A Glimpse Into the Future of Martech - Phil Gamache

Justin Norris: 00:00:08

You're listening to RevOps FM with Justin Norris.

2

: 00:00:14

Welcome to RevOps FM.

3

: 00:00:15

Today, I'm very excited to be speaking

with one of the humans behind the

4

: 00:00:19

Humans of Martek podcast, Phil Gomesh.

5

: 00:00:22

If you're not familiar with

it, Humans of Martek is one of

6

: 00:00:25

the top shows in our industry.

7

: 00:00:27

Phil and his co host John have

been at it for a jaw dropping

8

: 00:00:30

three years and Phil, I'm getting

tired just thinking about that.

9

: 00:00:34

And I noticed you're also just one show

away from your 100th episode milestone.

10

: 00:00:39

Super exciting there, too.

11

: 00:00:41

And one of the things that's super

special about this show, aside from the

12

: 00:00:45

huge diversity of guests and topics that

they have built up over a three year

13

: 00:00:48

span, is the incredible production value.

14

: 00:00:51

So each show that they do has this really

jaw dropping AI generated custom artwork.

15

: 00:00:56

There's a website with very rich and

detailed summaries for each show.

16

: 00:01:00

And you somehow managed to work a

busy full time job as director of

17

: 00:01:03

growth at Palago while doing it all.

18

: 00:01:05

So, Phil, thank you for making me feel

complete podcast inadequacy, and welcome

19

: 00:01:10

Phil Gamache: to the show.

20

: 00:01:11

Thanks so much, man.

21

: 00:01:12

A pleasure to be here.

22

: 00:01:13

Really excited to see what

you're, you're cooking up.

23

: 00:01:15

I know you've had a lot of

Awesome guests already so far

24

: 00:01:18

and some cool conversations.

25

: 00:01:19

So yeah, always happy to share

minds on another podcast.

26

: 00:01:23

Justin Norris: Appreciate that.

27

: 00:01:24

And seriously speaking, having, uh,

I'll be soon to publish my 10th episode.

28

: 00:01:29

So 1 10th of your journey and

just knowing that the effort that

29

: 00:01:32

goes into it, the work that goes

into doing a high quality show.

30

: 00:01:35

I'm amazed that you and John what

you've achieved and what you've

31

: 00:01:37

done over that period of time.

32

: 00:01:39

So serious.

33

: 00:01:40

Kudos to you.

34

: 00:01:41

I'd just love to start out maybe about

the podcast and a little bit of the

35

: 00:01:44

origin story and the why and what

was your journey like building it?

36

: 00:01:47

Yeah.

37

: 00:01:48

Phil Gamache: So John

and I met at Cliffolio.

38

: 00:01:49

He hired me there.

39

: 00:01:50

He was heading up growth

slash marketing at Cliffolio.

40

: 00:01:54

We were a Marketo shop.

41

: 00:01:55

So I came over from Pardot and had

to pick up Marketo and learn it.

42

: 00:02:00

And John and I worked together

there for close to three years.

43

: 00:02:03

He ended up moving to Revenue Pulse

after, but we stayed in touch.

44

: 00:02:07

We became close friends

while at Clipfolio.

45

: 00:02:09

During the pandemic, we were

chatting almost weekly on Zoom.

46

: 00:02:13

And one day we were just

like, why don't we hit record?

47

: 00:02:16

Maybe there's like parts of these

conversations that other folks

48

: 00:02:19

are going to find interesting.

49

: 00:02:21

I think a lot of people were starting

a podcast during the pandemic.

50

: 00:02:24

But I've always had this itch to share the

journey, like share some tips and tricks.

51

: 00:02:29

I've personally learned a ton

from other folks early in my

52

: 00:02:34

career reading and listening.

53

: 00:02:36

So I was just like, if I can help

a handful of people that are five

54

: 00:02:40

years before me in my journey,

it would be really cool to do.

55

: 00:02:43

MarTech at the University of

Ottawa locally here and did

56

: 00:02:50

it for three and a half years.

57

: 00:02:51

And it was awesome.

58

: 00:02:52

Like I still mentor a couple of students

from those classes and initially the idea

59

: 00:02:57

for the podcast was going to be this 101

debunking marketing automation, basically

60

: 00:03:02

repurposing the course after it shut down.

61

: 00:03:05

But it quickly morphed into ramblings and

just getting some cool folks on the show.

62

: 00:03:09

So we eventually settled on the journey

or the mission of how can we future

63

: 00:03:14

proof the folks behind all that MarTech.

64

: 00:03:16

I think a lot of the content

out there is tech focused.

65

: 00:03:19

How do we do X?

66

: 00:03:21

How do we do Y?

67

: 00:03:22

But not enough is about the

humans that are powering all

68

: 00:03:26

of that tech behind the scenes.

69

: 00:03:28

So we spent a bit of time on this

show talking about, I hate the word

70

: 00:03:31

soft skills, but like the people

side of, of MarTech productivity.

71

: 00:03:35

How do you stay sane?

72

: 00:03:37

How do you balance your, your

home life and your work life?

73

: 00:03:40

So yeah, that's kind of the genesis of the

74

: 00:03:42

Justin Norris: show.

75

: 00:03:42

And it's funny that you say, and

I know jokingly about soft skills,

76

: 00:03:45

but I think a recurring theme in the

discussions that I've had is those

77

: 00:03:49

are really the difference between the

practitioner and the leader, like the

78

: 00:03:53

technical skills get you to a certain

degree, but at a certain point, you

79

: 00:03:57

need those communication skills.

80

: 00:03:58

You need the ability to prioritize.

81

: 00:04:00

You need the ability to strategize to

get to that next level in your career.

82

: 00:04:04

Similar experience.

83

: 00:04:05

On your side,

84

: 00:04:06

Phil Gamache: definitely.

85

: 00:04:06

I think that platform skills

will take you a certain way.

86

: 00:04:11

And if you want to be icy for the rest

of your career, even then there's still

87

: 00:04:14

a lot of collaboration road mapping.

88

: 00:04:18

You still need to work with

a bunch of other folks.

89

: 00:04:20

So yeah, there's there's a lot

of people skills involved in.

90

: 00:04:23

I think a lot of roles, uh, but I

think marketing ops, uh, martech

91

: 00:04:27

specifically, you're at the intersection

of a bunch of different teams.

92

: 00:04:31

Uh, so definitely agree

93

: 00:04:33

Justin Norris: in terms of

building up your audience.

94

: 00:04:35

I mean, I'm obviously at a much

earlier phase of that process than

95

: 00:04:39

you, but I saw on your website,

so, you know, 18, 000 listeners.

96

: 00:04:43

Uh, it's a very well known show.

97

: 00:04:46

Does it feel like you get to a certain

place of like self sustaining growth or

98

: 00:04:50

is it always kind of like any business?

99

: 00:04:52

We're always trying to do new

things to get to that next level of

100

: 00:04:55

Listenership and audience building.

101

: 00:04:57

Yeah,

102

: 00:04:57

Phil Gamache: there were

growing pains for sure.

103

: 00:04:59

I mean John joked I like in the early

days even like maybe the first full year

104

: 00:05:04

of doing the show like we it almost felt

Like we were screaming into the void like

105

: 00:05:08

there wasn't A lot of listeners and like I

think the advantage you have is that like

106

: 00:05:13

you've gone the route of publicly sharing

stuff like read a lot of your LinkedIn

107

: 00:05:18

posts and have been a fan before I heard

that you were doing a podcast, right?

108

: 00:05:22

Like John and I didn't

necessarily have that benefit.

109

: 00:05:25

We didn't have an audience already.

110

: 00:05:27

We didn't have following

folks didn't know who we were.

111

: 00:05:30

So we were starting from scratch

and year one was almost like

112

: 00:05:33

just having our friends on the

podcast like friends in Martech.

113

: 00:05:37

And just getting like interesting stories

and stuff and trying to figure out like

114

: 00:05:41

what what was like our unique perspective

on things and I honestly like wanted

115

: 00:05:46

to be attached to John Taylor's name

like he's still a mentor today and like

116

: 00:05:51

I think he his brain works differently

than a lot of folks in Martech like he's

117

: 00:05:55

almost a web dev now like he's picked up

JavaScript and so It's been really fun

118

: 00:06:00

to see how the growth of listenership has

evolved, and it was really rough at the

119

: 00:06:06

start, like I won't joke, and there were

definitely times in like the first year

120

: 00:06:10

where we were just like what are we doing,

like there's a ton of time that gets

121

: 00:06:14

involved in this, like it's really what

we want to be doing with our spare time,

122

: 00:06:18

but it honestly just came down to you.

