Marin Istvanic joins Scalability School for a deep tactical session on what it actually takes to scale e-commerce brands to nine figures with paid media on Meta. Marin is one of the most hands-on media buyers in the DTC space, managing upwards of 5 to 7 million dollars a month in ad spend across his agency clients and his own internal brand (and this episode reflects that).
The conversation launches right into the hard hitting questions with the debate around ABO versus CBO and when each actually makes sense. Marin's full whitelisting flywheel (300+ influencer profiles, authority pages, dark posts running through multiple handles in the same session), cost cap weekend scaling and what he calls the "gladiators arena" ad set, how he uses different attribution settings as a signal to Meta that a campaign is structurally different, and why he still manually picks every thumbnail across all his accounts. He also shares how he used AI to synthesize his full YouTube library into a 50-page media buying philosophy document and then how he fed his top-spending ads into a landing page template that outperformed the control by 10%.
Key Takeaways
This episode of the Scalability School podcast is sponsored by NorthBeam and they just launched Northbeam Incrementality. Northbeam Incrementality gives you easy, automated, self-service incrementality tests, while protecting you from the major mistakes so many people make while running incrementality tests. Your MTA handles the daily tactics, your MMM guides the long-term planning, and Incrementality provides the causal truth. It's a closed loop that allows you to scale what works and cut what doesn't. Right now when you head over to www.northbeam.io/incrementality, they're offering Scalability School listeners 50% off unlimited tests for a year when you join. Just tell them we sent you!
To connect with Marin, send him a DM at https://x.com/IstvanicMarin
To connect with Andrew Foxwell send an email Andrew@foxwelldigital.com
To connect with Brad Ploch send him a DM at https://x.com/brad_ploch
To connect with Zach Stuck send him a DM at https://x.com/zachmstuck
Learn More about the Foxwell Founders Community at https://foxwellfounders.com/
Learn More about the The Hive Haus Creators Community at http://HiveHausUGC.com
Scalability School is your cheat code for building a more profitable DTC brand or agency.
We break down tactical strategies, real-time wins, and tough lessons from three powerful perspectives: operator, agency, and community. It's not theory—it's what's actually working in the wild. Skip the fluff, steal what works, and scale with confidence.
If your hit rate of the tests is high
enough, I think you can
squeeze more with ABO.
If your hit rate is
bad, then stick with CBO.
It never happens that a
good ad is not running.
You know, it's always running somewhere.
Maybe if it died in testing,
it's running somewhere else.
So like often the first thing that I do
when I take over a new ad account,
I go like a year or two back and I find
all ads that are above the target ROAS
that spend significantly.
I take their post ID and in most cases,
like seven or eight out of ten,
like ends up being performers.
People just forget that they had a winner
like, you know, a year ago.
That winner is a still winner.
Obviously, it's not
going to scale massively,
but I have some ad accounts where like
those old ads are like
running for like a few years.
It felt like I got a recognition when
Facebook confirmed
that a different thumbnail
actually means different entity ID
because like I was, I did not know why,
but like four or five years went back in
the day when I worked with Yudi,
I realized that a different thumbnail can
impact the performance of the same video
in a different way.
So still me and all of my media buyers
and my junior media buyers,
everybody are picking
thumbnails manually.
I think it has a positive impact.
And now let's take a listen to the
Scalability School podcast.
Welcome to another episode of the
Scalability School podcast,
your favorite podcast for tactical advice
on running your
e-commerce brand or agency
or agency accounts under that are brand.
So here's the thing.
We have a legend on the podcast today,
and I'm going to pronounce publicly his
name properly or attempt to.
Marin Ishtvan Itch.
Did I get close?
This is perfect.
Okay.
Who I have been mispronouncing my good
friend's name for three years
since we've known each other
or four years actually
we've known each other.
So I also, this is a theme of my life
because I also have the owner of socio,
Karen.
I've been calling Karan for six years.
And he recently just had the guts to tell
me that I was mispronouncing it.
So anyway, here we
are pronounced properly.
Everybody knows and loves this guy.
I've sent so many people to your YouTube
channel to learn how to do Facebook ads.
And an absolute legend.
Marin, welcome to the show.
Yeah, I appreciate it.
I'm not sure how I'm going to
justify all these big words,
but I definitely want to return the
gratitude because like I
was listening to your podcast,
you know, 2019, the e-commerce influence,
probably like I've stumbled upon it
randomly during one of my,
I know, searching on the
internet for a good content.
I was like, wow, how did I
did not know about this guy?
It was like in the early beginnings of my
media buying when you
were definitely a role model.
So I appreciate that.
I mean, welcome to episode
one of Glazing the Podcast,
where we glaze each other and talk about
all the good things.
Brad, Brad's really smart.
Brad's smarter than me.
So there's the pride glaze.
So let's just jump right into it.
By the way, Marin has spoken at three
events now for the
Foxhole founders in Europe.
And everybody always comes to me and is
like, that was so great.
Because you're such a tactician, Marin,
with your hands on the wheel.
So as a tactician, let's get into the
debate that everybody
wants to know right away,
which is ABO versus CBO.
Okay.
What are your thoughts?
We're only going to spend a few minutes
on this, but everybody wants to know.
What are your thoughts?
So if your hit rate of
the tests is high enough,
I think you can squeeze more with ABO.
If your hit rate is
bad, then stick with CBO.
