HomeVideos

10,000 Ads/Month: The Tournament System Replacing Media Buyers | Cody Schneider Interview

Now Playing

10,000 Ads/Month: The Tournament System Replacing Media Buyers | Cody Schneider Interview

Transcript

1051 segments

0:00

Cody Schneider, co-founder and CEO of

0:02

Grafft. He is uh continually doing so

0:07

many things with AI agents,

0:09

go-to-market, etc. He is famous

0:12

>> He's cracked.

0:13

>> that distribution is even more important

0:16

than the product itself. He is cracked.

0:18

We must bring him inside of the show.

0:20

Everybody welcome the one and only Cody

0:21

Schneider of Grafft. Stand up, Cody.

0:24

Get inside this show.

0:26

>> He's got the Steve Jobs swag.

0:28

>> Let's go. How you doing?

0:29

>> Let's go. We're doing even better now

0:31

that you are here, Cody. And I have to

0:34

say you're looking very you're looking

0:35

very Jobs-esque, very Steve Jobs.

0:38

>> rock look, you know? I just only do when

0:40

I do public. It's like everybody always

0:42

in the comments just blows it up. It's

0:44

like Anyway, it [snorts] it's really

0:46

just engagement bait, but it always

0:48

works. It's hilarious.

0:49

>> You do a very very rock. That's true,

0:51

very Dwayne Johnson.

0:52

>> meme of him and it just like always

0:54

rips, which is hilarious. So, anyways.

0:56

>> Cody

0:57

>> What were you guys talking about before?

0:58

Is it people buying pallets from Walmart

1:00

and then flipping them on eBay?

1:02

>> E-commerce.

1:03

>> I have a founder friend who's doing

1:04

this. They just like raised money for

1:05

this in Seattle, but they

1:06

>> Seriously?

1:07

>> Their whole setup, it's really

1:08

interesting. They do like

1:10

they do it regionally, so they buy from

1:12

like a store and then they go I don't I

1:13

can't remember the company name right

1:14

now, otherwise I would share it, but uh

1:16

they basically like you have to drive up

1:18

and pick it up physically. And it just

1:20

turns into the like your car you you

1:22

know you drive your car in they

1:23

literally like you know whatever it is,

1:25

you know, a pressure washer they load it

1:27

in the back of the car and then you like

1:28

drive away. Absolutely crushing it. I

1:31

didn't realize like this was such a huge

1:32

problem for them like both for the

1:33

companies like just trying to offload

1:35

this inventory, but anyways, super

1:36

random, you know, tangential thing

1:38

that's related.

1:38

>> There's

1:39

>> Wow.

1:39

>> There's opportunity everywhere

1:42

>> Absolutely, man.

1:43

>> on the interwebs.

1:43

>> There's so many different ways to make

1:44

money.

1:45

>> So many different ways to make money. Uh

1:47

now, Cody, I have heard I have heard in

1:50

all of your um agent building, agent

1:54

mobilizing, man, I have heard that

1:56

there's uh something interesting on your

1:59

side going on with agents monitoring

2:01

every competitor ad library and such,

2:04

Facebook and and more. Is this Is this

2:06

the case, Cody?

2:07

>> Yeah, 100%. I mean, you're seeing this

2:08

more and more like we I mean, you guys

2:11

are in the industry, you know how this

2:12

functions, right? It's like if you look

2:13

at like Rise Superfoods as an example

2:15

right now, I think they have like 2,500

2:16

plus like ads that are live currently

2:18

across the platforms and they're just

2:19

like at scale deploying. It's in the

2:21

range, you know, 100 plus daily.

2:23

Um

2:24

>> Phew.

2:24

>> Uh this is becoming like really

2:26

important. Is So, when you think about

2:28

inventory that's available on an ad

2:29

platform, right? Like something we spend

2:31

a lot of time on of like, okay, what is

2:32

my total addressable market? I work in

2:34

the B2B space, so it's a little bit

2:35

different, right? But we're like, what

2:36

is my total addressable market? How do I

2:38

get in front of them, you know, across

2:39

all of the places that I'm trying to

2:40

market? Um and then like I need to think

2:43

about basically, how is everybody else

2:46

in the market positioning themselves?

2:47

And then where is it like red ocean? So,

2:49

like every company that's a beef tallow

2:51

company is saying these exact, you know,

2:53

three same things with their ads. Okay,

2:56

well, there's actually this these seven

2:57

other benefits that we can talk about

2:59

that nobody is talking about once we

3:01

actually pull all of the creative. So,

3:04

if we go and we just make creative for

3:06

those topics that people aren't like

3:08

competing in, what we'll see is that

3:10

ROAS will like increase, it'll become

3:12

cheaper on a CPA, etc. And it's largely

3:14

just like

3:16

mapping, right? Again, how is the market

3:18

communicating themselves? Like So, first

3:20

it just take a step back. How do you do

3:21

this? It's like, okay, what are the pain

3:23

points and the outcomes of the person

3:24

that I'm trying to sell to? Like what

3:26

are the pain points they have? What are

3:27

the outcomes that they want? And then

3:29

once I've identified that, we like how

3:31

we do this is we build a matrix, right?