123

: 00:06:20

We were having fun doing it and we told

ourselves that the day that we weren't

124

: 00:06:25

having fun anymore, we were going to stop

doing it and we never really did it to

125

: 00:06:29

like become like a top podcast and have

like hundreds of thousands of of listeners

126

: 00:06:35

like there was never really the goal like

the goal is really just like sharing stuff

127

: 00:06:39

and helping a couple of folks in in Ottawa

and like our initial little area and when

128

: 00:06:44

it really kind of picked up steam was So I

think there's been like consistent growth,

129

: 00:06:50

like year over year, never like dramatic

growth like we did this summer, actually.

130

: 00:06:55

So this summer I, uh, had a baby,

so I was on pat leave for a full

131

: 00:07:00

month, month and a half or whatever.

132

: 00:07:02

And in between diaper changes and

during naps, it's sometimes like

133

: 00:07:06

her just like sleeping on me.

134

: 00:07:07

I would just like geek out on the podcast.

135

: 00:07:09

And this was right around the time

where like ChatGPT and a lot of AI

136

: 00:07:13

applications were, were picking up steam.

137

: 00:07:15

And I went super deep on the

topic and we did like a four

138

: 00:07:19

part series on AI for marketing.

139

: 00:07:21

And that's when like folks

really started picking it up.

140

: 00:07:24

Like it was, it was such a hot topic

and everyone was talking about it,

141

: 00:07:27

but I feel like the cover art helped

us like carve a unique interest and

142

: 00:07:33

yeah, it kind of picked up from there.

143

: 00:07:35

We started getting a lot of inbound

requests from guests to be on the

144

: 00:07:39

show, be interviewed, talk about AI.

145

: 00:07:41

And, uh, yeah, this year we

had Scott Brinker on the show,

146

: 00:07:44

which definitely helped, uh,

catapult our, our listenership.

147

: 00:07:47

We went really deep on, like, niche

topics, like composable CDP, warehouse

148

: 00:07:52

native tech, uh, email deliverability.

149

: 00:07:55

And so, yeah, it's been a fun

ride, and it's a journey for sure.

150

: 00:07:59

There's growing pains at the

151

: 00:08:00

Justin Norris: start.

152

: 00:08:00

I've noticed that you guys go deep.

153

: 00:08:03

And I think you really need to have a

unique way of doing things, you know,

154

: 00:08:07

for people to start to feel that trust

and that relationship with your show

155

: 00:08:11

and what you do doesn't have to be deep.

156

: 00:08:13

It could be short.

157

: 00:08:14

It could be topical, but.

158

: 00:08:16

I think going deep is a really useful

thing because in people, I was reviewing,

159

: 00:08:19

I'd published a post on the whole

Google Yahoo changes spam complaint

160

: 00:08:24

thing and in my research came across

your episode, you know, it was very

161

: 00:08:27

much the way I also personally think

like you just broke everything down

162

: 00:08:30

very exhaustively, you took it apart,

you put it back together again, and I

163

: 00:08:34

think that's really helpful for people

because not everybody has the time or the

164

: 00:08:37

inclination to do that sort of work when

you feel that someone else has done it.

165

: 00:08:41

And you're like, well, I

can trust these people.

166

: 00:08:43

They're a trusted resource.

167

: 00:08:44

I think it's a great angle and

you're doing it effectively.

168

: 00:08:48

And I also, you know what

you said about having fun.

169

: 00:08:50

I've had a few people ask me since I

started doing this, Hey, do you think I

170

: 00:08:54

should start a podcast or what's involved?

171

: 00:08:57

I was like, do not do that.

172

: 00:08:58

But I think unless you are the type of

person that just actually enjoys the

173

: 00:09:03

process inherently, like if you think

you're going to start it for fame and

174

: 00:09:06

fortune and riches, like not that, but

if you enjoy having conversations and

175

: 00:09:12

learning, like I do think sometimes,

even if nobody else saw them and they

176

: 00:09:16

never saw the light of day, it would

be a very valuable thing for me to be

177

: 00:09:20

doing just for my own personal growth.

178

: 00:09:22

You have to be someone that feels that.

179

: 00:09:24

Phil Gamache: Totally agree.

180

: 00:09:25

Selfishly, a lot of our topics are

things that I'm pondering myself

181

: 00:09:30

at my day job, like the whole

composable CDP deep dive that we did.

182

: 00:09:35

Like we were evaluating

whether to go package CDP or

183

: 00:09:38

composable at my current startup.

184

: 00:09:41

We were debating whether we rethink

our email deliverability strategy.

185

: 00:09:45

So a lot of like the themes

that you see on the podcast are.

186

: 00:09:48

Are self serving like I, I learned from

industry experts and I apply those,

187

: 00:09:53

those learnings to, to my day job.

188

: 00:09:56

So that's like part of the, the joy there,

but I think the, also the other part

189

: 00:10:00

that a lot of folks don't realize until

they get into it is like, this is how

190

: 00:10:03

it's all that happens behind the scenes

on, on the production side of things.

191

: 00:10:07

And I've picked up most of the slack

there from John, but I actually

192

: 00:10:11

think I enjoy most of it because.

193

: 00:10:14

Early in my career before I

wanted to go into MarTech, like I

194

: 00:10:17

was, I was big on graphic design

and just like media production.

195

: 00:10:21

My dad's a photographer and video editor.

196

: 00:10:23

So like I was working on Max and

GarageBand and iMovie when I was

197

: 00:10:28

a kid, like one day I wanted to be

like a cinematographer, you know?

198

: 00:10:31

So the idea of working in editing

audio and figuring out how to like

199

: 00:10:36

create visual imagery for, for the

show, even before we went into AI

200

: 00:10:40

and mid journey, we're Drawing images

and illustrations for each guest.

201

: 00:10:44

So I enjoyed that part of it.

202

: 00:10:46

That part of it like gave me energy

as opposed to a lot of folks are just

203

: 00:10:50

like, I want to chat with people.

204

: 00:10:51

I don't want to like worry

about all that stuff.

205

: 00:10:53

And there's school media

production agencies that that'll

206

: 00:10:55

do that side of it for you.

207

: 00:10:57

I enjoy doing both and that's

part of the joy for sure.

208

: 00:11:00

Justin Norris: There's a real sense

of craft and I think you have to

209

: 00:11:03

almost see yourself as a maker.

210

: 00:11:05

I mean, you are a maker when you

do a show, but it's different than

211

: 00:11:08

just like, we're a business and

we're going to do a podcast and

212

: 00:11:10

what's the ROI and it's what you do.

213

: 00:11:12

So I want to go to the

imagery and the use of AI.

214

: 00:11:16

You've really practiced what you

preach in the sense of incorporating

215

: 00:11:19

technology into the fabric of the

show, which makes a lot of sense.

216

: 00:11:23

I'm vaguely aware of all of

these tools and capabilities was

217

: 00:11:26

certainly not expert in them.

218

: 00:11:27

So I'd love for you to just educate

me and listeners on what your process

219

: 00:11:31

is and how you, how you create all of

220

: 00:11:33

Phil Gamache: this.

221

: 00:11:34

Yeah, we already recorded a pretty deep

episode on, on the whole process and we

222

: 00:11:38

were going to put this live, but we ended

up delaying that one until next 101.

223

: 00:11:45

In favor of like all the Google changes,

too many people are talking about this.

224

: 00:11:48

We're like, ah, wait, let's jump on this.

225

: 00:11:50

Let's let's chat about this.

226

: 00:11:51

But the short version of it

is we started using GPT to

227

: 00:11:56

basically uplevel our transcripts.

228

: 00:11:59

Like, I think a lot of folks for podcasts

will just like copy paste the transcripts

229

: 00:12:03

on, on a landing page and call it done.

230

: 00:12:06

Um, and like, I think

transcription tools have evolved

231

: 00:12:09

a lot and gotten a lot better.

232

: 00:12:11

They're still filled with likes

and ums and half the time,

233

: 00:12:15

like they're hallucinating and

it doesn't really make sense.

234

: 00:12:18

I don't think ChadGBT is really good

at writing and coming up with like, uh,

235

: 00:12:22

authentic content and brand new content,

but ChadGBT is really good at taking

236

: 00:12:28

a raw audio transcript and turning it

into something legible or interesting.

237

: 00:12:34

So I've got a couple of prompts that have

evolved over time, but I'm essentially

238

: 00:12:38

taking the transcript from otter.

239

: 00:12:40

ai, we use that tool for transcripts.

240

: 00:12:42

And I'm basically taking the question

and the answer from the guest and I

241

: 00:12:47

asked ChiGBD to turn it into a blog

post passage and it's actually really

242

: 00:12:53

interesting what it comes up with.