Yeah.
I think I don't disagree
with that for what it's worth.
Like I know I'm on the
opposite side, generally speaking,
when it comes to the ABO versus CBO.
I think if I had to
add my little summary,
I think the bet that you're making when
you choose between them
is that you are going to either be right
more often than meta
or at a bigger scale that kind of offsets
maybe the downside of being wrong.
And I think you've
said this before, right?
Like that's a function of hit rate, but
it's also a function of scale.
If you hit a winner that
meta missed on and it can scale
and weigh in excess of what you spent on
the creative test that didn't hit,
it probably doesn't
actually end up mattering.
The only reason I
generally start with CBO is,
and for a lot of people is maybe their
hit rate isn't great.
And it kind of prevents the downside,
especially as brands are
kind of getting to scale.
Yeah, definitely.
So CBO is like more conservative.
And like when the budget is not high,
when we have too many
tests or the hit rate is low,
I use CBO when I'm into scaling.
And I know that like from 10 tests,
I will get like at
least three, four winners.
That means I'm getting three, four
winners every week from the 10 tests.
If I test 20, I get seven, eight.
Maybe I will not be able to scale them
like, you know, from 100 to 2K.
But like if I scale it from 100 to 500,
and I have like seven
of those ads a week,
that adds up.
And I don't need to scale every ad.
I need like, I don't know,
I need one, two great ads.
And then I need like
10, 15 supporting winners.
And ABO allows me that
a bit more than the CBO.
But I definitely agree.
Yeah, sorry.
No, I didn't mean to cut you off.
I was just going to ask, like, have you,
have you gone down the route or do you
have a kind of a tactic for,
I saw somebody on Twitter X recently say,
you know, they will
run their creative tests.
And then if they don't work,
they'll still maybe throw
them in to a cost cap campaign
or kind of keep it live somewhere else.
Like when I go CBO, it's
generally, I'm keeping the ads on.
I just don't generally turn them off
unless there's a good
reason to turn them off.
Maybe they do end up sucking
or they're super out of season
or something like that.
But I'm just wondering
if there's like small,
little tactical things
that you pull forward.
Yeah, so I don't have that
kind of like zombie campaign.
The reason is anytime
ads shows attraction,
I would take it post ID and I
would run it somewhere else.
So I never, it never happens
that a good ad is not running.
You know, it's always running somewhere.
Maybe if it died in testing,
it's running somewhere else.
So like often the first thing that I do
when I take over a new ad account,
I go like year or two back
and I find all old ads
that are among the target rows
that spend significantly.
I take their post ID and in most cases,
like seven or eight out of 10,
like ends up being performers.
People just forget
that they had a winner,
like, you know, a year ago.
That winner is a still winner.
Obviously it's not gonna scale massively,
but I have some ad
accounts where like those old ads
are like running for like a few years.
Yeah.
Do you have a cadence for like going back
and revisiting old ads?
Cause like maybe you guys
even have turned things off
in the past.
Yeah, that's a good thing.
As I said, like there's always,
this ad that was decent,
it's always running somewhere.
So when I take over,
I find those ads and
they're always running
with some kind of budget,
with a different audience,
with a different attribution setting,
with a different bidding.
You know, they're
always running somewhere,
at least with some budget.
So I don't get into position
that there's an ad that was good
that is currently not active.
Yeah.
Well, you said
something interesting there.
I don't want to, you know,
we can start to deviate off the ABO
versus CBO a bit here,
but you mentioned the different audience.
And so I think some people hear audience
and they're like,
wait, you're actually doing,
like you're doing more than just broad.
And so maybe,
maybe just want to
spend a couple of minutes
talking through that.
Yeah, I know it's 2026.
And like,
if I said that,
like I'm using interest,
people are saying like,
okay, what, why?
But the thing is,
like I find they can have cheaper CPM,
they can get me lower frequency,
and it's my way of
amplifying the winners.
So let's say if I have
an ad in testing that,
I don't know,
it was running for two months,
then it died,
I would take it post ID,
I would group it with some other winners
that also died,
and I would run them with
some interest audiences.
And as I said,
like it's,
it's a nice addition.
It's not something that's scalable.
It's not something that like,
you know,
is massive unlock.
But let's say if I'm
spending 20K in testing,
I'm spending 20K in scaling.
And then like,
there's like 5K a day
that's running in interest audiences
where I badge the winners.
Are these ads that
you're running predominantly?
So I would say 70% of my clients are US,
10% is Canada,
10% Australia,
10% UK.
Yeah, yeah.
Yeah,
it's,
I,
so it's interesting about the,
the interest thing.
I think I agree with you
that there tend to be marginal wins,
especially when you're spending
at a really high level with interest.
And I don't think
that's necessarily wrong
to do that.
So I appreciate you calling that out.
So let's talk,
let's transition to talking about
tactics that are working right now.
There's a lot of
things that you talk about
on a regular basis.
One of them is whitelisting at scale.
This is talking about
a ton of whitelisting profiles
that are running,
you know,
not partnership ads,
dark posts that don't show
on the influencer's timeline,
multiple influencer touch points
in the same session.
Yeah.
So you went and said,
whitelisting,
you spent 1.2 million versus brand
and the cheaper CPC,
it was a dollar cheaper CPC
and a way better North Beam ROAS
at 1.29 versus 0.76.