3:32

Of like, here are all these emotional

3:34

triggers, here's these situations that

3:36

they could be in. So, like as an

3:37

example, like one that we did recently

3:39

was like um I took this restaurant over

3:42

from my dad and I'm afraid I'm going to

3:43

run it into the ground. That's an

3:45

emotional trigger, right?

3:46

>> Okay. [laughter]

3:47

>> Situation that they could be in is like,

3:48

oh, it's the Friday night rush. It is

3:51

the uh you know, what like it it's uh

3:55

the weekend and it's off. I know their

3:56

phone is ringing. I hear it ringing. I

3:58

don't want to answer it cuz I'm like

3:59

shutting down, you know, etc. Um and I

4:01

can make ads across all those different

4:03

variations of the matrix. So, imagine I

4:04

have like, you know, five different

4:06

emotional triggers and I have like 10

4:07

different situations. Like from that I

4:09

can create, you know, 15 different

4:11

concepts of ideas. When you map once you

4:13

created that

4:15

when you go and you pull the ad creative

4:17

that all of your competitors are

4:18

running, what you'll find is that if you

4:20

like, you know, put that to the matrix,

4:22

you'll find that certain parts of the

4:23

matrix are lit up. Like oh like only

4:26

these like, you know, only these cells

4:28

within the top left corner. Everybody's

4:30

talking about the product in the exact

4:32

same way, but there's really like, you

4:33

know, 10 different ways that I could

4:35

talk about this and then nobody is. And

4:37

so if I go make ad creative talking to

4:39

those pain points, [clears throat]

4:40

I'm actually going to be able to

4:41

outperform, you know, who I'm competing

4:43

against because I'm differentiating the

4:45

messaging to the same, you know, exact

4:47

customer, but like I you know, talking

4:48

about it in a different way. So,

4:49

anyways, I I think that that's something

4:51

that we're really excited about and

4:52

we're seeing like really great results

4:54

with. So, just like to pass that on.

4:56

Also just the scale that you have to do

4:57

this at now. I mean, we're at this like

4:59

the point where it's like every day, you

5:02

know, we are publishing for example for

5:04

one company it's like 20 or

5:06

it's five new ad sets, five new ads

5:08

within those ad sets every day. The ads

5:11

compete against each other over like a

5:12

two to three-day window. Losers

5:14

automatically turn get turned off. So,

5:16

the winners get left within that ad set

5:18

and then the winners at winning ad sets

5:19

are to a winners pool and then the

5:21

winners pool is competing against each

5:23

other for like the ad budget that's

5:24

available, right? So, it's almost a

5:26

different game than like historically

5:28

like I would like find an ad and I would

5:29

try to go and scale that. Like I would,

5:30

you know, how much like what is the the

5:32

ceiling that I'm going to? Now with

5:34

Indromena, again, I'm in the B2B space

5:36

so it's a little bit different. The game

5:37

that we're playing is like how much ad

5:39

creative can I just give the model

5:42

because it's almost like I I you know, I

5:44

had somebody else describe this where

5:45

it's like it's this one-to-one ad

5:47

algorithm, right? Where it's like it's

5:49

just going to go find this random person

5:50

that it can serve this ad to. And maybe

5:53

like, you know, that one person it's

5:54

going to cost $30 to get that lead from.

5:56

It's not going to find anybody else,

5:57

right?

5:58

>> Yeah.

5:58

>> But we can go like it's so cheap to make

6:00

that creative now that we can go and

6:02

just like feed basically Andromeda for

6:04

this. So, anyways, happy to answer any

6:05

questions about that or like how we're

6:07

actually doing that or like you know

6:08

technically, etc.

6:09

>> I I find that fascinating. I find that

6:12

fascinating. I'd love to know like how I

6:14

when you start talking spreadsheets, I

6:16

mean, Cody, my eyes just like I'm I'm my

6:19

face gets tired. When my face gets tired

6:23

>> I understand, yeah.

6:24

>> all of this is just like it's just code

6:26

under the hood, right? Like but for

6:27

example, if you're pulling your ad

6:28

libraries, like you can use Appify,

6:30

you're just like monitoring

6:32

the ad libraries of your competitors and

6:35

then have it where it's like a daily

6:36

cron job that's like Okay, go into that

6:40

you know, Facebook ads library, pulling

6:42

out the net new ad creative, it adds it

6:44

into the database of like what is the

6:46

active ads, and then from that you're

6:48

basically okay, like what are the core

6:50

concepts? Like you can have a vision

6:51

model watch the actual ad if it's a

6:53

video and be like, what is this? Like

6:55

explain [laughter] to me like what's

6:57

happening like you know, second by

6:59

second within this video. Okay, cool.

7:01

Like based off of this like now let's do

7:03

that for 10 different competitors and

7:05

then from that aggregate, okay, this is

7:07

how this is the things that these are

7:09

the larger trends that we're seeing,

7:10

right? Oh, for some reason claymation

7:13

ads are now going viral within

7:14

supplements that are for, you know,

7:15

women's health, okay. Let's let's can we

7:18

go and create make ad creative using

7:20

seed dancing that as well, right? Etc.

7:23

Um,

7:23

but yeah, and then I mean

7:25

this isn't like a person

7:28

doing this, right? [laughter]

7:29

This is just this is just software. Like

7:30

what the [ __ ] is an agent, you know, to

7:32

be

7:32

>> Yeah. I Yeah, what Yeah, what is it? I

7:35

mean, it's it's software, right?

