
It becomes the challenge for every single team.

656
00:47:15,360 --> 00:47:20,760
I think people have to be a little bit understanding that they're going to make a lot of mistakes.

657
00:47:20,760 --> 00:47:22,360
And so I'm seeing this a lot.

658
00:47:22,360 --> 00:47:27,400
I cringe every time I look at my Facebook or Instagram and I see, hey, here's how to

659
00:47:27,400 --> 00:47:30,160
do keyword research with ChatGPG.

660
00:47:30,160 --> 00:47:32,160
And I'm like, don't do it.

661
00:47:32,160 --> 00:47:35,040
No, you don't get it.

662
00:47:35,040 --> 00:47:37,440
Because the outputs look so good.

663
00:47:37,440 --> 00:47:40,920
The outputs look legitimate.

664
00:47:40,920 --> 00:47:48,180
And so a new wave of sharp people are going to be building technology and implementing

665
00:47:48,180 --> 00:47:52,980
technology to take advantage of people implementing it poorly.

666
00:47:52,980 --> 00:47:55,080
And that's a wild thing to think about.

667
00:47:55,080 --> 00:48:02,240
So it's effectively saying, if you're using keyword density or you're using term frequency

668
00:48:02,240 --> 00:48:10,920
to optimize your pages as an SEO association, it's really easy to beat you in organic search.

669
00:48:10,920 --> 00:48:13,920
If you're a shark, if you've been doing this for a long time, right?

670
00:48:13,920 --> 00:48:19,120
And you're going to now have that.

671
00:48:19,120 --> 00:48:24,060
It's really easy to beat you if you're using generative AI badly.

672
00:48:24,060 --> 00:48:27,160
It's like today's spam, today's bad spammers.

673
00:48:27,160 --> 00:48:32,880
Like it's really, I mean, and there's even new waves of things spiking in the SEO community

674
00:48:32,880 --> 00:48:36,860
and in the content community where people are saying, hey, this is super easy.

675
00:48:36,860 --> 00:48:40,840
Just launch 10,000 pages and go for it.

676
00:48:40,840 --> 00:48:46,560
I would love nothing more than for my clients, competitors to go do those things because

677
00:48:46,560 --> 00:48:53,480
it's like, I'm just waiting for the sword of Damocles to no longer have a rope, right?

678
00:48:53,480 --> 00:48:56,480
And I think it has a rope in the end.

679
00:48:56,480 --> 00:48:59,480
The thing that cuts your head off.

680
00:48:59,480 --> 00:49:06,360
And that's what you're experiencing right now is I'm watching train wrecks.

681
00:49:06,360 --> 00:49:09,000
I'm watching them before they happen.

682
00:49:09,000 --> 00:49:11,400
And I'm just like train wrecks coming.

683
00:49:11,400 --> 00:49:16,720
The product reviews update was a great example of that where we were ahead of the game and

684
00:49:16,720 --> 00:49:23,720
trying to advise all of our clients to get religion on that a year ahead of time.

685
00:49:23,720 --> 00:49:29,760
Because at some point, Google is not going to be okay with content that pretends to review

686
00:49:29,760 --> 00:49:32,960
something that doesn't actually review it, right?

687
00:49:32,960 --> 00:49:36,040
You're going to see that times a million.

688
00:49:36,040 --> 00:49:39,360
The search generative experience with Google, right?

689
00:49:39,360 --> 00:49:42,160
It changes every day, right?

690
00:49:42,160 --> 00:49:46,760
Currently the three units of pay, if you don't know what search generative experience is,

691
00:49:46,760 --> 00:49:52,960
it's the generated search results on Google, which there are some, they're in a test plan

692
00:49:52,960 --> 00:49:55,240
clearly where they're assessing it.

693
00:49:55,240 --> 00:49:58,480
The query deserves to have this experience.

694
00:49:58,480 --> 00:50:03,480
And then they're showing effectively a mediocre book report on the topic.

695
00:50:03,480 --> 00:50:07,920
And then sometimes they're showing products, sometimes they're showing some pages from

696
00:50:07,920 --> 00:50:10,040
the learner's journey.

697
00:50:10,040 --> 00:50:13,960
If you want to, you can type in another question and continue that experience.

698
00:50:13,960 --> 00:50:16,840
And then you'll go into their normal search result.

699
00:50:16,840 --> 00:50:18,120
It's clearly a work in progress.

700
00:50:18,120 --> 00:50:21,760
And Google has shown to us that they get things right over time.

701
00:50:21,760 --> 00:50:23,040
It takes them some time.