Talk about whitelisting scale
and your sauce there.
Yeah.
So basically we realized
that no matter the client,
when we introduce whitelisting,
especially from the authority pages,
so let's say if you're in pet space
and you find a veterinarian,
if you're in,
I don't know,
pain,
then you find a like pediatrician,
if you're like in a
skincare dermatologist,
so like those people
tend to have like bigger influence
on regular users.
They add credibility.
The fact that
when you're running through
the multiple pages
at the same session,
you can see different ads
at the same time,
although it is the same advertiser
just running through
the different pages.
When you run through
the brand page,
you can see one ad per session.
So it creates that sense
that everybody's using the product.
So we then just like
built a massive flywheel
of these influencers
that we actually put
into our ad account.
So like for our internal brand,
I think we tested
more than 300,
400 whitelisting profiles.
So basically every week,
I'm testing like between
five and 10 new profiles.
And like, again,
it creates a sense
that like every single authority
in this space
is using our product.
Obviously, this is quite expensive
at the scale
because, you know,
these influencers
that are authorities,
like I have some girls
that we're paying
like 20K per video
plus usage, right?
But those are like,
you know,
those are proven influencers.
Those are authorities.
Those are people that I know
no matter the ad I launch,
it will be a winner.
So I pay 20K,
but I spend 20K a day.
So basically I offset that
in a week, you know.
Yeah, there's,
and there's a bunch of different
like levels to this.
So I'm curious,
like does the,
does the partnership,
whitelisting,
however,
like all of that,
does that differ
depending on the scale
that the brand is at?
Like if they're spending
less than 2K a day
or less than 5K a day,
are you approaching
in a different direction?
It sounds like maybe
you try to go
for the expert and authority people
first,
regardless of scale.
Yeah,
obviously in clothing,
you cannot,
there's no expert.
So you just go people
that fit your avatar.
Obviously the less you,
the less you're spending,
the less people
you can actually have
because they're
sometimes more expensive.
But we try,
I would say like
as you scale,
you get actually
more of those people.
So that's the,
that's the main,
it kind of like
goes with the volume.
Yeah.
And the two follow-up
questions to that,
how are you finding
these people in general?
And then the follow-up
to that would be,
how are you,
and I know it depends,
you just,
you just said it depends
based on the person,
but like,
what does the general
partnership agreement
look like with that person?
It's fixed plus percent
to spend,
it's just percent to spend,
who cares what that
would look like?
Okay,
good question.
So first question,
how do we find them?
So we have,
for our internal brand,
we have like four people
in the influencer team
that is just like
finding people
and negotiating.
We leverage some tools
like Facebook
Creator Marketplace.
It was a terrible tool
until a few months.
Now it's amazing tool
and my influencer team
tells me that like
they find more people there
than any other software.
Also,
we started using,
I think it's Super Bloom.
It's an agency
that helps that.
Basically,
but like we're on that level
that we can pay,
you know,
four or 5k
for an influencer
because we know
there's a script
that works.
Obviously,
when you're smaller,
it's just like,
you know,
reaching manually
to people on Instagram
or to create a marketplace.
Facebook marketplace
is better
because you know
these people already know
what that process looks like.
They already have
the pages set up.
So it's easier
in terms of the payment.
It's mostly,
so initially we have deals
that we pay them
for content
and then we repurpose
that content
for an ad.
Now we are just paying them
to get the content.
So you don't even have
to post it on your page.
I just want your handle
and I just want
to be able
to run the ads
through your page.
They give us video,
raw video.
We chop it up
into three,
four different hooks.
We send it for approval.
We send the ad copy
for approval.
We send the access
to get the,
a request
to get the access.
Once they approve
that,
I have the access.
I launch that ad
as a new ad set
in my testing campaign.
And literally,
there's no brand
that we work with
that did not have
a positive impact
from introducing
whitelisting.
Yeah.
Do you see
that there's,
does the,
does the type
of influence
or maybe type
is the wrong question.
It's like,
does the scale
of their following matter?
Have you seen
any correlation
or not really?
So,
the big ones
definitely work better,
but
smaller ones
can surprise
and can also,
you know,
get you
some surprising wins,
especially if the,
if the video is good,
but I still think
the sauce is in the script.
It's just like,
this is like
amplification.
Why I don't like,
you know,
running partnership
because when I see
a person
with brand,
it screams
that this is an ad
to me.
Like,
it's just,
I know this is an ad
and I scroll past it.
When I see just a person,
okay,
this like looks something
that I could,
I could engage with
and I think
that's why it works.
Well,
I forgot to ask
and answer the question
about the pricing.
So,
it really depends.
Usually,
it is,
it is a fixed fee.
We try to get them
on an affiliate,
like on a percentage
of spend
because that's better
for me.
There's a lower risk
and I'm okay paying
you more
if you're actually
getting me the results.
It will be
hypercritical for me
because I charge
percentage of spend.
So,
you know,
I think that's a fair deal
to get to the influencer,
but majority of them
are still like,
I don't know.
I would rather you pay me
and then like
in 30 days,
we evaluate.
If we have 30 days access,
we evaluate
whether
the ROAS
that they got
and the spend
is like revenue,
whether that offset
their fee
and whether we actually
made money.
If yes,
then we kind of like
trade to negotiate.
I have some person,
I have some people
that like we are extending
rides from July
last year,
you know,
because it just works.