7:37

>> software under the hood. It's a software

7:38

with a thinking loop and some type of

7:40

data like feedback, right? And like

7:42

that's all it is. It's like cool, like

7:44

it's an app when you think about a media

7:45

buyer, like what are they doing, right?

7:47

They have like an algorithm that's in

7:48

their head of like, okay, I'm publishing

7:50

new creative, I'm looking at what's

7:51

working. I'm like doubling or I'm

7:54

removing the losers and I'm doubling

7:55

down on the winners. That can be like

7:58

turned into code now. Like but what do

7:59

you need to accomplish that? You have to

8:00

have a data stream and you basically

8:02

have to have like an agent that can like

8:03

think on top of that data and then the

8:05

ability to make new ad creative. You If

8:07

you've solved all of that, you can just

8:09

have this engine that's like running

8:11

media buying now. And this is like where

8:14

this is all headed, right? And so

8:15

anyway, I I I think with all that said

8:17

though, let's take a step back. Like the

8:18

real opportunity I think is that you can

8:20

now identify like what are all the ways

8:23

that I could talk about this? What are

8:25

all the ways my competitors are talking

8:26

about this? What are the ways that they

8:27

aren't that they should that I should

8:29

be? Let me go play in that blue ocean

8:31

space and like I this is just like you

8:33

with hypothetically cloud code, you

8:35

know, figuring this out. Totally

8:36

possible. So.

8:38

>> That's juicy.

8:39

>> I think this is so important because I

8:41

think a lot of people are the smaller

8:43

players within our space not realizing

8:45

that they're going against behemoths. We

8:47

talked about IM 8, you know, raising not

8:50

raising but getting a billion dollars of

8:51

cap that they can basically deploy

8:54

against ads. And so if you're running

8:57

the same exact thesis for ads as IM 8

9:00

and you're trying to compete with them,

9:02

and you're you're like that doesn't even

9:04

work at all because

9:06

they're going to win. They're going to

9:08

win nine to 10 times out of 10. You

9:11

don't have the capital to keep up with

9:12

them. But if you zig when they zag and

9:15

you're talking about completely other

9:16

things, that's a way to compete with

9:19

them, not not capital to capital. It is

9:22

just angle to angle. I love this I love

9:24

this concept so much.

9:25

>> information is out there too. I think

9:27

this the challenge is like historically

9:28

to go and collect all this would be I

9:30

mean you'd literally pay.

9:32

Oh my god.

9:33

Yeah, it'd be insane, right? Like you

9:34

got you You both know what this actually

9:36

takes to do. In in contrast now it's

9:38

like I can just have an agent go for

9:39

like 12 hours straight just read Reddit,

9:42

aggregate all of [laughter] the

9:43

information for all the different

9:45

situations that people are in.

9:47

Um you know, both [laughter] good and

9:48

bad like right?

9:49

>> Both good and bad.

9:50

>> But then from that I can then basically

9:52

again map out this persona and like this

9:55

is how I think about Andromeda more and

9:56

more is like I am trying to map out this

9:59

emotional trigger plus a situation that

10:01

they're in and that situation is like

10:03

I'm using language that they wrote that

10:05

resonates with them, right?

10:07

And from that I can like you know,

10:09

really clearly see like there's only so

10:10

many ways that I can talk about beef

10:13

tallow, right? But like maybe you know,

10:15

people

10:16

are spending less time talking about it

10:17

as like an eczema solution or whatever,

10:19

right? Okay, cool. Then you should spend

10:20

your you should spend your time there in

10:22

contrast to like oh, here's this like

10:24

you know, it's it's replacing what

10:26

people do with like Vaseline for

10:27

slugging, right? Or whatever.

10:29

>> [laughter]

10:29

>> Yeah.

10:30

>> So anyway, yeah. But

10:32

again I the scale the volume of the the

10:34

creative that you can read This is where

10:35

this is all going is like and I mean

10:37

we're spending so much time trying to

10:38

figure this out of like how can I employ

10:40

seed dance to like at scale make

10:42

creative that feels cinematic that feels

10:45

like you know, these Pixar ads rip

10:47

across the board

10:48

by

10:49

any industry any category etc. And I I

10:52

don't think we're there yet like but

10:53

we're close and we're probably like six

10:55

weeks away to where it's like you're

10:57

just going to like plug into

10:59

this ad machine and it just starts

11:01

creating ad creative, right? And like

11:03

you know, you can have that

11:04

automatically upload to Facebook. Now

11:06

the problem becomes understanding like

11:08

if I'm publishing 10,000 ads a month,

11:10

what the [ __ ] is working? Like that's

11:11

going to be the hardest part of this,

11:13

right? So anyway.

11:14

>> Well Cody, it seems to me. I'd love your

11:16

take on this. I mean it seems to me like

11:18

Zuck wants to build this inherently

11:22

inside of Meta. It's like we can make

11:24

your ads, we can run your ads, you just

11:27

tell us a CPA and who you want and like

11:29

we'll go make it happen. I I I um

11:32

I genuinely like don't really trust them

11:36

to do that.

11:37

>> [laughter]

11:38

>> I JUST DON'T.

11:39

I JUST DON'T.