702
00:50:23,040 --> 00:50:31,560
One thing I'm seeing is that Google is referencing what they consider to be great resources,

703
00:50:31,560 --> 00:50:32,560
right?

704
00:50:32,560 --> 00:50:37,840
And one example of this for e-commerce sites is, but they're actually anchoring down to

705
00:50:37,840 --> 00:50:39,160
the text.

706
00:50:39,160 --> 00:50:43,820
So some of these e-commerce sites, their category pages are cracking and it's anchoring down

707
00:50:43,820 --> 00:50:44,820
to the footer.

708
00:50:44,820 --> 00:50:53,560
So you click, so you're in this like funky, unexpanded filter funnel experience.

709
00:50:53,560 --> 00:50:58,680
You basically just like, you know, so there's so many different layers of this.

710
00:50:58,680 --> 00:51:05,160
Because imagine, right, if you put your FAQ below the fold on your category page and Google's

711
00:51:05,160 --> 00:51:07,200
like, yeah, that was a good answer.

712
00:51:07,200 --> 00:51:08,200
I got that.

713
00:51:08,200 --> 00:51:12,480
They're actually dragging you down to the bottom on it.

714
00:51:12,480 --> 00:51:14,360
And nobody's ready for that right now.

715
00:51:14,360 --> 00:51:18,560
So there's just a lot of things that people aren't ready for the implications of the bad

716
00:51:18,560 --> 00:51:27,120
decisions they've made to put the customer's experience second.

717
00:51:27,120 --> 00:51:32,960
And I think that what the big change to come somewhere with this landing pad is every business

718
00:51:32,960 --> 00:51:39,520
is going to have to examine where they've made a decision for traffic, for lead gen,

719
00:51:39,520 --> 00:51:44,640
for e-commerce sales that wasn't customer first.

720
00:51:44,640 --> 00:51:46,920
That's going to end up being the story.

721
00:51:46,920 --> 00:51:49,480
It's going to take us a couple of years to get there.

722
00:51:49,480 --> 00:51:52,680
But over time, that's going to end up being the story.

723
00:51:52,680 --> 00:51:58,320
Yeah, I think SGE is, we don't have access to it in the UK at the moment.

724
00:51:58,320 --> 00:52:06,920
So I'm very much following from afar, just watching the likes of Lily Ray and other commenters

725
00:52:06,920 --> 00:52:10,320
just doing all of these experiments every day.

726
00:52:10,320 --> 00:52:16,360
And I do want to dive into it because it is fundamentally going to change the way that

727
00:52:16,360 --> 00:52:20,400
organic content, how we engage with the organic.

728
00:52:20,400 --> 00:52:26,720
I think in some respects, Google for a long time, Google generates money through selling

729
00:52:26,720 --> 00:52:27,720
ad units, right?

730
00:52:27,720 --> 00:52:31,080
So they want to deliver you the best user experience.

731
00:52:31,080 --> 00:52:36,760
That just so happens to have been historically by giving you organic content and the best

732
00:52:36,760 --> 00:52:38,560
of that content.

733
00:52:38,560 --> 00:52:43,440
But they don't really care about organic truly.

734
00:52:43,440 --> 00:52:45,440
They want to sort out the, what is it?

735
00:52:45,440 --> 00:52:47,160
What's the mission?

736
00:52:47,160 --> 00:52:51,480
To organize the world's information.

737
00:52:51,480 --> 00:52:59,160
So if they can do that and organic jumps 10% further down the page or 50% further down

738
00:52:59,160 --> 00:53:04,160
the page in some instances that I've seen, while still presenting ad units, in fact still

739
00:53:04,160 --> 00:53:05,840
getting more ad units.

740
00:53:05,840 --> 00:53:11,960
Those Google shopping ad units are all over SGE as far as I can see.

741
00:53:11,960 --> 00:53:13,480
Organic is a lower priority.

742
00:53:13,480 --> 00:53:20,080
As long as they can access the information to then put that into the SGE, then they're

743
00:53:20,080 --> 00:53:21,080
happy.

744
00:53:21,080 --> 00:53:24,040
This isn't search results, right?

745
00:53:24,040 --> 00:53:27,600
So if someone's coming to a search engine and they're getting stuff that isn't search

746
00:53:27,600 --> 00:53:33,380
results, it has to be validated that that's the right thing to show, right?

747
00:53:33,380 --> 00:53:38,400
Or it has to be validated in some other way that it's appropriate for their business to

748
00:53:38,400 --> 00:53:39,400
show.