So,
it's like running
for like nine months
and we're just extending
every month.
Yeah,
I have one more question,
Andrew,
and then I'll pause
and shut up for a second.
But this might transition
actually into one of the
follow-up questions here
or the next topics,
but are you doing anything
special with the
account structure
for these tests
or are they just going
into their own,
they're going into
their own creative
testing campaign?
They're just going
into their own ad sets.
So,
each creator
is in its own ad sets.
So,
if I'm launching
20 tests a week,
I have 20 ads
a week
that I set
and I launch
tests from
Wednesday or Thursday.
I avoid launching
stuff on Monday
and Tuesday
because my goal
is to get
at least one
winning variation
of these
video that they
get me.
So,
I know it can
sound overwhelming
because like,
I know,
in a month,
I could potentially
have 100 tests,
which is 100
ad sets.
So,
someone could see
that and like,
okay,
what is this?
But actually,
out of those 100,
I'm getting like,
you know,
40 that are
decent performers.
Very good background
on whitelisting
and how it's
working for you.
So,
I appreciate you
getting into that.
One other thing
you've talked about
is cost cap
weakened scaling,
which is
ASC
with cost cap
using different
attribution windows.
seven-day click
versus seven-day click
plus one-day view.
And,
essentially,
the different
attribution forces
people kind of
Facebook to find
different types of people.
And,
with the cost cap,
your point is like,
look,
Meta does the scaling
for me.
You talk about
this strategy
and what else
is how you do it,
how you set it up
more specifically.
All right,
friends,
quick break.
This episode
is brought to you
by Northbeam,
the marketing
attribution platform
that we love
over here
at Scalability School.
If you're a marketer,
I mean,
you probably agree,
right?
Incrementality testing
is broken.
These tests require
tons of labor,
spreadsheets,
statistical reference,
and the designing
of a test
is slow and error-prone.
How are you supposed
to know what to test
in the first place,
right?
There's a million factors
that can make
incrementality testing
difficult.
But,
the good news is
Northbeam has completely
changed the game
and they fixed this.
Only Northbeam
incrementality
automates your
incrementality test
design and monitoring
while protecting you
from the major mistakes
that plague
incrementality testing.
Northbeam builds tests
using your own
MTA data,
launches them directly
in ad platforms,
and automatically
monitors them
throughout.
If something
threatens validity,
you're alerted
before it ruins
the result.
With Northbeam,
you can launch
tests in platform
with a single click,
saving you tons
of time and reducing
any chance
for errors.
And when the
test concludes,
your results flow
straight into your
MTA,
meaning you can
easily see results
alongside the rest
of your data.
Incrementality
was broken.
Northbeam fixed it.
Point is like,
look,
Meta does the scaling
for me.
You talk about
this strategy
and what else
is how you do it,
how you set it up
more specifically.
Yeah,
so everybody
knows that I don't
like Cost Cap
for testing,
but I really,
really love them
for scaling.
I realized that
two years
for one Black Friday
and I was like,
okay,
I have this scaling
campaign,
but I realized
that I don't want
to have the same ad
in a different campaign
without a reason.
So,
okay,
if it's different
optimization,
okay,
highest volume
or highest volume,
value,
yeah,
yeah,
so if I move that
or if I change
the attribution setting
or if I add interest,
that in my eyes
tells Facebook,
okay,
this is slightly different.
So,
I'm okay
with kind of like,
I have the tests
in my testing
and I keep scaling them.
Sometimes I get
from 100 to 300.
Sometimes I get
from 100 to 5K
just because tests work
and I continue
to increase the budget.
When I move the winners
into scaling campaigns,
so my scaling campaign
is on 7 Day Click,
I call this
winners of the winners.
It's like a gladiators arena.
It's just one ad set
and all of the winners.
I don't care
which ad is going to win
because all of them
are winners
and I'm comfortable
using cost cap
because all of them
are winners.
So,
if I put cost cap
of 50,
Facebook spending 55,
I'm cool
because it giving me volume
but I know
it will never spend 80
because this is a proven ad.
So,
I'm okay,
you know,
leveraging manual bids
for scaling.
What you mentioned,
Advantage Shopping Plus,
the fact is
that's the old
Advantage Shopping Plus
where we could define
that percentage
of spend for buyers.
The thing is,
like,
I never turned it off.
So,
in 7 out of,
like,
say,
10 ad accounts,
I still have that OG
Advantage Shopping Plus
running.
If I do not,
then it's just one campaign
with one ad set.
As I said,
I limit it to 7 Day Click
because I found that
that old Advantage Shopping Plus
was doing a bit more retargeting
regardless,
you know,
similar setup.
So,
I don't want Facebook
over attributing
on view sales
that is not
responsible
and what's the beauty
of this weekend scaling?
I realized that
conversion rate is better
because more people
are on the phone.
So,
there's,
I never knew
to what extent
I can scale on a weekend.
I know I can scale
but I don't know
whether I can increase
from 1K to 2K
or from 1K to 5K
and how to increase
that budget
without actually disturbing it.
So,
if you would,
like,
look a graph
of my campaigns,
Andrew,
I think I shared that
on the Elizabethan presentation.
My testing campaign,
like,
goes like this
because I,
like,
I just increased
the budget
of what's working.
My interest campaign
is,
like,
it's always steady.