11:40

>> HIRED US, RIGHT? Cuz like how many times

11:42

have you gotten on a call with like a

11:43

Google Ads specialist and they're just

11:45

like, "You should increase budget." And

11:46

it's like, "That's not what I'm here to

11:47

do, right?"

11:48

>> [laughter]

11:50

>> Thanks for the advice. Really appreciate

11:52

the advice.

11:52

>> Totally. And like also like when you

11:54

think about like what they're probably

11:55

training off is it's to the average,

11:56

right? And like what we're advocating

11:58

for here, what we're actually talking

11:59

about is like I don't want the average.

12:00

I don't want like what all my

12:01

competitors are doing. I'm trying to

12:02

look for like what are the outliers

12:04

because that's with the better place to

12:06

allocate the limited resources I have,

12:08

time, money, mental energy, right? Um

12:10

and so I think, you know, for the app

12:14

I was talking to I was at a dinner in

12:16

San Francisco recently and this guy was

12:17

saying like, you know, when they

12:19

initially built like the self-service

12:21

like Facebook Ads platform, a lot of it

12:23

was just like we're just trying to get

12:24

to like very normie like the boost a

12:27

post, you know, that that boost a post.

12:29

The very normie SMB owner. Like there

12:31

was just so much money that they could

12:32

like basically accumulate from that. And

12:34

they have no idea what the what is the

12:36

actual outcome, what does this actually

12:37

do, but [ __ ] like 10,000 people in my

12:38

geography saw it, that's awesome.

12:40

>> Yep.

12:40

>> Um but I think that the

12:43

like for the average person, probably

12:45

solves it. You're an HVAC company, so

12:47

you're like, "I hate my agency. Like I

12:49

despise them, you know? Like all they

12:52

do, you know,

12:53

they just lie to me, full stop, right?

12:54

Like they're not actually like giving me

12:56

the real information, they're not

12:57

delivering." You're telling me that I

12:58

can just plug into Facebook and I'm

13:00

like, "Cool. Here's this form that I

13:02

want people to complete. Go find me

13:04

people that like, you know, do my get me

13:06

leads for my tree trimming business,

13:08

right?" And it like makes it a creative

13:09

that solves that. I think that that for,

13:11

you know, again, the SMB makes a lot of

13:13

sense. But now it's like you want to

13:15

compete in the top 1% or, you know, top

13:17

10%. The only way to create And this is

13:19

with all the AI like generate

13:21

AI is just a tool that is like utilized

13:24

by the person that or that, you know,

13:25

the the the strategy that's behind it,

13:27

right? Like if I say, go write me a blog

13:29

post about X, right? Like it's going to

13:31

write to the average of the bell curve,

13:33

it's going to be terrible, it's going to

13:34

be super vague. In contrast, if I'm

13:35

like, hey, here's 10 podcasts from

13:38

experts talking about this specific

13:39

topic, right? Um

13:41

for example, like hotel marketing in

13:43

2026, go write me a piece of content

13:47

based off of the source material that I

13:49

provided you, the outcome that you're

13:50

going to get from that is going to be,

13:51

you know, top 1% because I I created

13:54

this walled garden that it has to think

13:55

in, and I gave it source material for

13:57

It's the same idea here, right? Like of

13:58

like, what is a good ad for like me to

14:01

be able to like be publishing or me to

14:02

publish to it the the channel, I have to

14:04

basically like context-oriented,

14:07

like this is this is the scope. Like

14:08

this I'm pointing you in the direction

14:09

of what actually Now, I can give it

14:11

leeway within that, right? where it's

14:12

like, okay, yeah, we're going to go

14:13

generate 100 different like, you know,

14:15

static variations based off of these

14:18

different persona like maps that we're

14:19

going after and these different, you

14:20

know, these pain points, these outcomes

14:22

that they want. But it's still very

14:24

specific to like, this is what I know

14:27

Like I'm still giving it the the

14:28

direction. This is what north looks

14:30

like. You can go northwest, you can go

14:32

northeast, I don't care, but you have to

14:33

go north in some capacity. So.

14:36

>> Cody, oh, go ahead, Kyle. Go ahead.

14:38

>> I was just going to say, Cody, I would

14:39

love to shift gears into what you've

14:41

been talking about with TikTok real

14:42

farms because I am

14:44

>> [laughter]

14:44

>> so insanely fascinated by this. I and I

14:48

will say I will say selfishly, I have

14:50

tried to crack this. I was unable to

14:52

crack it for me personally. I I think

14:55

there's something there. I would love

14:57

for your take on this because I think

14:58

it's

14:58

>> Yeah, yeah, yeah. I'm happy and I'm not

15:00

an expert on this. There's way better

15:01

people than like I have friends that

15:03

this is all they do and like they're

15:04

incredible, right? But um so, what a

15:07

TikTok real farm is, it's basically like

15:10

imagine literally an Android phone that

15:12

is in a warehouse in Brooklyn that has

15:14

like an eSIM that's separated and

15:16

they're getting a local like IP address

15:18

and they like have like they're like

15:20

doom scrolling on this phone to make it

15:22

look like it's a real person, right?

15:24

So company like our partner that we use

15:26

we love them double speed AI they're

15:28

great. Um

15:29

>> You use them?