243

: 00:12:55

Sometimes it, it gets a bit confused

when we're talking about CDP or GBT,

244

: 00:13:00

it doesn't always get the acronyms

correctly on the transcription.

245

: 00:13:02

So there's a couple of things in

there that I tweaked the prompt for.

246

: 00:13:05

I always like reread the prompt.

247

: 00:13:07

Sometimes it's not perfect.

248

: 00:13:08

Sometimes the logic is kind of

missed, but oftentimes the output is

249

: 00:13:13

way greater than the raw transcript

that someone would end up reading.

250

: 00:13:17

And so we ended up having this

like long form, like 3, word blog

251

: 00:13:22

post for each of our episodes.

252

: 00:13:24

And we're like, how do we augment this?

253

: 00:13:26

And, uh, around the time I was on pat

leave, I just started playing around

254

: 00:13:29

with mid journey, got a paid account and

was just like playing around with it.

255

: 00:13:33

And I was like, man, it would

be cool to, to make our cover

256

: 00:13:35

art, uh, for mid journey.

257

: 00:13:36

But then I was thinking like, what if

every question that we have, which is

258

: 00:13:39

essentially an H two on our landing page,

what if we have like an accompanying

259

: 00:13:44

image, uh, to, to support that, and

then we can use that for social shares.

260

: 00:13:48

Instead of just sharing the one full

episode, we can share specific pieces of

261

: 00:13:53

questions and takeaways from the show.

262

: 00:13:56

So it's, it's evolved into like

using GPT in mid journey now.

263

: 00:14:00

And we always leverage GPT to come up

with an H2 for that blog post passage

264

: 00:14:06

and summarize a practical key takeaway

for the audience at the end as well.

265

: 00:14:10

But yeah, it's definitely evolved a lot

over time and it's funny, we joke that

266

: 00:14:15

like, we hear a lot of folks on social

that discover us and a lot of the comments

267

: 00:14:19

that we get when we share our episodes are

not like, wow, the content's amazing, and

268

: 00:14:23

I learned so much, most of the comments

are just like, this cover art is badass,

269

: 00:14:27

like, wow, this image is awesome, but

270

: 00:14:30

Justin Norris: we'll take it.

271

: 00:14:31

The imagery is awesome.

272

: 00:14:33

I haven't gone deep into Midjourney,

but I've played around with a few AI

273

: 00:14:36

image generating tools, and some of

them have been pretty underwhelming.

274

: 00:14:41

Some of them are pretty cool.

275

: 00:14:42

Just the other day, actually, I was able

to successfully use one and get an image.

276

: 00:14:44

I'm like, okay, actually, I'll use

this as a feature image on a blog

277

: 00:14:47

post because it's pretty good.

278

: 00:14:48

You have obviously learned the

prompt engineering expertise to be

279

: 00:14:53

able to get exactly what you want.

280

: 00:14:55

How do you get somebody and like very

specific things that are unique to their

281

: 00:14:59

personality and incorporating it in this

futuristic landscape and all that sort

282

: 00:15:03

Phil Gamache: of stuff?

283

: 00:15:03

The secret to making the illustrations

actually look like our guests comes

284

: 00:15:08

from a secondary bot that we connect to

Discord, which we use for mid journey.

285

: 00:15:14

So it's called Insight face swap.ai.

286

: 00:15:16

And uh, I can share you, uh, links

to that, so to put in the show notes.

287

: 00:15:21

But essentially it's saving

features of the face of our guests.

288

: 00:15:26

So I have it in a private server on

Discord, and I'll upload a bunch of images

289

: 00:15:30

of the guests that I'm having on the show.

290

: 00:15:32

And then I use that bot to save one

of those images, like front facing,

291

: 00:15:37

nice, clear, no glasses, good quality.

292

: 00:15:40

And then the bot remembers

features of that face.

293

: 00:15:42

And then I'm using a bunch of different

prompts to come up with an illustration.

294

: 00:15:47

Usually, like, it's the prompt

style is flat illustration.

295

: 00:15:51

This is something I tell a lot of folks

that are getting started in Midjourney,

296

: 00:15:54

is that, like, you can learn a ton about

how to do prompts from their guides.

297

: 00:15:59

There's a bunch of YouTubers and creators

that have been using Midjourney for

298

: 00:16:04

several years now that are like light

years ahead of me and I've learned a

299

: 00:16:07

ton from them as well, but like pick a

style, there's so many different styles

300

: 00:16:11

that you can use in Midjourney and that's

like The difference between Mid Journey

301

: 00:16:16

and some of the other image generators

is that, like, once you pick a style, you

302

: 00:16:20

can actually get pretty consistent type

of colors, type of flat illustration, or

303

: 00:16:26

if you want to go, like, geometric, or

if you want to, like, do cubism, or you

304

: 00:16:31

can do, like, real life photography also.

305

: 00:16:33

So, like, pick a style, like Discover,

you can see a bunch of other people

306

: 00:16:37

publicly using Mid Journey and

get a taste of, like, what the

307

: 00:16:39

prompt was and what the output was.

308

: 00:16:42

But the secret sauce to making the

guest cover art actually look like

309

: 00:16:46

our guest is that we'll basically do

a bunch of prompts to get some flat

310

: 00:16:50

illustrations that look kind of okay.

311

: 00:16:52

Never really look like our guests,

sometimes they're like, we get

312

: 00:16:55

lucky, and it looks pretty similar

to our guests, and they'll be cool

313

: 00:16:59

with us just running with that one.

314

: 00:17:00

But, most of the time, we're just

like, right clicking on the output,

315

: 00:17:04

and then we select the inside face

swap bot, and it applies the guest's

316

: 00:17:08

real face on top of the illustration.

317

: 00:17:11

We're applying like a real photo on top

of an illustration, so the output isn't

318

: 00:17:15

always like, perfect, sometimes it just

breaks the flat illustration style.

319

: 00:17:19

So there's like a bunch of like

test and learn, uh, which is part

320

: 00:17:23

of the fun and the addictive nature

of using these image generators.

321

: 00:17:27

But yeah, ultimately we always

end up on one that looks kind

322

: 00:17:31

of okay and we run with it.

323

: 00:17:33

Justin Norris: What is your

perspective on what this process is?

324

: 00:17:37

Is it a new type of graphic design,

whereas instead of using point

325

: 00:17:41

and click inside of Photoshop?

326

: 00:17:43

You're just interacting with

a different set of tools.

327

: 00:17:46

Is it replacing the work

of the creator to a degree?

328

: 00:17:50

What is your point of view

329

: 00:17:51

Phil Gamache: on that?

330

: 00:17:52

I think for now it's not replacing the

designer, especially not the illustrator.

331

: 00:17:57

It can do concept art way faster than

those illustrators and those designers.

332

: 00:18:03

But it still can't, like, create

something that's specific to a brand.

333

: 00:18:07

Like, a lot of these designers

and illustrators have a set brand

334

: 00:18:11

guidelines that they're following.

335

: 00:18:12

And you can instruct Midjourney

to use specific colors and

336

: 00:18:17

themes, but at the end of the day,

it's never going to be perfect.

337

: 00:18:20

It's never going to be as on brand as an

actual designer and illustrator that's

338

: 00:18:25

creating, like, Retargeting ad banners

like that's always like, at least for

339

: 00:18:30

now, going to be better in the future.

340

: 00:18:32

Maybe there's a way to like upload

your brand guidelines or like a

341

: 00:18:36

specific format within the tool

itself to change those settings.

342

: 00:18:40

That could be game changing if they're

working on that, but I think for now,

343

: 00:18:44

uh, there's still a lot of upside to

using an illustrator and a designer.

344

: 00:18:49

If you've got a big company and you've

got like a media production within it.

345

: 00:18:54

But yeah, the other thing I'll say is

that Midjourney is still not perfect.

346

: 00:18:57

They've gotten better at hands and, uh,

you still can't do words in letters.

347

: 00:19:01

Uh, ChadGBT and Dolly have

started playing around with that,

348

: 00:19:05

but it's still not perfect yet.

349

: 00:19:07

Like if you say like upload

something that says humans of

350

: 00:19:09

Martech on it, it's going to be

misspelled like nine out of 10 times.

351

: 00:19:13

So like the words are still not

really great with image generators.

352

: 00:19:16

That's where like word art artists

and graphic designers are still like

353

: 00:19:20

a level up on top of, of these tools.

354

: 00:19:23

And even if you have like

something super specific to do,

355

: 00:19:25

like I'll give you an example.

356

: 00:19:27

I was trying to come up with an image,

you know, those traditional like

357

: 00:19:30

iceberg images where you see like

the tip of the iceberg, but under the

358

: 00:19:33

water is like the meat of the iceberg.