749
00:53:39,400 --> 00:53:44,680
And so ad units are a great example of that.

750
00:53:44,680 --> 00:53:50,880
Query refinement actions are all have always been a big part of Google's mission.

751
00:53:50,880 --> 00:53:52,920
It's not just search results and that's a key.

752
00:53:52,920 --> 00:53:57,920
So refinement of queries.

753
00:53:57,920 --> 00:54:03,840
Someone wanting to go from a search to a browse experience has always been or a search to

754
00:54:03,840 --> 00:54:06,100
a filtration experience.

755
00:54:06,100 --> 00:54:11,280
So for me, I'm just seeing this as another someone wants to go from a search to a blah,

756
00:54:11,280 --> 00:54:12,760
blah, blah experience, right?

757
00:54:12,760 --> 00:54:15,280
And that's why I think you hear it as generative experience.

758
00:54:15,280 --> 00:54:16,640
They don't know where that's going to land.

759
00:54:16,640 --> 00:54:23,220
But what we can be confident in is that they're going to make the right decision over time

760
00:54:23,220 --> 00:54:27,160
as to which queries should have this and shouldn't.

761
00:54:27,160 --> 00:54:32,680
When the novelty wears off and it becomes ubiquitous, is it really going to be a great

762
00:54:32,680 --> 00:54:37,820
experience for me to type in, you know, best Bluetooth speakers for a pool party.

763
00:54:37,820 --> 00:54:42,280
And it's like Bluetooth speakers are this type of thing.

764
00:54:42,280 --> 00:54:48,840
And if you go to a party, you know, you might want this feature or that feature or that

765
00:54:48,840 --> 00:54:52,580
feature and you're like, well, that's not helpful at all.

766
00:54:52,580 --> 00:54:58,080
And then they're showing me three links and I'm clicking and it's like early stage awareness

767
00:54:58,080 --> 00:54:59,160
consideration purchase.

768
00:54:59,160 --> 00:55:03,000
By the way, if you want to, if you want an observation of the week in your SGE, the right

769
00:55:03,000 --> 00:55:06,160
rail tend to not have the same intent.

770
00:55:06,160 --> 00:55:09,000
They tend to have almost a quickie journey.

771
00:55:09,000 --> 00:55:10,680
It's a really good research.

772
00:55:10,680 --> 00:55:13,840
That's worth the price of admission for this podcast today.

773
00:55:13,840 --> 00:55:20,240
It's usually pretty good as to what intense Google favors for the journey on the topic.

774
00:55:20,240 --> 00:55:24,720
And then you can scale down and there's actually two views in the SGE.

775
00:55:24,720 --> 00:55:29,000
One is a pack and the other one's a stack.

776
00:55:29,000 --> 00:55:31,440
And so the stack, you can actually see each segment.

777
00:55:31,440 --> 00:55:37,400
It can go, you know, 3000, 4000 pixels if you look this top right button.

778
00:55:37,400 --> 00:55:42,120
And so you start to learn, you're trying to learn about what I might need to be, who I

779
00:55:42,120 --> 00:55:47,640
might need to be to be part of this, to be for my stuff, to get part of this.

780
00:55:47,640 --> 00:55:53,120
And it's a little bit ambiguous right now, but we can be confident that it validates

781
00:55:53,120 --> 00:56:00,080
Google's desire for you to have content at all stages of the learning journey or the

782
00:56:00,080 --> 00:56:01,800
buying journey.

783
00:56:01,800 --> 00:56:04,400
And for that to all exhibit expertise.

784
00:56:04,400 --> 00:56:07,920
The persons that are going to get hit hardest here are e-commerce and e-commerce reviews.

785
00:56:07,920 --> 00:56:08,920
Yeah.

786
00:56:08,920 --> 00:56:11,720
I just, I mean, just, just, it would just be blunt.

787
00:56:11,720 --> 00:56:15,880
And then the second tier that's going to get hit hardest is early stage awareness.

788
00:56:15,880 --> 00:56:22,600
People that have fallen into early stage awareness content as their perception of how to rank.

789
00:56:22,600 --> 00:56:26,120
And so if you've done really, really weak early stage awareness content only on your

790
00:56:26,120 --> 00:56:36,560
site, you are a frail little deer and about to get walloped by a very large truck.

791
00:56:36,560 --> 00:56:44,360
And that I'm looking at periodical publishers, B2B tech companies who rely solely on a low

792
00:56:44,360 --> 00:56:47,720
quality glossary.