15:31

>> Yeah, we use double speed. They're

15:32

awesome. We've been super satisfied with

15:34

the results and the like what we've seen

15:36

and their team has been just great

15:38

partner. Um but the what it ends up

15:41

looking like

15:42

is in practice they basically like they

15:44

manage this like infrastructure for you.

15:46

Um I think it's like $100 per account

15:49

per month or don't quote me on that.

15:52

And we've got 10 accounts and it's I

15:53

think it's like $1,500 is what we pay

15:55

per. Basically what the math comes out

15:57

to is like if you're doing

15:59

about a thousand or 500 to a thousand

16:02

dollar or sorry 500 to a thousand views

16:04

per video across like 10 accounts that

16:06

you have over a month the CPM on that

16:09

ends up being about two to three

16:10

dollars.

16:11

>> That's nice.

16:12

>> on Tik Tok is like you know in the range

16:14

of $10 right? When you just like are

16:16

just doing like ads to the general

16:18

audience. So What what the what it ends

16:20

up looking like that's that's the whole

16:22

arbitrage here is you're basically

16:23

getting cheaper costs on impressions on

16:26

one of these platforms. You can do this

16:27

on like Instagram as well there's like

16:30

less sophisticated providers Tik Tok is

16:31

where like majority of people spend the

16:33

time. What the media so how this

16:35

actually functions like double speed has

16:36

these accounts they give you an API and

16:38

then we use the nano banana to make

16:40

slideshows where it's like literally

16:41

like

16:42

>> Slideshows baby.

16:43

>> Exactly. So it's like super cheap. We're

16:44

not doing videos for this. We tried

16:46

videos it doesn't work as effectively

16:48

just like slideshows always work.

16:50

The tricky part is finding a format that

16:52

is already working that you're then

16:55

going and replicating. And so this is

16:56

where you have to go and do this

16:58

research. You can just brain rot

16:59

yourself like go to Tik Tok and like

17:01

scroll within your category you're going

17:02

to find like what's working and then

17:03

pull out those winners and then turn

17:05

that into a repeatable format. Or you

17:07

can use a tool like one of them's

17:09

Viralo it's like v i r l o

17:12

they have an API that enables you to go

17:15

it like with Claude code and you can

17:16

basically be like hey I'm in the beauty

17:18

category show me all the like viral

17:20

posts in the last 7 days and it's going

17:21

to like bring up it it has this huge

17:23

database like a repository. Gives you

17:25

that source material.

17:26

So once you make these

17:28

slideshow images the actual like

17:30

tactical pieces is like you have some

17:32

type of hook like imagine day in the

17:33

life of a corporate girly and then it

17:35

like it's like sliding them through

17:37

them.

17:37

>> Yeah.

17:37

>> What new like their day in New York

17:39

looks like you include the product

17:41

within the slideshow. Like as a natural

17:44

like part of it and then you can also

17:46

include that you know in the for example

17:50

the the caption of of the the thing the

17:53

the the actual like account you just

17:55

like make it look like a normal person.

17:58

>> [laughter]

17:58

>> Not that it's like branded whatsoever

18:01

it's just

18:01

>> Yeah.

18:02

>> And every I mean if you I guarantee you

18:03

guys like

18:04

I would not be shocked I have all the

18:06

internet is fake right like I'm I'm in

18:08

problem with this.

18:09

>> Your dead internet theory.

18:10

>> I am I'm 100% it's already done like

18:12

it's it's game over.

18:13

>> It's over.

18:14

>> It I don't know if there's a way to just

18:16

there isn't a way to beat it

18:18

I mean at the end of the day it just

18:19

comes into like your

18:21

we're just hunting for impressions right

18:23

like I'm just like looking for like what

18:24

is the way to get in front of my target

18:26

customer as cheaply as I can to create

18:27

the outcome that I'm looking for. But

18:29

anyways yeah when you do that at scale

18:30

though like the master is crazy right so

18:32

it's like okay 10 accounts you know if I

18:34

can get whatever 300 to 500,000 views a

18:36

month from those like if I scale that up

18:38

to 100 accounts if I scale that up to

18:39

1000 what does that look like and when

18:42

you see um

18:44

like in if you see a

18:46

like a musical artist like come out of

18:47

nowhere

18:49

and why are they all over social this is

18:51

how this is happening.

18:52

>> Yeah.

18:52

>> If you see like a brand come out of

18:54

nowhere and it's like all over social

18:56

this is how it's happening.

18:58

If you see you know kind of universally

19:00

this is like the tactic that's that's

19:01

this is how discovery happens for new

19:03

products so it's a great way to do like

19:05

brands uh

19:07

like build like brand notoriety or

19:09

authority and just like also like brand

19:12

just like do they does it do they do

19:14

they have brand recognition etc. of like

19:16

you know what you're trying to build?

19:17

It's not very tactical like it's not

19:19

very bottom of funnel. Like again, I'm

19:21

B2B we don't like we don't see any crazy

19:23

results from it, but it's like we use it

19:24

as like a again, we're trying to build

19:26

brand within this category for a

19:28

specific customer that we're selling to

19:29

etc. Um there's just like another way to

19:31

show up in front of them. For e-comm you

19:32

can get this to work though where it's

19:33

like cool, here's this literally the

19:34

supplements that I take on slide three

19:36

hits next when I when I take my when I

19:38

drink my matcha I also take this, right?