359

: 00:19:36

And so I was trying to

convey that with a concept.

360

: 00:19:38

I forget what it was, but like.

361

: 00:19:40

Mid journey wasn't able to do it.

362

: 00:19:41

Like I couldn't come up with

a prompt that outputted that

363

: 00:19:46

similar type of iceberg image.

364

: 00:19:49

Sometimes you can even like upload, uh,

an image of something that you want,

365

: 00:19:53

and then you can just like tweak the

prompt on it, reference that image.

366

: 00:19:57

But it, like I couldn't do it,

couldn't do it with mid journey.

367

: 00:20:00

Like the time it took me to

like figure out how to do it.

368

: 00:20:02

The designer would have have

been done already with that.

369

: 00:20:05

So it's still not replacing folks,

but I think at some point you got

370

: 00:20:09

to think that it will interesting.

371

: 00:20:11

Justin Norris: Those strange limitations.

372

: 00:20:13

I saw another one the other day where

it was like it had misspelled linked in.

373

: 00:20:17

It's like you've got, you know, a

quadrillion billion petabytes of data

374

: 00:20:21

and you can't spell linked like It's so

strange that it has such a capability

375

: 00:20:26

and yet limited in those specific ways.

376

: 00:20:29

Maybe we turn to Martech generally.

377

: 00:20:32

You mentioned a number of the different

technologies that you've used.

378

: 00:20:35

Apartheid Shop, then you

moved over to Marketo.

379

: 00:20:38

And you've gone through a

number of other platforms.

380

: 00:20:41

What's been your point

of view on that journey?

381

: 00:20:43

A lot of people tie their career to

one platform and kind of plant their

382

: 00:20:46

flag there and make that their brand.

383

: 00:20:48

And you've taken a different path.

384

: 00:20:49

How do you

385

: 00:20:50

Phil Gamache: feel about that?

386

: 00:20:51

Yeah, I'm definitely biased on

that point just because it is

387

: 00:20:54

the path that I stumbled into.

388

: 00:20:57

So I started off with Pardot in a

Salesforce shop, so I was really

389

: 00:21:00

deep in Pardot for like four

years and became a craft and I

390

: 00:21:04

definitely wanted to stick to it.

391

: 00:21:07

I saw the value of it and I went to like

all the conferences and I saw all these

392

: 00:21:11

like experts that were specific Pardot.

393

: 00:21:15

Specialists, right?

394

: 00:21:16

And I was like, I could do that.

395

: 00:21:18

Like I could become a part hot

specialist and I don't think there's

396

: 00:21:20

anything wrong with going that route.

397

: 00:21:22

I ended up going the route of trying

a different tool just because I

398

: 00:21:27

wanted to join a different company.

399

: 00:21:29

Clipfolio in Ottawa was like one of the

hottest startups and uh, still kicking

400

: 00:21:33

around and doing good stuff today.

401

: 00:21:35

And they just happened to be

using a different tech stack.

402

: 00:21:37

So it wasn't a matter of, I want to find

another cool company that's using Pardot.

403

: 00:21:42

It was, it was time for me to like

leave the company that I was at and try

404

: 00:21:46

something new, like look for something

greater, bigger problems to solve.

405

: 00:21:50

And they happened to be using Marketo.

406

: 00:21:52

And then when I moved over from Clipfolio

to close, they were using customer IO and.

407

: 00:21:57

So on and so forth, like now I'm using

iterable and I've used a bunch of these

408

: 00:22:00

different tools and I like to give this

idea that there is a lot of importance

409

: 00:22:05

being platform agnostic in Martech because

I am a big fan of strategy over platform.

410

: 00:22:13

Maybe like later on in your career

too, like I think initially when

411

: 00:22:16

you're an IC or a specialist.

412

: 00:22:19

And you're at a company, like, I

think there's nothing wrong with being

413

: 00:22:21

an expert in the platform itself.

414

: 00:22:23

But as you like grow up in different

levels in martech or in marketing

415

: 00:22:27

ops, you get involved in a lot

more strategy and budget planning.

416

: 00:22:30

And at the end of the day, like

all these automation tools,

417

: 00:22:33

they're relational databases.

418

: 00:22:36

They all have list building

and segmentation capabilities.

419

: 00:22:39

They're all rule based automation

workflow builders, right?

420

: 00:22:43

If this, then that, yeah, if

this, then that and or statements,

421

: 00:22:48

they're all so similar, right?

422

: 00:22:49

So I give that advice because I would

rather chase cool companies in my career,

423

: 00:22:55

then limit myself to the companies

I could work at because they use a

424

: 00:23:00

specific platform or a specific tool.

425

: 00:23:03

Like, I don't want to limit

myself to a certain stack.

426

: 00:23:06

At the end of the day, there's like

11, 000 martech tools out there, right?

427

: 00:23:09

It's impossible to be

a pro in all of them.

428

: 00:23:11

And the last point I think on that is

like, legacy platforms are slow to change.

429

: 00:23:17

Pardot hadn't changed in like X

years after I stopped using it.

430

: 00:23:21

Like Marketo is still using the same UI.

431

: 00:23:23

It's always used.

432

: 00:23:25

Let's not talk about that.

433

: 00:23:28

It's too painful.

434

: 00:23:30

It's not a dichotomy, right?

435

: 00:23:31

It's not like one or the other.

436

: 00:23:33

I think folks can make a beautiful

career being a Marketo or an

437

: 00:23:38

iterable or a HubSpot consultant.

438

: 00:23:40

I think that folks that go the

strategist way, the platform agnostic

439

: 00:23:44

ways, is also valuable as well.

440

: 00:23:47

Justin Norris: I agree with you.

441

: 00:23:48

I had a almost diametrically

opposed career in that I

442

: 00:23:50

went deep into one platform.

443

: 00:23:52

I spent seven years at a, uh, at a

consulting shop that was deeply embedded

444

: 00:23:57

in the Marketo and then the Adobe

ecosystem, but I think what you see is

445

: 00:24:01

that, yes, the technology is, is big

and it's complicated and they will need

446

: 00:24:05

specialists, but unless you really plan

to just hitch your wagon to that one

447

: 00:24:11

horse and hope That is going to keep

going for 20 years at a certain point.

448

: 00:24:16

You'll probably have to pivot.

449

: 00:24:18

You can see it already.

450

: 00:24:18

I love Marketo.

451

: 00:24:19

I think it's still a marketing

automation platform of choice.

452

: 00:24:22

I'd probably buy it again if I

was building a stack from scratch

453

: 00:24:26

today, but it doesn't have the place

that it once had in the ecosystem.

454

: 00:24:30

Like you said, it's been

acquired its legacy.

455

: 00:24:33

It's slow to change.

456

: 00:24:34

They still can't get the U.

457

: 00:24:35

I.

458

: 00:24:35

Together products.

459

: 00:24:36

They ossify and fossilize once they

get ingested by the maw of one of

460

: 00:24:42

these big companies and then they just

kind of stagnate and younger dynamic

461

: 00:24:46

companies come in and overtake them.

462

: 00:24:48

And so is that something you

really want to worry about?

463

: 00:24:52

Um, and then I also think ultimately,

if you aspire to some sort of in house

464

: 00:24:56

leadership, it's, you're not going to

go to a C level meeting and there'll

465

: 00:24:59

be like, we need your technical skills.

466

: 00:25:01

You know, they want someone who knows

those technical skills and it can be.

467

: 00:25:04

A strategic operations leader and

think about business focus problems.

468

: 00:25:08

So I think the path that you took makes

sense and I think it more and more makes

469

: 00:25:12

sense as the tech skills become not

unimportant, but more commoditized, more

470

: 00:25:16

easy to, to replicate or to offshore or,

you know, to do different things with.

471

: 00:25:20

I'm curious given all that though, let's

say you were starting a new company today.

472

: 00:25:24

What would your stack be?

473

: 00:25:25

Where have you landed as like the

best of breed for the modern era?

474

: 00:25:29

Yeah, it's

475

: 00:25:30

Phil Gamache: interesting that you say

Marquetta would still be your choice if I

476

: 00:25:33

was starting from scratch And it would be

a startup at first like I think you know

477

: 00:25:38

maybe one day when we become a big team

like we look at like the The bigger tools

478

: 00:25:43

in the space, but I think there's too

many less expensive tools and tools with

479

: 00:25:49

a smaller learning curve to get started.

480

: 00:25:51

So like, I think automation for me,

like my two favorite platforms are

481

: 00:25:55

iterable and customer IO slightly biased.

482

: 00:25:58

Cause they are the last two tools

that I've used and, and kind of like

483

: 00:26:01

dove in and really expanded on, on

the capabilities and the beautiful UI

484

: 00:26:08

and building like visual builders and

visual journeys and stuff like that.