It's always on,
like,
let's say 5K
but the cost cap one
is,
like,
you know,
it's like this.
On Tuesday,
it spent,
I don't know,
3K.
On Thursday,
it spent 7K.
On weekend,
it spent,
like,
25K
just because it can.
And I'm okay
giving it,
getting a higher,
slightly higher CPA
because it brings me volume.
I have a million
follow-up questions
and,
like,
randomly specific things.
So,
I might be bouncing
around a lot here.
You mentioned
the attribution settings.
You mentioned
a bunch of different things.
Like,
you could change something
about the campaign
or the targeting
or whatever.
Like,
some of those settings
to tell meta
there's something
different about this.
Exactly.
Do you actually
scale by changing?
Like,
are you setting up
maybe new ad sets
with 7-day click,
7-day click,
1-day view,
7-day click,
1-day engage view,
1-day click?
Are you actually playing
with those
as a scale mechanism
at all?
So,
my first campaign
is ABO testing
on highest volume
on 7-day click,
1-day view
unless there's
too many view conversions.
So,
basically,
when I check
into the ad account,
if there's more
than 25% view conversions
and AOE is not 400+,
I would use
exclusively 7-day click.
If,
you know,
everything looks fine,
I would use 7-day click,
1-day view.
Then,
I would move
the winners
into that audience campaign
as I mentioned.
So,
that's kind of like,
okay,
that's a different direction
because I'm using audience.
And then,
the third direction
is this 7-day click
on manual bids,
which is combining
two different things.
So,
I think I pretty much
covered that.
Sometimes,
recently,
I started playing
with incremental
attribution
or when the AOE
is higher
or there's,
like,
a lot of skews,
then I would play
with the volume.
But I think I cover,
like,
four out of five things
in those three campaigns
of mine.
So,
you mentioned
the OG AFCs
and,
like,
the ability
to set the percentage.
Do you have,
do you have,
like,
a playbook
for retention campaigns
or actually trying
to go and reacquire customers?
Most of my customers
do not
want to,
like,
reacquire customers.
Some,
I'd say,
like,
okay,
but,
like,
if someone did not
bought from you
for,
like,
I don't know,
six months,
do you consider
that person
a new customer?
If they say yes,
then I would create
a campaign,
but I would create
a separate campaign
that's targeting.
I would never allow
that to be
into my prospecting
campaign
because you can
never trust
Facebook with that
if you just go
for the lowest
conversion.
So,
I would set up
a different campaign
going only
for those customers.
And that's mostly,
you know,
in some skincare
or clothing.
And also,
during the promotions,
people are,
brands are okay,
they're fine
with going after
those customers.
So,
then I would set up
a campaign like that.
Yeah.
I have a question
about just overlap
in general.
Like,
you have multiple campaigns,
you have a ton
of different ad sets
going on.
Do you concern yourself
with overlap?
Or is like,
hey,
the number is going up,
I'll pay attention
to CPMR
or something like that,
but like,
otherwise,
the number is going up,
it's fine.
Yeah.
So,
I think,
like,
I definitely track
CPMR and frequency
and like,
I'm cool with
leaving an ad
active if it has
a bad CPA,
but it has a very
low frequency
because it tells me
that it's feeding
top of the funnel.
But I'm not going
to turn off an ad
that has a good
performance just because
it has a bad
CPMR.
And I think
that it is
healthy conversation
that like,
everybody are starting
to measuring that.
But I think
we are going
like in a bit
like too extreme
that like,
okay,
CPMR is better
than the CPA.
That like,
that's not the case.
Obviously,
it is important,
but it's also important
that people see
your ads
multiple times.
And when I have
like,
you know,
30 different influencer,
I have 10 different angles,
I have,
you know,
that creative diversification,
they're not actually
seeing one ad
for eight times.
They're seeing like seven
different ads.
So I think that actually
also is beneficial
to an extent.
Obviously,
I'm tracking that
with a tool
that Andrew and Kurt
built,
which is super helpful
for net new reach.
And that's obviously
one of the metrics
that I measure,
as well as the percentage
of new visits
in Norbeam.
But it's not,
you know,
I'm not going to turn off
a good ad just because
it has a bad frequency.
Yeah.
Okay.
Then my last question
related to this is,
I mean,
it's kind of overlappy,
but are you using
the actual like experiments
tab in Meta
at all for anything?
No.
Okay.
Really interesting.
So no holdout tests,
no A-B testing
directly in Meta?
No,
no.
If you're doing holdouts,
it's mostly through
House or WorkMagic.
When you've done
holdout tests,
what have they
shown you
that's been useful?
Cool.
I ask because
there's a lot of
discussion about this now
and, you know,
Northbeam has
an internal tool
that is doing this,
which I think
is actually pretty good.
And it's,
I think it's something
that as,
especially as budgets
go up and complexity
starts to get bigger
that like,
it's huge for people
to understand like,
okay, look,
you know,
so much of our traffic
from YouTube
is actually converting
on Amazon or whatever.
So like,
what types of stuff
are you,
have you seen
that's been surprising
or helpful to you?
Yeah,
potentially I can touch
base on the incremental
attribution.
Again,
like I know this is
like a hot topic now,
but when I do the breakdown
in my ad accounts,
like incremental
turns out to be
90 to 95%
of seven day click attribution.
So it's not really,
you know,
something game changing.