19:41

And like that is like a part of like

19:43

that sales process. Um then again, the

19:45

it's just very hard to track attributed

19:47

like

19:48

this is like why it's we we don't do it

19:50

a lot of it or like why we see people

19:52

not wanting to do it. Hard to track on

19:54

an attribution side. Um you have to

19:56

think about it in the aggregate. People

19:58

are always hunting for like outliers.

19:59

Like that's not the game that you're

20:01

playing here. You're playing more of

20:02

like I'm getting you know if the average

20:05

views again can be a thousand per video

20:08

like a thousand per slide show, awesome.

20:09

Like we've we've we've accomplished

20:11

again like what that you you know

20:12

hitting the CPM number that we're trying

20:14

to to to come to from a price

20:15

standpoint. So

20:16

>> I I

20:17

you know in in D2C and e-comm like

20:20

people are spending gobs of money per

20:23

month.

20:23

>> gobs

20:24

>> gobs of money

20:25

>> gobs

20:26

>> What is what is like fifteen hundred

20:28

bucks for a two to three dollar CPM?

20:31

That's like crazy.

20:32

>> And Chris we we had Ryan Beltran CEO of

20:35

Original Grain on the show and he said

20:37

specifically yesterday he said

20:38

impressions matter. Impressions matter.

20:41

I cannot quantify it to my brand. It's

20:43

not something I can like tie back into a

20:45

conversion and blah blah blah. He's

20:47

running a a multi multi million dollar

20:49

watch brand and he is saying impressions

20:52

matter. At the end of the day

20:53

impressions matter for my brand.

20:55

>> I think it's just I mean you can see it

20:56

too, right? Like you can see it in the

20:58

branded search. That's the only like you

20:59

know trailing indicator where it's like

21:01

you can go to Google search console and

21:03

you can be like our branded search is up

21:05

month over month. Like what does this

21:06

look like? Or we had a viral video. This

21:08

happens all the time. It's like we have

21:10

a banger day, and like do we see

21:11

impressions go up? Like branded search

21:12

impressions increase, and branded clicks

21:14

increase? Absolutely. Always like you

21:16

can you can absolutely quantify it.

21:18

Tying that all together though, like

21:19

what is my ROAS? What is my number?

21:21

There's no way like you're going to

21:22

actually do that. People are trying to

21:23

solve this. I I've been in this game for

21:25

like

21:26

10 years.

21:27

>> [laughter]

21:28

>> Everybody has told me they're going to

21:29

solve this over my entire career. It's

21:31

never been solved. [laughter] It's not

21:32

going to be solved anyways.

21:34

>> For For that like banger video that

21:36

you're talking about, that quote-unquote

21:37

like viral run, is it the same form Like

21:40

are you talking about like a a slideshow

21:43

video that pops off? And then if so,

21:45

like what would you consider that to be

21:48

a Okay, we got a viral one on our hands.

21:50

Is it like a hundred? Like a hundred K

21:52

impressions?

21:52

>> It's an outlier that's like 5x or 10x

21:54

what the average is. Um

21:55

>> Got it.

21:56

>> Like So, for example, like this is a

21:58

strategy, right? It's these like TikTok

21:59

real farms. Another strategy is I put

22:01

creators on payroll. Like I use

22:02

something like a lot of people are

22:04

starting to use slideshow. I haven't

22:05

used it personally. I don't know how it

22:06

functions, but how I've ever done this

22:08

is like I go and I

22:10

scrape 20,000 creators in beauty, and I

22:12

cold email them, and I'm like, "Hey, uh

22:15

I want you to do a video per day across

22:17

Facebook or sorry, across uh Instagram

22:20

Reels, YouTube Shorts, and TikTok. Uh

22:22

we'll pay you whatever, you know, 900 a

22:24

month for that."

22:25

>> Yeah.

22:26

>> And then I go and I get 20 different

22:28

creators. They're all making a video per

22:30

day, so I have 20 videos per day going

22:32

out. And then I'm They're basically

22:34

hunting for viral formats.

22:36

As soon as I see a viral format, then I

22:38

roll that out to the whole all the

22:39

creators that are within my like cohort,

22:41

right? So, it's like I'm But that that

22:44

same idea, like this is how this can

22:45

work together is like, "Cool, I found a

22:46

viral format. I now take that and I

22:48

apply that to the TikTok real farms."

22:52

And maybe they don't perform as well as

22:53

the actual like original. Like but even

22:56

if it's only whatever, you know, if it's

22:57

10% of a hundred Say your average, you

23:00

you views per video or per slideshow is

23:01

a a thousand. And that video did a

23:04

million, and you only get, you know,

23:05

10,000 on the TikTok real farm in

23:07

comparison to the million that it did.

23:09

Um, it's still, like, that's still an

23:12

outlier. There's still a way to double

23:13

down on that. So, it can work in tandem

23:15

together, but it typically is like one

23:16

direction, like where it's like, I have

23:18

the creator identified, that human

23:19

identifies the format, and then I'm

23:21

rolling that out to like this this real

23:23

farm. Um, this is in contrast to, like,

23:26

that's creator marketing, right? This is

23:27

totally in contrast to influencer

23:28

marketing. Which is its own animal,

23:31

which is like, this person has like, you

23:32

know, incredible reach, and also like,

23:34

they have trust and notoriety that's

23:36

built into them.