485

: 00:26:12

I think that they lean a bit more on the

B2C side of things, specifically iterable.

486

: 00:26:18

My current company is B2B2C.

487

: 00:26:21

It's a super weird business model,

but we're on the B2C side of things.

488

: 00:26:24

We market to consumers and eligible

employees for the Pelago Health Benefit.

489

: 00:26:30

And so iterable is, is well versed

for, for that use case itself, the size

490

: 00:26:35

of database that we have in iterable,

like it would cost us an arm and a leg

491

: 00:26:39

to house that in, in a Marketo or any

of these other bigger ones, but I'm

492

: 00:26:45

a big fan of the modern data stack.

493

: 00:26:47

Like I know we, I said, we went

super deep on the composable CDP

494

: 00:26:50

architecture and I'm a big fan of it now.

495

: 00:26:53

I've completely converted, uh, at

previous companies, I was a big fan

496

: 00:26:56

of segment in, in the packaged CDP.

497

: 00:26:59

I've even used MParticle as well for a

few consulting gigs, but I've, I've opened

498

: 00:27:06

up to the composable CDP now and we can

get into some of the benefits there.

499

: 00:27:10

But like, I, I just think

that like the data warehouse.

500

: 00:27:14

Is the source of truth today.

501

: 00:27:16

And for marketers, it used to

be the CRM and then it used to

502

: 00:27:19

be the automation tool, and then

it used to be the package CDP.

503

: 00:27:23

And in all of those instances, we're

essentially making a copy of our user

504

: 00:27:28

database and we're bothering our data

team or engineering team to cause,

505

: 00:27:33

you know, like none of those CDPs off

the shelf are like, you don't need

506

: 00:27:37

a developer, even though they all

claim that you don't need a developer.

507

: 00:27:41

At the end of the day, like I think

the data warehouse, You always

508

: 00:27:44

Justin Norris: need a developer.

509

: 00:27:45

You

510

: 00:27:46

Phil Gamache: always need it, yeah.

511

: 00:27:47

Especially for data warehousing,

and I think that like most of these

512

: 00:27:51

companies, startups, especially in tech,

like the data warehouse is a central

513

: 00:27:55

piece of the stack today, and they're

making all of this work, building

514

: 00:27:59

all the pipelines to like ingest the

data from all your different tools.

515

: 00:28:03

They're structuring it in the warehouse,

and then we're using reverse CTL

516

: 00:28:07

tools that push that into other tools.

517

: 00:28:09

A CDP, a packaged CDP, is essentially

copying all the work your data team

518

: 00:28:14

did in the warehouse, and you're paying

for a whole other copy of that data,

519

: 00:28:19

just like you are in Marketo already,

or in Iterable, like the, it's not

520

: 00:28:23

warehouse native yet, I think at some

point a lot of these tools will become.

521

: 00:28:27

But so yeah, I think that like, you

know, whether it's snowflake or big

522

: 00:28:31

query or redshift, it's become an

essential part of the stack for Martek.

523

: 00:28:35

Not that Martek folks are really

Engineering the data warehouse, but they

524

: 00:28:40

are architecting a lot of the tools that

they pick based around the idea that the

525

: 00:28:44

warehouse is the central point of truth.

526

: 00:28:47

And then we would use like snowplow

rudder stack to do that event collection,

527

: 00:28:53

replace GA for like, don't have to worry

about all of those limitations there.

528

: 00:28:58

And then you use a reverse detail

tool like census to push the data

529

: 00:29:02

that you have in your warehouse to all

the tools that you use in marketing.

530

: 00:29:07

So your ad platforms, your iterables,

your CRM, so that you don't have

531

: 00:29:11

to worry about, like, is my CDP,

the data in my CDP, is it matching

532

: 00:29:16

with what I have in the warehouse?

533

: 00:29:17

Is it match what I have in the CRM?

534

: 00:29:19

All of these tools have the

warehouse as the central point.

535

: 00:29:23

Of failure were positive there, but

yeah, so those are the core of what I

536

: 00:29:28

think would be part of it from scratch.

537

: 00:29:30

There's a bunch of other ones too.

538

: 00:29:32

Like I'm still a big fan of Zapier.

539

: 00:29:34

I love the world of async and having like

internal wikis for like documentation,

540

: 00:29:39

big fan of notion and using loom.

541

: 00:29:43

All these like white boarding tools

like, uh, we use Figma and, and Fig Jam

542

: 00:29:47

at, at work for just workflow processes,

asynchronous design thinking sessions,

543

: 00:29:53

experimentation, C-M-S-C-R-M, I don't

have too many hot takes on those.

544

: 00:29:57

There's a lot of options in that space.

545

: 00:29:59

Obviously I spend a bit

of time at WordPress.

546

: 00:30:01

I have a, a bit of a bias for that

as, as the CMS, but yeah, there's

547

: 00:30:05

still a ton of options, right?

548

: 00:30:06

Justin Norris: We certainly

have no shortage of options.

549

: 00:30:08

Nobody's ever complained that we

don't have enough MarTech yet.

550

: 00:30:11

As the Scott Brinker landscape

map becomes, like, approaching

551

: 00:30:15

infinity in its size.

552

: 00:30:17

Let's, I want to turn to this concept

of composability and maybe You've

553

: 00:30:21

obviously thought deeply about it.

554

: 00:30:23

For those for whom it might be

a newer concept, can we just

555

: 00:30:27

build it from the ground up?

556

: 00:30:29

Clearly it sounds like what we're

talking about is taking different

557

: 00:30:32

components and putting them together

to create something that maybe you

558

: 00:30:35

would have to buy off the shelf as a

single unit of functionality otherwise.

559

: 00:30:39

Yeah,

560

: 00:30:39

Phil Gamache: I think the easiest way

to think about it is point solutions

561

: 00:30:43

versus an all in one platform.

562

: 00:30:46

Composable for me is like this idea of

best of breed components that some of

563

: 00:30:51

them are purchased, some of them are

created in house, whether you're data

564

: 00:30:54

engineers or building your reverse ETL

pipeline in house, or the ETL pipeline,

565

: 00:30:59

like the best of breed idea, whether

it's purchased or created in house.

566

: 00:31:03

It just means all of these components are

integrated together, and maybe they're

567

: 00:31:08

different pieces, but you're not locked

into one vendor on like a two year, three

568

: 00:31:13

year contract, and you're being billed

by a number of people in the database.

569

: 00:31:18

There's this idea that flexibility is

currency in today's smart tech landscape.

570

: 00:31:24

I think Arun, the founder of Castle.

571

: 00:31:26

io said this on our podcast,

it like it stuck with me.

572

: 00:31:29

Flexibility is currency.

573

: 00:31:31

And so opting for a composable stack

or like a composable CDP or even like

574

: 00:31:36

unbundling pieces of your Martech

over this idea of like an all in one

575

: 00:31:41

tool, it can provide you this like

elbow room to innovate and adapt.

576

: 00:31:46

Faster than if you were

locked into one vendor.

577

: 00:31:49

One of the examples there would be

the advancements of AI right now.

578

: 00:31:53

Like if you're locked into one platform

that doesn't have AI features yet.

579

: 00:31:59

What are you doing?

580

: 00:32:00

Like you want to play

with those AI features?

581

: 00:32:02

Like, are you waiting two

or three years for that?

582

: 00:32:05

Like big legacy platform to

finally come out with AI features?

583

: 00:32:10

Or are you trying some of these new

startups, these new point solutions that.

584

: 00:32:14

Have this capability already that can

actually hook up to your data warehouse.

585

: 00:32:18

That's the idea of the composable stack.

586

: 00:32:20

It's a way more flexible.

587

: 00:32:21

There's options for a bit more.

588

: 00:32:23

Innovation is in some cases it is cheaper.

589

: 00:32:25

In some cases you have to factor in

like the troubleshooting when you

590

: 00:32:28

have like seven tools instead of one.

591

: 00:32:31

And something breaks like the

troubleshooting piece, like it

592

: 00:32:34

comes into becoming a bit trickier.

593

: 00:32:36

Like you have to figure out

which one is breaking there.

594

: 00:32:39

Sometimes it isn't cheaper, but

sometimes folks will pay for it

595

: 00:32:42

just for the added flexibility.

596

: 00:32:45

But this idea of not being locked

in to one platform and you can't

597

: 00:32:49

try any other features until that

one platform comes up with it.

598

: 00:32:53

And maybe they were purchased by a

large enterprise and they're still

599

: 00:32:56

integrating it with the bigger company.

600

: 00:32:58

AI features by the time they

come out, there's going to

601

: 00:33:01

be like way more advancements

already and like point solutions.