I think it's like
a tool in a toolbox,
but I would not necessarily
change my whole ad account
to incremental
just because
Twitter is talking about it.
In terms of the tests,
so I did try,
you know,
add to cart
and view content,
never worked for me.
Like I did five tests
with two different platforms,
never worked for me.
So I'm not sure what,
how come that's working
for Ridge,
but I mean,
I guess it's working.
In terms of the other channels,
we found out definitely
what you mentioned,
YouTube has a way bigger impact
on Amazon
compared to just Shopify.
We found that
Google third-party pages,
third-party funnels
have bigger impact
on overall performance
than if it is visible
in the ad account.
We also had some success
with Vibe
on the CTV,
but we realized
that it's very hard
to, you know,
understand whether Facebook,
which is spending like,
you know,
20, 30 times more
than the Vibe,
has an impact
on good performance
of Vibe
or Vibe has a good impact
on the Facebook.
So we are going to do
some additional tests there.
Besides that,
still did not do
app loving tests,
but that's on one
of my clients' maps.
So interested to see that
because I know everybody
are, again,
think how they scaled
app loving.
I never managed
to scale it.
No matter if I'm spending
50K a day on Facebook,
I could not get app loving
to spend more than 2, 3.
So I'm not sure
if I'm missing something
or,
but yeah,
those are kind of like
the main takeaways,
at least from,
you know,
a couple of clients
that are doing those tests.
The only thing I would say
on app loving,
the only,
we've seen,
we've seen similar things
to you where it's like
most people just don't hit scale.
But we had a conversation
with Miranda months ago
and that was like,
I wrote down a whole,
I had a whole notebook
of notes,
which I'm about to have
with you.
And I went back
and I just like picked
a couple of things
and the thing that hit
the best for us
was changing the hooks.
We did port over
the meta
creative directly
and she said,
she basically said,
don't do that.
But she says like,
as a starting point,
it's fine to do,
but that's not where
you're going to get
the biggest bang for your buck.
The biggest thing
that made a difference
for us from a creative perspective
was literally just
swapping out the hook
to like be something
about addressing the fact
that we just interrupted
their game.
Like, hey,
we'll get you right back
to your game
or sorry to interrupt your game
or like something
literally directly calling it out
and it made a huge difference.
Like the top performing ads
were all starting
with that hook
and the rest of the,
the rest of the,
the video is the same.
So interesting.
I'll be curious
to hear what you...
How did you then try
to get that creator
to film a hook
or was it the hook
totally unrelated
to the creator?
Yeah.
Because if you get
like 20 winners
and then like you get to,
you have to ask
like 20 people
to actually refilm the hook.
So was it just generic hook?
Yeah, that's a really good question.
In this specific case,
the ones that stood out the most,
it was an AI voiceover
with creator B-roll.
So we just,
we were able to,
the videos that were working
in general
were just kind of like
B-rolled with voiceovers.
So we were just able
to swap out
the messaging
with an AI voiceover.
That's...
But I bet you'd have
more success
by like maybe taking
your five best creators
and asking them
to film something
really specific.
That could be interesting.
I haven't been able
to do that yet.
So if you do,
you have to let me know
how it goes.
Sounds good.
Yeah, you should start
with like,
hey loser,
get off your phone.
Yeah, you're just
close to life.
Yeah, exactly.
Your kids in front of you,
like why aren't,
you know,
make them feel really bad.
That usually works.
So let's talk about
the hack segment
a little bit briefly,
which is like,
what's the closest thing
right now to a hack
that you've seen
inside meta?
So you change this,
it's like,
and it is a dramatic shift.
What are like one
to two things
that you,
that you're adjusting
that you're like,
man,
that's really an unlock?
I mean,
I like,
I like hacks,
but like,
I don't think like hacks
are just a nice addition,
you know,
like having a proper exclusion,
having not reliance
on existing buyers,
no viewed conversions,
like knowing when to scale,
knowing how much to scale,
that matters the most.
And although I'm a media buyer,
like,
you know,
product,
offer,
funnel,
creative,
they have way bigger impact
than the media buying.
I can make results
when like client is spending,
like everything else
is dialed in,
then the media buying
again makes a difference.
But you know,
at like,
at lower level,
probably like focus
on the offer,
offer,
like I've seen tremendous changes
in the hit rates
just by changing the offer.
So that's definitely
one thing outside
of the ad account.
Also remember,
everything else
outside of the ad account
has a tremendous impact
on the ad account.
In terms of the ad account,
I still pick
all of my thumbnails manually.
And like,
I just kind of like,
it felt like
I got a recognition
when Facebook confirmed
that a different thumbnail
actually means
different entity ID
because like I was,
I did not know why,
but like,
I like for four or five years
when back in the day
when I worked with Yudi,
I realized that
a different thumbnail
can impact the performance
of the same video
in a different way.
So still me
and all of my media buyers
and my junior media buyers,
everybody are picking
thumbnails manually.
I think,
I think it has
a positive impact.
Although I must say,
Facebook is getting better
and picking that
automatic thumbnail
compared to,
let's say,
a year before.
Yeah.
It's funny.
You basically don't like hacks
and I agree.
Like we put it in there
intentionally
because it kind of like
pushes you to say,
like pick,
pick,
have an opinion about hacks.
And it's,
it's hilarious
because like you probably feel
like you're repeating yourself
pretty often
and maybe the content
that you make
and you still get people
that are like,
they don't have exclusions.