23:37

>> Yeah.

23:37

>> And that's its own game, right? And a

23:39

lot of the time that's for brand, like,

23:41

you're building brands, right? Like, for

23:43

example, why did, you know, Meta Ray-Ban

23:45

like just work with Kylie Jenner, right?

23:47

It's like to elevate the brand identity.

23:49

>> Yeah. Yeah.

23:50

>> Um, and that there's that's that can,

23:53

you know, strategically fit into it, but

23:54

it just depends on like what you're

23:55

trying to do. Like, are you, is this

23:56

transactional, or is this like more

23:58

again, like, am I going for impressions?

24:00

Am I going for that brand affinity? Cuz

24:02

we know that branded search happens, we

24:03

just don't know when. Like, it's the

24:05

only, like, impressions are the way that

24:06

you influence word of mouth, right? And

24:08

and then the the example I always come

24:10

back to, like, I had a I had a boss who

24:11

said this to me, he's like, why do you

24:12

think that like accident lawyers buy

24:15

every billboard in a city, you know, a

24:16

tier two, tier three city? Like, do you

24:18

think it's not working? Like, do you

24:20

think it that they're buying billboards

24:21

cuz it's not [laughter] working? They're

24:23

always buying billboards, so like,

24:24

something is obviously happening, and

24:26

this is like why Frank Wright is like

24:28

getting all the leads [laughter] for

24:30

accident lawyers in like, you know,

24:31

whatever Tuscaloosa. So.

24:33

>> Yeah. Yeah.

24:34

>> Something's working.

24:35

>> Something's working.

24:36

>> That's why they're doing it.

24:37

>> That's so funny.

24:37

>> Cody, I you've always got so much sauce.

24:41

>> Sauce. You're dropping sauce all the

24:43

time.

24:44

>> And but here's here's the thing. Here's

24:46

the thing. Like, I I have never really

24:50

like fully got my hands around grafft.

24:54

But bro, YOU CAN DEPLOY AI AGENTS WITH

24:56

GRAFFT. LIKE, THAT IS THIS IS LIKE THE

24:58

THING. I MEAN LIKE

25:00

IS THIS am I the only one who's shocked?

25:02

I mean I've maybe I should have known

25:04

this but like you can do you can do the

25:06

stuff that you're saying with Graft.

25:08

>> I so we initially started out as like

25:10

data pipeline and warehouse tool and

25:11

then like that's what I was told it was.

25:13

Yeah. Totally and we knew agents were

25:14

coming on some horizon and like we were

25:16

going to build marketing agents and

25:17

sales agents and like fit into that

25:19

space. We only work with B2B right now

25:21

just like that's largely just

25:23

where we have expertise and like where

25:24

we're seeing the like the adoption of

25:26

this make the most sense. Long term we

25:28

totally see this working for e-com which

25:29

is again like you're trying you're uh

25:33

you know you're a brand that's trying to

25:34

compete against like Rise right? Yeah.

25:37

How are you going to compete against

25:38

them? Like okay well this is you're

25:40

going to have to have agents that are

25:41

being functioning as like media buyers

25:43

and also like production like media

25:45

production you know for ads and voice

25:46

right? But anyway yeah so what we build

25:48

is basically data pipeline data

25:49

warehouse and then also the

25:50

infrastructure to deploy the agents onto

25:53

and then also the infrastructure to

25:54

deploy artifacts too. Say you wanted to

25:56

make a reporting dashboard. So an

25:58

example of this is like so how we

26:00

function right now is we basically for

26:01

to plan engineer and we do this work for

26:03

company. They're like I just most

26:05

companies are like I just want to buy

26:06

the outcomes. I don't care how it

26:07

happens. That's where we fit into this.

26:09

Long term where this all evolves to is

26:11

basically like you're in cloud code and

26:13

you build some type of system that's

26:14

like running your Facebook ads for you

26:16

automatically and you're like damn I

26:18

wish I could deploy this into a cloud

26:20

that like has access to my live you know

26:23

Facebook ads data Google Analytics data

26:25

my like Shopify data that's all in a

26:27

single place.

26:28

That's basically the infrastructure that

26:29

we're building that enables you to just

26:31

like one stop shop you know deploy that

26:33

out. So anyway yeah that's the product

26:35

and why we're helping companies right

26:36

now. But

26:37

>> That's incredible.

26:38

>> We could talk about agents all day

26:39

though. I know we're coming up on time

26:40

but it's it's crazy man. Like if you we

26:41

had this conversation 6 months ago I'd

26:43

have been like agents are [ __ ] and

26:45

[laughter]

26:46

now I'm just full stop I don't know

26:48

where the the boundary the only boundary

26:50

we found so far like on our side is long

26:52

form video editing. Um

26:54

>> Okay.

26:55

>> Everything else like if a human can do

26:55

it on a computer,

26:57

um we can almost get, you know, we can

27:00

basically get an agent to do it at this

27:01

point. Um

27:03

>> I

27:03

>> This is an example like we have we're

27:04

probably launching launching in the

27:05

range of like 10,000 to 15,000 ads a

27:08

month across all the companies that

27:09

we're working with.

27:10

>> Wow.