602

: 00:33:04

So that's, that's the idea behind it.

603

: 00:33:06

Like best of breed components that

all work together, but it does

604

: 00:33:10

Justin Norris: feel like best of

breed being taken down to another

605

: 00:33:14

higher level of resolution because

time was where best of breed meant.

606

: 00:33:17

I'm going to use Marketo

and Salesforce instead of.

607

: 00:33:19

Pardot and Salesforce, you know,

like kind of like VHS Betamax Wars of

608

: 00:33:24

Best of Breed versus like buying the

Super Stack from one company was, you

609

: 00:33:28

know, I'm not not going to just buy

everything from Salesforce or everything

610

: 00:33:30

from SAP or everything from Adobe or

whatever I'm going to pick and choose.

611

: 00:33:33

And now it's you're even

taking some of those things and

612

: 00:33:35

breaking them down even further.

613

: 00:33:37

It feels like where Yeah, you have a

warehouse instead of just having Marketo.

614

: 00:33:41

Maybe you have one tool

for your workflow building.

615

: 00:33:44

Another tool is, uh, ingesting

activity data like digital clickstream

616

: 00:33:47

data, like you referred to snowplow.

617

: 00:33:49

And I find that inherently appealing

because like for all the reasons that

618

: 00:33:53

you mentioned, the one area where I think

A single vendor sometimes makes sense

619

: 00:33:57

is HubSpot, if you're smaller, helping

a friend with a business project and he

620

: 00:34:01

was on HubSpot and I was just amazed,

they've got kind of everything at 80%.

621

: 00:34:06

And if you're a small company just

getting started, that maybe that's

622

: 00:34:09

kind of like good enough for you.

623

: 00:34:11

I don't need Chili Piper, I have

Calendar Building and I don't

624

: 00:34:13

need Sales Loft because I have my

sequencing and I don't need Marketo.

625

: 00:34:17

It's just all right there.

626

: 00:34:19

But I think once you're past a certain

size, you want Like you said, I don't

627

: 00:34:22

want to wait for features, I want

to have the best of everything, and

628

: 00:34:25

I want them to play well together.

629

: 00:34:27

The flip side to that, then, though,

if I think about it is what you alluded

630

: 00:34:32

to with the troubleshooting point.

631

: 00:34:34

And I've even found we have a

great data team, but it's hard

632

: 00:34:38

working with a data warehouse.

633

: 00:34:40

There's discrepancies.

634

: 00:34:41

There's why is this opportunity in sales?

635

: 00:34:43

As soon as you have more than one of a

thing, then all of a sudden you're Like

636

: 00:34:46

you said, bringing worlds together.

637

: 00:34:48

And is your experience that this

is achievable in a smaller org or

638

: 00:34:52

does it require big teams or is

there inherent overhead and like

639

: 00:34:56

friction and cognitive load that

comes with breaking things up in this

640

: 00:34:59

Phil Gamache: way?

641

: 00:34:59

Definitely I think like on a

troubleshooting perspective on

642

: 00:35:03

like a QA perspective, making sure

that data is matching up in all

643

: 00:35:07

these tools because you're still

loading that data differently.

644

: 00:35:10

You're converting that data in

different format based on like

645

: 00:35:13

the two of the endpoint solutions

that you're sending it to.

646

: 00:35:16

Um, so yeah, maybe, maybe it's

not one of the first things

647

: 00:35:19

you do as a, as a startup.

648

: 00:35:22

And I think even like a couple of years

ago, this idea of like having a data team

649

: 00:35:27

at inception of a startup was, was crazy.

650

: 00:35:31

But I think today it's,

it's almost a necessity.

651

: 00:35:34

And, um, I've worked for startups

where there wasn't a data team, right?

652

: 00:35:37

Like there was maybe an analyst or

two, but like when it came to getting

653

: 00:35:41

data in your automation tool, It was

up to the marketer to figure that out.

654

: 00:35:45

Or if there was too technical stuff,

you had to bother one of the engineers

655

: 00:35:49

who was working on product stuff to

like stop working on customer facing

656

: 00:35:53

product things, help the marketing team

on JavaScript and like cookies and stuff

657

: 00:35:58

like that on on the front end side.

658

: 00:36:00

Today, I think a lot of startups

from day one will have at

659

: 00:36:04

least a person owning data.

660

: 00:36:07

And they're working on source of truth

and maybe some customer facing stuff,

661

: 00:36:13

but at the end of the day, their,

their role is like stitching together

662

: 00:36:16

different teams and making sure that

teams are getting the data they need.

663

: 00:36:20

They're coming up with analysis

like eventually they want to.

664

: 00:36:23

You know, go to funding and,

and, and get extra stuff.

665

: 00:36:25

So like all the fundraisers right now are

asking like, what are you doing with AI?

666

: 00:36:29

And part of the key there

is having a ton of data.

667

: 00:36:32

You can't do anything useful

with AI if you don't have like

668

: 00:36:36

a raw database already started.

669

: 00:36:38

That's, you know, structured

or someone thought about that

670

: 00:36:41

from day one from inception.

671

: 00:36:42

So that's like the need or the,

the exception or the expectation

672

: 00:36:46

of having a data team from day one.

673

: 00:36:48

And so does it make sense to have like

a warehouse, you know, in the first

674

: 00:36:53

couple of days and you're using HubSpot,

like maybe not, but I think at some

675

: 00:36:57

point, if, especially if you have a

data person on the team, like it's a

676

: 00:37:01

matter of figuring out how does the

marketer work with the data person?

677

: 00:37:05

What are the limitations?

678

: 00:37:07

What's going to be the hiring roadmap?

679

: 00:37:09

Are we going to invest

more in that data team?

680

: 00:37:11

And we're going to have

capabilities to figure.

681

: 00:37:13

All those QA issues out and, um, the cost

differences versus like building a package

682

: 00:37:20

CDP, like an MParticle or a segment, like

those discussions and all need to happen

683

: 00:37:25

with like a, the data team and our tech

team, because at the end of the day, like

684

: 00:37:29

when you're just getting started, I don't

think there's anything wrong with like

685

: 00:37:32

doing the HubSpot route, like on top of

all the things you said they do, like they

686

: 00:37:36

also do a CMS, you can have your blog and

HubSpot, they do all the forms for you.

687

: 00:37:41

But at some point you'll

need to make that decision.

688

: 00:37:43

Like, are we still going

to stick to HubSpot?

689

: 00:37:45

Do we want like more best of breed tools?

690

: 00:37:48

And then the whole migration

discussion needs to happen.

691

: 00:37:51

And maybe you kick that down the

road because you don't want to

692

: 00:37:53

take on that migration project.

693

: 00:37:55

And if you've got everything invested in

HubSpot, like you've got like three years

694

: 00:38:01

of SEO data in the HubSpot CMS tool, like

how excited are you about moving that

695

: 00:38:07

over to Ghost or, or, uh, or WordPress?

696

: 00:38:10

Like, That's a big migration

project in and of itself, and I'm

697

: 00:38:13

speaking from experience there.

698

: 00:38:16

Justin Norris: Yeah, those growing

panes are extremely expensive.

699

: 00:38:18

And it's easy to forget, I don't

know, seven, eight years ago, it was

700

: 00:38:22

kind of revolutionary to have a cloud

first warehouse and that wasn't,

701

: 00:38:27

you know, necessarily managed by IT.

702

: 00:38:30

I remember him being consulting,

thinking about like, I want to come

703

: 00:38:33

up with like a totally cloud based

BI solution, cloud based warehouse,

704

: 00:38:36

cloud based ETL, cloud based BI.

705

: 00:38:39

I mean, today that would be

an entirely mundane concept.

706

: 00:38:41

It's like, yeah, like, of course,

but at that time it was a big lift.

707

: 00:38:45

It was different.

708

: 00:38:46

And I think I mean, that's why Snowflake

is a quadrillion dollar company that it

709

: 00:38:50

is because everyone is using it now, but

it's easy to forget, I guess, that that

710

: 00:38:54

was not commonplace, you know, not so long

ago for some of us, at least, even for a

711

: 00:38:59

Phil Gamache: lot of Martech vendors,

like it's not a reality for them yet,

712

: 00:39:02

like a lot of the automation vendors

are not cloud based, like they don't get

713

: 00:39:08

data from your warehouse, some of them

are setting up sinks and like direct

714

: 00:39:12

pipelines into it, but at the end of the

day, they're still copying that data.

715

: 00:39:16

iterable and Marketo and customer

IO, they all charge you based on the

716

: 00:39:20

number of users that you're storing

in the Marketo database, right?

717

: 00:39:24

But you're also paying snowflake for

storing those users in that database.

718

: 00:39:28

So I think in the future, I don't

know how far down the line this is.