They're not using,
you know,
it's like you still see
a million people
who are just doing crazy shit
and wondering why nothing's working.
And sometimes it's just like
those simple things
ends up being the hack for them
because they weren't doing
in the first place.
Yeah.
So no,
that's funny.
One thing that we've,
I'll throw in,
I'll throw in one little hacky thing
just as something
that continues to work for us
is just like customers
that are uploading images
as reviews,
just taking them as is
and uploading it as an ad
gets you creative volume,
gets you creative diversity.
They're probably super blurry
and that's the point.
Like it,
it just works really well
and we just keep doing that on repeat.
Yeah.
Now that you mentioned that
there's like something thing,
like usually I found that
for most problem solving product,
images do mostly retargeting.
We just do top of the funnel
while like on,
you know,
on clothing,
jewelry,
images actually do the prospecting.
So we found that like,
I could not,
like there's one client,
like we're spending 50K a day,
like 99% of my ads are videos.
I could not get images to work
and then we just replicated.
It's a beauty client.
We just replicated,
you know,
it felt like a story.
It was like a text,
organic text overlay,
like what you would put on a story
and it was just an image
and like those ads are getting,
I would say like 50% of delivery
on story and reels
and they feel organic
and this is something
that we finally managed to break out
with images
and got some traction.
Again,
I can't think because it looks organic.
It's so interesting.
I've been literally in the last,
I mean,
I've been thinking about it
for longer than this,
but like in the last two weeks,
I've had this like weird obsession
with like formats
where like you can communicate
the same message
in a bunch of different ways,
but also the type of content
that somebody is consuming
to your points,
like stories and reels
is different.
Like Zach's a great example of this.
Like they have compression socks
for Holo
and there's probably 80 year old dudes
that are buying it
and 30 year old nurses
and the type of content
that they're consuming
is very different.
It's like they expect
a different format
and so maybe it's the same,
it ends up being the same message
which your legs are going to hurt less
or whatever it is,
but how you communicate it
and just even the style of creative
can unlock different pockets
of audience.
So I like that.
That's a good one.
Nice.
And those are easy to make too.
Like those are easy.
Yeah, exactly.
They're super easy.
Slap a question and answer on it.
Exactly.
So how are you just,
how are you using AI by the way?
Like, and by the way,
if you haven't checked out
the new Foxhole podcast,
releasing or will be released
by the time this comes out,
AI D2C WTF,
which is a new podcast
I started up to talk about
all this stuff even more deeply.
But how are you using AI now?
Is it with reporting
or creating different variations
or is it all that stuff?
And if so, like what are you doing?
Yeah.
So I'm not on the level
that Will is.
Like when I see his conversation,
Will, that was on some of your podcasts
and like when I see the presentation
that he did in this,
I was like, okay, wow.
Like I would never be able to do that.
I did actually follow his template
and I managed to get close,
but I'm personally nowhere near that.
What I like to do it
is like for research,
for like synthesizing the data.
For example,
I gave it my whole YouTube video
for my whole YouTube channel
and like I told it to create
my media buying philosophy.
It created like 50 pages document
and then I uploaded that to Claude.
And now like, you know,
whenever there's a question,
I can get an answer from that
or like I can get a new script
for YouTube based on my whole knowledge.
That's one thing.
The other thing is
in terms of the ads,
I gave it my top five spending ads
and then I gave it one of my template
of a listicle page that we are running
and I told it to change the reasons
based on the reasons,
based on the angles from the ads.
And that landing page is currently
like that listicle is outperforming
our current listicle,
you know, by 10% of click rate
and like it's getting a nice traction
in Orbeam as well.
So that would be one
like practical example
of me using AI.
Besides that, you know,
like when you're doing media buying,
there's not much you can do
compared to when you're a copywriter
or developer or creative strategist.
So it's just like mostly trying
to kind of like improve my workflow
and save some time either on analysis
or on the reporting or
synthesizing the data.
I might be leaking something here.
What do you use for creative reporting?
It's mostly Motion or Atria
depending on the client.
Yeah.
Motion has an MCP connector now.
This is what I'm potentially leaking.
I don't know if it's
available to everybody,
but they gave me access to it.
It's awesome.
It's like, I don't know if you use
the chat feature inside of Motion at all,
but it's really nice.
You should reach out to them
and just ask if they can turn on
the MCP connector for you
because it can pull all those AI tags.
So like here's an example.
You mentioned landing pages.
You could say,
hey Motion, can you go grab
the top trending tags for me
in the last 30 days
for my messaging angle?
Okay, so it's going to grab that
and it's going to say,
here's how the spend breaks down
as a percentage of your ad account
for this angle, this angle, this angle.
You take the top spending one
and then it can grab the messaging.
So it can pull the scripts for those ads
and it can give you the scripts
and then it can synthesize it
and write a landing page
in the customer language
from those top ads.
If you wanted to,
if you could set up a schedule in Cloud,
like you could have that do it,
just like present you
a new landing page every week.
That's a little overkill probably,
but it's been fun.
Yeah, basically that's what I did,
but I did manually through Manus.
Hey, here are the ads.
Hey, can you transcribe me?
Hey, here's the landing page.
Hey, make it a landing page.
So basically it was me involving,
but if I can do that without me actually
instructing all of that,
that would be amazing.