793
00:56:47,720 --> 00:56:53,440
I've seen every which way this is going to shake out and those folks should be quaking

794
00:56:53,440 --> 00:56:58,520
in their boots because they, you know, they were focused on a tactic instead of strategy.

795
00:56:58,520 --> 00:57:04,120
And that's, that's the answer that you said at the top end is if your entire, if your

796
00:57:04,120 --> 00:57:15,440
entire unpaid traffic collection channel setup is all coming from one tactic, you're in a,

797
00:57:15,440 --> 00:57:16,960
you're in a bind, a big bind.

798
00:57:16,960 --> 00:57:22,800
And a lot of these tactics that have been employed by what I call gray hat SEOs in the

799
00:57:22,800 --> 00:57:28,160
past, they're usually the ones pointing at things on videos.

800
00:57:28,160 --> 00:57:33,360
And you know, dancing about their results is a lot of those tactics, just they're completely

801
00:57:33,360 --> 00:57:38,800
irrelevant if you've got, you know, they may work here and there and you may be able to

802
00:57:38,800 --> 00:57:39,800
get some money out of it.

803
00:57:39,800 --> 00:57:42,680
But if you're a real brand, it's going to catch up with you.

804
00:57:42,680 --> 00:57:44,040
And we're already seeing that happen.

805
00:57:44,040 --> 00:57:50,840
I mean, I, I can't obviously say any brands on this, but we're, we're seeing tragic stuff

806
00:57:50,840 --> 00:57:59,200
happen with people who had had marginally hike marginally low quality or low quality

807
00:57:59,200 --> 00:58:01,000
as a tactic in the past.

808
00:58:01,000 --> 00:58:02,000
And they're like, what do we do?

809
00:58:02,000 --> 00:58:05,160
And I'm like, you have to update 700 pages like yesterday.

810
00:58:05,160 --> 00:58:08,360
And they're like, how do we do that fast?

811
00:58:08,360 --> 00:58:13,680
And I'm like, unfortunately the situation you're in is that doing that fast is going

812
00:58:13,680 --> 00:58:19,800
to cost you a lot of money and require expertise that you may not have.

813
00:58:19,800 --> 00:58:22,680
And then what do you bring?

814
00:58:22,680 --> 00:58:24,880
You bring what you bring, what makes you special.

815
00:58:24,880 --> 00:58:30,920
If you bring what makes you special to that and bring your skills, you become a little

816
00:58:30,920 --> 00:58:32,240
bit unstoppable.

817
00:58:32,240 --> 00:58:36,680
And that's, you know, those, that's what we're going to be up against is it's the big publisher

818
00:58:36,680 --> 00:58:43,320
who realizes that they can, they're so confident in the quality of their work that they can

819
00:58:43,320 --> 00:58:50,080
base all of the world's information that they present to you off of something.

820
00:58:50,080 --> 00:58:51,080
Right.

821
00:58:51,080 --> 00:58:57,800
It's that the potential for a search engine as a concept to not be valuable.

822
00:58:57,800 --> 00:59:05,720
There's a potential for an ad server as a concept to be irrelevant.

823
00:59:05,720 --> 00:59:10,480
And I've been exploring that in my brain for about five years.

824
00:59:10,480 --> 00:59:16,100
I did not know what would happen this quickly, but it is, it's going to be very, very fast

825
00:59:16,100 --> 00:59:26,360
and it's going to be ugly that some of these previous ways one access information no longer

826
00:59:26,360 --> 00:59:29,920
are relevant and then new technologies are going to take over.

827
00:59:29,920 --> 00:59:36,900
And it's the world's changing, but we're doing away with a lot of really add customer

828
00:59:36,900 --> 00:59:42,720
experience and information retrieval, negative information retrieval experiences.

829
00:59:42,720 --> 00:59:48,880
And I think you're, you're going to see a couple of years, you're going to see SEOs,

830
00:59:48,880 --> 00:59:54,680
certain SEOs are saying, I can't believe this tactic still works or this tactic will always

831
00:59:54,680 --> 00:59:55,680
work.

832
00:59:55,680 --> 00:59:58,160
And, and it's just, it's kind of like trite.

833
00:59:58,160 --> 01:00:05,340
Yeah, we've had similar conversations on the podcast many times that short-term ism.

834
01:00:05,340 --> 01:00:10,720
If you're thinking long-term, you, you've got to be looking at different approaches.

835
01:00:10,720 --> 01:00:14,800
So over the next 12 months, where what's new for market views, what's coming?

836
01:00:14,800 --> 01:00:16,520
You've mentioned a new tool there.