719

: 00:39:32

And there's some startups that are

kicking around with this idea already.

720

: 00:39:35

Some bigger platforms on

B2C already doing this.

721

: 00:39:38

But this idea of like warehouse native

martech, like connecting directly to

722

: 00:39:43

the database, like sitting on top of

the database, querying it when you're

723

: 00:39:47

creating a segment versus copying it.

724

: 00:39:49

I go super deep on this on the podcast.

725

: 00:39:52

Like there's a lot of ideas and

opinions about this, like copying data.

726

: 00:39:55

Sometimes there's still

upsides to doing that.

727

: 00:39:58

Like you get that data closer to you.

728

: 00:40:00

So when you want to

act on it, it's faster.

729

: 00:40:03

But then there's like latency issues and

it's not necessarily real time anymore.

730

: 00:40:07

So there's, there's a

lot of debates around it.

731

: 00:40:09

But I think eventually the space is

moving to this idea of warehouse native.

732

: 00:40:15

We'll get a

733

: 00:40:15

Justin Norris: selection from your

back catalog of relevant episodes for

734

: 00:40:18

people that want to go deeper into this.

735

: 00:40:21

Before we turn away from the Martech

talk, this is sort of perhaps an

736

: 00:40:24

existential question in some ways,

but as technologists, as operators,

737

: 00:40:29

we all have our preferences.

738

: 00:40:30

Zooming out a level and taking

like a business lens on it,

739

: 00:40:34

how much does it matter?

740

: 00:40:36

We all need this tech to get jobs done.

741

: 00:40:38

Is it a 5 percent competitive advantage?

742

: 00:40:40

Is it a 10%?

743

: 00:40:41

I know that's impossible to quantify,

but what is the ultimate value to the

744

: 00:40:45

business of all of this technology?

745

: 00:40:47

Phil Gamache: Yeah.

746

: 00:40:48

I think if you're putting on a short

term lens and you're trying to hit

747

: 00:40:52

short term goals, it doesn't really

matter whether it's like X or Y, they're

748

: 00:40:58

probably still going to figure out a

way to do it and, and get to like your

749

: 00:41:01

endpoint for your short term goals.

750

: 00:41:04

I think it's important for companies

and that are thinking more long

751

: 00:41:08

term, like when, when you're building

your Martech roadmap and I know

752

: 00:41:12

you had Daryl on the show and we

just finished recording with him.

753

: 00:41:15

Like, he's really big on

this idea of thinking of your

754

: 00:41:18

Martech stack as a product.

755

: 00:41:20

And there's a lot of

things that come into that.

756

: 00:41:22

But like, one of the cool ideas is

like, Productizing your Martek roadmap

757

: 00:41:28

and looking into the future, like

not just next year, but like two,

758

: 00:41:31

three, four years down the line.

759

: 00:41:34

What does that tech stack look like?

760

: 00:41:36

And what do we need to be able

to do by the time we hit like

761

: 00:41:39

our goals in four or five years?

762

: 00:41:42

The tools that we have today, maybe

they're hitting our short term

763

: 00:41:44

metrics, but we're going to have a

data team that's going to be 20 people.

764

: 00:41:49

Are we making the most use out of

those folks if we're using a tool

765

: 00:41:53

that's copying our customer data and

it's not leveraging our warehouse?

766

: 00:41:57

So like with the long term lens, I

think this becomes super important.

767

: 00:42:01

There's obviously a lot of unknowns,

like there's things you can't predict.

768

: 00:42:05

And like, what does the MarTech

world look like in 5 10 years?

769

: 00:42:09

Like, who the hell knows with AI?

770

: 00:42:11

But I think those discussions

are important to have when, when

771

: 00:42:13

you like think of your MarTech

roadmap as, as a product and

772

: 00:42:17

you're trying to productize it.

773

: 00:42:19

On the

774

: 00:42:19

Justin Norris: AI subject, obviously

that's been the buzzword of the year.

775

: 00:42:23

And so I'm not questioning as

AI, it's a fad, it'll disappear.

776

: 00:42:26

I really don't believe that.

777

: 00:42:28

And I don't think that it's.

778

: 00:42:29

Very defensible to believe that

right now, but I do wonder sometimes.

779

: 00:42:33

What is the reality?

780

: 00:42:35

What is the hype?

781

: 00:42:36

What is overblown?

782

: 00:42:38

Was it coming for our jobs?

783

: 00:42:40

And you clearly a very thoughtful

person and have worked more closely

784

: 00:42:44

with it than I think anyone That I've

had the chance to speak to in depth.

785

: 00:42:49

So I'm curious What's your

take next year this time?

786

: 00:42:53

What will the role of AI be in our

daily work from your point of view?

787

: 00:42:57

Phil Gamache: These discussions have

been happening all year with a lot of

788

: 00:43:01

MarTech vendors on the product side

and trying to figure out Do we just add

789

: 00:43:07

a little copy generator on our tool,

a copy assistant, and call it done?

790

: 00:43:12

Like we have an AI feature?

791

: 00:43:14

Or do we really think mindfully

about the future of the space?

792

: 00:43:18

How do we move away from rule

based automation tools that

793

: 00:43:24

are and, or, ifs, ands, whys?

794

: 00:43:27

to letting AI take the wheel when

it comes to deciding what message

795

: 00:43:33

should a certain segment get.

796

: 00:43:35

And I think some tools in the

space that are less legacy have

797

: 00:43:40

already started moving, uh,

towards some of those features.

798

: 00:43:44

But I think that there's some really

interesting capabilities in the

799

: 00:43:47

enterprise space and I've spent

most of my time in Martech and my

800

: 00:43:51

career in startups, but I did have

a stint at automatic and wordpress.

801

: 00:43:55

com and they're like

a: 8000

802

: 00:43:59

So not like a massive enterprise, but

some really, really smart data scientists

803

: 00:44:05

that I had the pleasure of working with.

804

: 00:44:08

And I think that they're like light

years ahead in terms of Building

805

: 00:44:12

stuff in house that eventually is

going to be part of Martek vendors

806

: 00:44:18

is just going to be table stakes.

807

: 00:44:19

One of those things is propensity models,

or this idea of like, uplift modeling.

808

: 00:44:25

Right now when we're doing like

a discount campaign, we just

809

: 00:44:28

finished Black Friday, right?

810

: 00:44:29

Like, folks are sending out an email

to their entire database with a 15

811

: 00:44:34

percent discount notification, right?

812

: 00:44:37

When you send that discount to all of your

customers and a customer who was going to

813

: 00:44:42

buy at full price anyways gets your offer

and buys at discount, you're technically

814

: 00:44:47

losing revenue from that person.

815

: 00:44:49

So there's a lot of literature around

this like uplift modeling idea.

816

: 00:44:54

I think like the early Facebook

data engineers were some of

817

: 00:44:58

the proponents around this.

818

: 00:44:59

Usually like puts folks into four buckets.

819

: 00:45:02

So there's like the sure things.

820

: 00:45:03

Those are the people that are going

to buy, regardless of whether they

821

: 00:45:07

get your black Friday offer or not.

822

: 00:45:09

There's the lost causes.

823

: 00:45:10

They have moved on from you.

824

: 00:45:12

They maybe started a free trial,

but they're not interested

825

: 00:45:14

in buying your product.

826

: 00:45:16

There's the sleeping dogs.

827

: 00:45:17

Those are the folks that might react

negatively to actually getting your offer.

828

: 00:45:21

Maybe they were going to renew, but they

got your email and it pissed them off.

829

: 00:45:24

And now they're not going to

renew because they remembered how

830

: 00:45:27

much they were spending with you.

831

: 00:45:29

But the Persuadables is kind of

the holy grail of, of marketing

832

: 00:45:32

and, and this idea Uplift Modeling.

833

: 00:45:35

This is the most important group

because these are the folks that are

834

: 00:45:38

on the fence about buying something

and are going to be the most receptive

835

: 00:45:42

to getting your Black Friday offer.

836

: 00:45:44

So how do you let AI take over

when it comes to figuring out?

837

: 00:45:49

Your target users, what

bucket do they fall into?

838

: 00:45:52

And how do you only send that discount

to the persuadables and focus on them?

839

: 00:45:57

So yeah, at automatic, I worked with a

data science team that had built out an

840

: 00:46:02

internal modeling engine called pipe.

841

: 00:46:05

And it was so powerful for marketers

and the growth team because we could

842

: 00:46:09

essentially build a model for any of

the events that we were tracking from

843

: 00:46:14

our users from a predictive standpoint.

844

: 00:46:17

So we could build a list and target

specific users for an email campaign

845

: 00:46:21

based on a question that we had.

846

: 00:46:23

Like.