27:10

>> You know, in the aggregate. Across

27:12

Facebook ads, we understand what the ad

27:14

is, how it's doing, how does it fit in

27:16

the persona, what's actually working,

27:18

and then there's like a learning loop on

27:19

top of that. Like the volume is

27:21

unfathomable, but sorry, go ahead.

27:23

>> So you would say still that your the

27:25

bottleneck here is running up against a

27:27

very, very long form video that needs to

27:30

be edited?

27:31

>> For sure.

27:31

>> Say like a a two-hour live stream show

27:34

about e-commerce and one of those.

27:38

>> [laughter]

27:38

>> Absolutely. There are solutions that are

27:40

coming, and I think they're like really

27:42

close.

27:42

>> Um yeah.

27:44

>> Next six months I think we're going to

27:45

be in a place that the challenges that

27:46

you have to have So the things that you

27:47

have to solve you is you have to do a

27:49

word-by-word transcription

27:51

of like what's being said within it, and

27:52

then you have to like the agent has to

27:54

understand like frame by frame what is

27:56

happening.

27:57

>> Yeah.

27:57

>> And it's just expensive, right? You can

27:59

do this. It's just very expensive, and

28:00

it doesn't make sense. It's cheaper to

28:01

go and, you know, hire a homie out of

28:04

the Philippines for

28:05

1,600 a month, and they're just like

28:07

heads down basically, you know.

28:08

>> Yeah.

28:09

>> Like

28:10

pumping this out with

28:11

automations within Adobe Premiere,

28:13

right? Um

28:13

>> Yeah.

28:14

>> But it'll get to that place where it's

28:15

like cool, vision's got cheap enough,

28:17

like transcription is already cheap

28:18

enough. It's largely just that vision

28:20

piece of like understanding what's

28:21

happening in the video and like like

28:23

being able to stitch that together. Um

28:25

Once that's solved, it's like this is

28:27

done. Like that's

28:28

I mean then there's all this is like

28:30

what does B-roll look like and those

28:31

pieces, but I think there's I mean we

28:33

we're working with a company right now

28:34

that is like solving a lot of this

28:36

motion graphic stuff, and I'm like,

28:37

"Dude, this is insane." Like you're just

28:39

like giving it random

28:42

I I mean honestly, for your audience

28:43

it's probably great. It's the the

28:44

company's called Styleframe AI.

28:46

>> Um

28:46

>> Styleframe

28:48

>> You give it a key frame. So for example,

28:49

it's like here's four like key frames

28:53

of the

28:56

like I want an animation to occur. So

28:58

for say for example, like I think one of

29:00

the ones they have on their home page is

29:01

like I'm trying to make like like a

29:03

watch like a motion graphic for a watch

29:05

like that's like cinematic almost feels

29:06

like an Apple like style, right? You can

29:08

do that within their platform. Like it's

29:10

it's ridiculous like what is possible

29:12

and what's like on the forefront. So

29:14

anyways, yeah, just another nugget for

29:15

you guys. Again, if you're

29:17

like trying to build that type of

29:18

creative that feels incredibly like, you

29:21

know, cinematic polish, right? This this

29:23

there's there's tooling out there now

29:25

that that is actually possible to do.

29:27

This is pretty incredible.

29:28

>> Cody Schneider, man. Wow.

29:30

>> Let's go.

29:31

>> Love you guys. Love being here.

29:32

>> Hey, hey, everybody watch how far you've

29:34

come to. I've been like just you know,

29:36

in the

29:36

>> You've been along the ride, bro.

29:38

>> Awesome, yo. It's so

29:39

>> You've been along the ride, man. I I

29:41

think everybody needs to go to graft.com

29:43

B2B go talk with Cody. Cody, I think of

29:46

you as like Mr. GPM. I mean to me you're

29:49

Mr. Go to Market.

29:51

And anybody who wants that and wants to

29:53

get forward deployed

29:56

>> [laughter]

29:56

>> in the AI age should go and do that.

29:58

Brother, we appreciate you.

29:59

>> right now that are like trying to figure

30:01

out how they do their like wholesale

30:03

motion.

30:04

And like if that's interesting to you

30:05

guys, you know, anybody that's listening

30:07

we're all trying to discover that,

30:09

right? And like

30:11

build basically the system to accomplish

30:12

that where you're like, yo, I'm trying

30:13

to get into every boutique in

30:15

for example, the south eastern part of

30:17

the United States. So how do we go and

30:19

create a system around that? That's

30:20

something that's totally possible. So

30:21

reach out if that's in your wheelhouse

30:22

and you're trying to accomplish that.

30:23

>> Jack, we need to clip that and then we

30:25

need to put it [laughter] out for

30:26

everybody in CPG especially.

30:29

Oh my gosh. Cody, we appreciate you.

30:31

Cheers, man. Thank you, brother. Look

30:33

forward to having you back. Wow.

Interactive Summary

Cody Schneider, CEO of Grafft, discusses his innovative approach to go-to-market strategies using AI agents. He emphasizes the importance of data pipelines, analyzing competitor ad libraries to find 'blue ocean' messaging opportunities, and automating the testing of ad creatives at scale. He also explains the concept of 'TikTok real farms' as a way to generate affordable impressions and discusses the future potential for AI-driven ad production.

Suggested questions

3 ready-made prompts