So I'll definitely check that out.
Manus.
I haven't played with Manus too much,
but I feel like Zaka is going to
ban me from the world
if I don't start using it.
You know they're going
to be talking about it
at the Meta Performance Marketing Summit
quite a lot.
So outside of the things
that we've talked about,
you know, I think one of the things
that people look to you for
is very, you know,
deep tactical knowledge
and deep understanding of this.
How do you recommend people
upskill themselves on these things?
Like, I think there's so many sources.
I mean, and,
but I know that you sort of set out
in building your name around education
as well, right?
Like you're, you're very good at that.
How do you recommend people
start to get better at this
and really start to dive in?
So, I mean,
I don't know when I started,
I went through the like whole Google
and Facebook literature support
to understand how the auction works.
Obviously you don't have to do that.
I did that like eight years ago,
but like now I'm like spending five mil,
seven mil a month
and I'm still listening, you know,
to five different
podcasts on media buying.
I'm going to probably not heard
probably 90% of the stuff I already know,
but you know,
hearing Zach,
hearing Brad,
how he does,
you know,
that overlay of a testimonial image.
Okay.
That's a cool idea.
Hey,
doing this with MCP with Twitter,
you basically just need one good idea
because there's not
much that you're missing.
Obviously,
if you're on a lower level,
you have communities besides the podcast,
obviously,
Voxfield founders,
shout out.
If you're not there,
you have to be there
if you want to be serious
about DTC and e-com
and you have a library
of free content on YouTube,
like there's literally,
you can start a brand tomorrow.
You can start being,
I don't know,
whatever you want
because the information
is already out there.
You just have to,
you know,
stay disciplined,
stay consistent
and try to like put it into action.
It's not enough
that you just like read about AI.
You just have to try
to build something with it.
And I think that like,
you know,
transition from just absorbing
the information
to using information
is actually what makes a difference.
Well, Martin,
thank you so much for joining us.
A very rich episode
full of so much knowledge
and so much information.
And, you know,
I'm so grateful to you
for all that you've taught me,
all that you've taught founders
and all that you've now taught
listeners to the podcast.
So thanks for coming on.
And for those of you
that don't know Martin,
I recommend following him
on all the channels
that you can
because I always learn
something from your content.
So thanks for putting it
out there, my friend.
Yeah, I appreciate you having me.
I really enjoyed this episode.
It was really good conversation.
I think we covered
a lot from the technical stuff
and tactical all the way
to the holistic.
And yeah,
if you have,
if you guys have any questions,
just hit me up on Twitter.
I'll promise I'll respond
unless you try to sell me
your DM setting service.
Maybe,
maybe,
maybe throw in a couple arguments
on X as well.
You know,
a little banter back and forth.
All right,
take care.
This episode is brought to you
by the Foxwell Founders membership
that Andrew and his wife,
Gracie run.
It has been absolutely pivotal
for not just the Homestead team,
but the Easy Street Brands team.
We've had,
I don't even know
how many members
are currently in there
that are part of our ecosystem.
But when it comes to anything
from learning ads
to understanding what's going on
to building an agency
to knowing retention,
it's been absolutely useful
for our team
when they get stuck
or they need help
to just go there
and resource all the other experts.
So definitely would recommend it
for anybody that's looking to,
you know,
take it a step deeper,
try to get a little bit more knowledge
on growth marketing
and all the world DTC is.
Yeah,
I think one of the most
incredible things about it
is you can just like
open up the Slack group
every single day.
You can pin your favorite channels
for the topics
that you care most about.
And like every day,
there's going to be somebody
who just like,
because they want to contribute
something valuable
to the group,
you can go learn something
every single day
and it's going to be
extremely useful.
There's some ballers in there
that you just get like
the benefit of learning from
that like for the cost,
like you couldn't pay them
that for their time,
but through the membership,
like you get access
to some incredible people
and tons of resources.
Yeah,
I mean,
I think the biggest resource
to me too
is like the events
that,
you know,
Foxwell Founders does.
They've been able to do some,
even in Wisconsin,
even in the boring state
of Wisconsin,
which is pretty awesome,
getting people together
in person
and able to have really
just like honest conversations
of what's going on,
what's working for them now,
you know,
where they're at
in their business
and knowing that
there's going to be,
like Brad said,
some real killers
in the space
in this membership
that can help
and are willing
to take the time
and help.
So that's been
a huge part of why
a lot of our team
have really enjoyed it as well.
And the applications
are now open
if you're looking to join.
Foxwell Founders,
yeah,
foxwellfounders.com.
Go check it out.
Go apply.
The only way
that we grow this podcast
is by you sharing it
with your friends.
Honestly,
like reviews kind of
don't really mean anything
too much anymore.
They're really meaningful,
but they don't do a lot
for the growth
of the podcast.
And so sharing YouTube links,
sharing Spotify links,
sharing Apple,
whatever we call it
under the podcast app now,
anything you can share,
the better we're going to be.
Guys,
anything else
you want to say on this?
Yeah,
please go check us out
on YouTube.
Rack up those views for us.
We'd love to see it.
And then subscribe.
Make sure to subscribe
on YouTube as well.
And I relentlessly refresh
the YouTube comments
because it dictates
my mental health for the day.
So please say something
nice about all of us.
Thank you, everyone.
Thanks for listening.
Honestly.