847

: 00:46:24

Which free users are going to buy a site

by day number 30, or which paying users

848

: 00:46:30

are going to churn by day 90, like we

could target specific messaging to those

849

: 00:46:36

users before they actually do that thing.

850

: 00:46:38

Abandoned cart emails happen

after the fact and sometimes

851

: 00:46:42

people have already moved on.

852

: 00:46:44

But when you are able to predict

someone who's going to have something

853

: 00:46:47

in their cart or isn't going to buy

something at the end of a certain

854

: 00:46:51

day, You can get ahead of that and

try to prevent it from happening.

855

: 00:46:55

So there's a whole, like,

bunch of use cases around that.

856

: 00:46:59

And I think that, like, you know,

a lot of folks are still living

857

: 00:47:01

in this, like, rule based nature.

858

: 00:47:03

And the future is really letting these

models be part of the MarTech vendor

859

: 00:47:09

stack so that humans and MarTech folks

are less prescriptive about the campaigns

860

: 00:47:14

and the touch points that are going out.

861

: 00:47:16

And you're letting models predict

whether someone is going to do something.

862

: 00:47:20

And so you're sending them the best

message at the right time to do that.

863

: 00:47:24

Justin Norris: That's funny.

864

: 00:47:25

What you described reminds me a lot of

something I studied when I was first

865

: 00:47:29

getting into marketing talking about 10

or 15 years ago now, I read a book called

866

: 00:47:35

drilling down by a fellow named Jim Novo.

867

: 00:47:37

He did a lot of database marketing, like

shopping network or shopping catalog

868

: 00:47:42

type of thing where you're, you know,

people that are buying stuff like that.

869

: 00:47:44

But they were looking at like recency,

frequency, monetary value, RFM, maybe

870

: 00:47:50

you or listeners are familiar with.

871

: 00:47:52

But it was a very similar idea.

872

: 00:47:53

You divide people into

quintiles, I think it was.

873

: 00:47:56

And then you adjust your

strategy accordingly.

874

: 00:47:58

You have a discount ladder so that you're

not offering big discounts to people that

875

: 00:48:03

we're going to buy anyways, and you're

offering greater discounts to people.

876

: 00:48:06

That are maybe much less likely to come

so you entice them back with a stronger.

877

: 00:48:11

So it makes me feel like many things

that maybe it's applying a much more

878

: 00:48:16

sophisticated set of rules or taking,

you know, a much bigger set of data

879

: 00:48:19

points aside from just those three.

880

: 00:48:22

To do, to take playbooks that have

existed and apply them, you know,

881

: 00:48:25

perhaps in a more intelligent

way or more automated way.

882

: 00:48:28

Yeah,

883

: 00:48:29

Phil Gamache: letting the machine take

the wheel and you're still kind of, uh,

884

: 00:48:33

running point on stuff and monitoring

things and doing QA, but I think like

885

: 00:48:38

gone are going to be the days and five

ish years where the marketer is creating

886

: 00:48:43

rule based automation based on loose

data as opposed to just like models

887

: 00:48:49

with like a bunch of historical data

and events that are way better than

888

: 00:48:54

us at predicting potential behaviors.

889

: 00:48:56

Will

890

: 00:48:57

Justin Norris: LinkedIn by that time?

891

: 00:49:00

Phil Gamache: Maybe not and like the

hands are still going to be weird.

892

: 00:49:03

Justin Norris: Touching on the

personal side, now you mentioned

893

: 00:49:05

you're a new dad, you have a full

time job, you have this side gig, this

894

: 00:49:10

project, really almost a second job.

895

: 00:49:12

How do you balance

896

: 00:49:13

Phil Gamache: all that?

897

: 00:49:13

Yeah, this is a key theme for us in every

episode that we chat with folks, so.

898

: 00:49:19

I think, for me, we pride ourselves on

this idea that, like, balance is like

899

: 00:49:25

this fixed point in the journey that,

like, it's something that we, we need to

900

: 00:49:29

reach, uh, but I think it's just this,

like, this continuous journey rather

901

: 00:49:33

than a, a, a final destination and it

involves a ton of stuff and every human

902

: 00:49:39

is different and different things.

903

: 00:49:41

Energize people and different things like

reduce the battery on, on a lot of folks.

904

: 00:49:47

So for some folks like passion

and alignment and making sure

905

: 00:49:51

that you're doing meaningful work,

that's like a big part of my job.

906

: 00:49:55

I work at Pelago that

helps conquer addiction.

907

: 00:49:58

And, you know, a lot of companies

like to say that they're saving

908

: 00:50:01

the world and they're saving lives.

909

: 00:50:03

We're not doctors, we're not saving

babies, but we are saving people from

910

: 00:50:07

their addictions, and we have saved

lives, and I think that's a, a super

911

: 00:50:11

powerful, uh, thing to be able to say,

like I, I help run the MarTech stack

912

: 00:50:15

and, and run the growth team that runs

Essentially, like when we convince

913

: 00:50:19

someone to sign up, like they quit

their tobacco addiction, they quit their

914

: 00:50:23

opioid addiction or they reduce their

drinking and it has like meaningful

915

: 00:50:26

impact on, on other people's lives.

916

: 00:50:29

The other thing for me that drives me is

this idea of like personal recharging,

917

: 00:50:33

being able to take time away from work.

918

: 00:50:36

Like I love MarTech obviously, like

I'm able to have a podcast about

919

: 00:50:39

it that I'm doing on my free time.

920

: 00:50:42

But I'm still a big fan of escapism,

healthy escapism, going for walks

921

: 00:50:46

with with my dog, my wife, my newborn,

taking part into TV shows, a big fan

922

: 00:50:51

of like science fiction and reading

books and just like escaping the

923

: 00:50:55

world that is like day to day grind.

924

: 00:50:58

But for me, age old advice that actually

comes out in a lot of the answers on

925

: 00:51:02

the show when we ask this question

is To never underestimate the power

926

: 00:51:06

of a well timed no when you get a new

request or you get a new opportunity.

927

: 00:51:11

That's the key to

maintaining that balance.

928

: 00:51:14

Like we had Lauren on the show

and she put it really nicely.

929

: 00:51:17

I don't know if she came

up with this or not.

930

: 00:51:18

She was like, life is about like

juggling and you have a bunch of balls

931

: 00:51:22

in your hands that you're juggling.

932

: 00:51:23

Some of them are made of glass

and those are the important

933

: 00:51:27

ones that you can't drop.

934

: 00:51:28

So it's about figuring out what are

the most important things that you have

935

: 00:51:32

going on, making sure you're prioritizing

those, and not feeling stressed about

936

: 00:51:36

saying no when other stuff comes up

that won't have the bandwidth for.

937

: 00:51:41

Justin Norris: It's good advice.

938

: 00:51:42

It's good to know you're

a, uh, a sci fi fan.

939

: 00:51:44

We didn't go into Star Trek and Martek,

but I think there's a whole other show

940

: 00:51:47

there we could do, but I really appreciate

you spending the time with me today, Phil.

941

: 00:51:51

You're a super interesting

guy and inspiring what you've

942

: 00:51:55

achieved with the podcast.

943

: 00:51:56

And I also just think how it's informed

your perspective or the interplay between,

944

: 00:51:59

you know, the learning and discussions

you've had through the podcast and your

945

: 00:52:03

thoughtfulness as a Martech leader, as

a growth leader and what you're doing.

946

: 00:52:07

We'll include lots of links in the show

notes again to some choice cuts from the

947

: 00:52:10

back catalog for folks that want to go

deeper Into human smart tech, but really

948

: 00:52:15

appreciate you chatting with me, and

I hope we get a chance to speak again.

949

: 00:52:19

Yeah, it's a

950

: 00:52:19

Phil Gamache: super fun thanks for

having me and really excited to see the

951

: 00:52:22

trajectory of this show and happy to

support you and Excited to keep listening.

952

: 00:52:27

I've listened to all all episodes so far.

953

: 00:52:29

So Keep it up, man.

954

: 00:52:31

Hey

955

: 00:52:32

Justin Norris: everyone.

956

: 00:52:32

I want to invite you over to the

Rev Ops FM substack community

957

: 00:52:36

where you can sign up to get rough

transcripts, show notes, longer form

958

: 00:52:39

articles, and other bonus content.

959

: 00:52:42

Just head over to rev ops fm slash

subscribe to get free access.

960

: 00:52:47

I'd also love to know what you thought

of the episode and to hear suggestions

961

: 00:52:49

for topics you want to learn about.

962

: 00:52:51

Feel free to leave a comment

on substack or send me an

963

: 00:52:54

email at Justin at rev ops fm.

964

: 00:52:58

Thanks for listening.