837
01:00:16,520 --> 01:00:18,560
What have we got to look forward to?

838
01:00:18,560 --> 01:00:26,800
So we have, we have a, our reintroduction of our proprietary language model technology,

839
01:00:26,800 --> 01:00:34,320
which is going to go into our editor experience will be the first wave of that.

840
01:00:34,320 --> 01:00:36,680
And then, so that'll happen in a month.

841
01:00:36,680 --> 01:00:43,560
So I've been recommending people get a paid account right now because there will be a

842
01:00:43,560 --> 01:00:46,840
packaging change is coming.

843
01:00:46,840 --> 01:00:53,320
And then when that happens, and then in the, hopefully knock on wood in Q4, we will have

844
01:00:53,320 --> 01:00:58,880
another release of our implementation of artificial intelligence solution that may sneak into

845
01:00:58,880 --> 01:01:06,160
Q1, but doing something that no one's ever done before with this type of technology.

846
01:01:06,160 --> 01:01:13,840
So that is, that's going to be an ex just, it's an exciting time to work in those areas.

847
01:01:13,840 --> 01:01:20,800
But most more recently, I think some things we've launched are basically a quality and

848
01:01:20,800 --> 01:01:26,960
a cluster quality analysis solution is baked into one of our applications.

849
01:01:26,960 --> 01:01:32,600
We also have a cluster or universe keyword or topic, universe gap analysis solution.

850
01:01:32,600 --> 01:01:39,120
So if you're monitoring a site, you can search for a topic and see where you have strengths

851
01:01:39,120 --> 01:01:41,600
and weaknesses on that cluster.

852
01:01:41,600 --> 01:01:46,480
And those things have been out for a couple of months and are, are worth your view.

853
01:01:46,480 --> 01:01:48,760
I always say one of them is about a 10 and a half hour workflow.

854
01:01:48,760 --> 01:01:50,980
And one of them is about 30, 30 hours.

855
01:01:50,980 --> 01:01:56,820
So if you clearly see the benefit of those things, it can typically make the ROI easily

856
01:01:56,820 --> 01:01:57,820
justifiable.

857
01:01:57,820 --> 01:02:00,280
Well, whatever comes next, I'm looking forward to it.

858
01:02:00,280 --> 01:02:05,600
The last bunch of updates that rolled out, the ones that stand out for me were the, the

859
01:02:05,600 --> 01:02:07,800
SERP X-rays, brilliant feature.

860
01:02:07,800 --> 01:02:10,040
I love diving into that.

861
01:02:10,040 --> 01:02:12,200
So much insight, a lot of value for that.

862
01:02:12,200 --> 01:02:14,480
So yeah, excited to hear what comes next.

863
01:02:14,480 --> 01:02:19,360
Did people want to get started on Market Muse and Market Muse.com?

864
01:02:19,360 --> 01:02:20,360
Absolutely.

865
01:02:20,360 --> 01:02:21,360
Yeah.

866
01:02:21,360 --> 01:02:22,360
Go to Market Muse.com.

867
01:02:22,360 --> 01:02:27,160
Shoot me an email, jeff at Market Muse.com or DM on the X.

868
01:02:27,160 --> 01:02:28,160
Okay.

869
01:02:28,160 --> 01:02:31,360
So if you're, if you're, if you're writing more than two or three pages a month right

870
01:02:31,360 --> 01:02:34,520
now, this solution is going to be right for you.

871
01:02:34,520 --> 01:02:35,520
And that's the key.

872
01:02:35,520 --> 01:02:40,480
It may require you to do some, you know, soul searching and some cleanup.

873
01:02:40,480 --> 01:02:44,840
And there's really no solution that's going to tell you how to stay differentiated while

874
01:02:44,840 --> 01:02:49,920
you clean stuff up except Market Muse, because we make sure that when we give recommendations,

875
01:02:49,920 --> 01:02:56,840
it's not just copy your competitor, it's differentiate and make sure that it's high quality and you

876
01:02:56,840 --> 01:03:00,360
know, you really can't go wrong as long as you're putting your best foot forward as a

877
01:03:00,360 --> 01:03:01,360
business.

878
01:03:01,360 --> 01:03:05,440
Thank you for listening to Artificially Intelligent Marketing.

879
01:03:05,440 --> 01:03:11,500
To stay on top of the latest trends, tips and tools in the world of marketing AI, be

880
01:03:11,500 --> 01:03:13,260
sure to subscribe.

881
01:03:13,260 --> 01:03:20,480
We look forward to seeing you again next week.