Latest / The Yield Doctor with James Deaker / Retail vs Commerce Media Networks: Growth, Data & AI with Danilo Tauro

Transcript

Retail vs Commerce Media Networks: Growth, Data & AI with Danilo Tauro · The Yield Doctor with James Deaker

0:00
Every publisher wants a bigger slice of
0:02
ad revenue. But in retail, growth isn't
0:06
just about selling more ads. It's about
0:08
proving those ads actually drive
0:11
incremental sales. That's why retail
0:14
media networks and commerce media
0:16
networks are the hottest battlegrounds
0:18
in digital advertising right now. The
0:21
opportunity is massive, but only if you
0:24
can connect the ad spend to their
0:26
business outcomes. I'm James Deika, the
0:30
yield doctor, and today we're going to
0:31
be doing a deep dive into how to drive
0:34
that growth with Danilo Toro. Danilo has
0:38
a unique series of experiences.
0:41
He was the global director of media,
0:44
tech, and data at Proctor and Gamble. He
0:47
was the head of product for Amazon ads
0:49
for connected TV first and then later
0:53
for holding co agencies. and he was the
0:56
general manager at Uber Eats advertising
0:59
globally. He is currently a managing
1:02
partner at Cardigraphy AI, a company at
1:05
the intersection of startup accelerator
1:08
and enterprise matchmaking and adtech
1:10
and martekch. Danilo, that's a mouthful.
1:13
I am so pleased that we have the chance
1:16
to sit down and talk today.
1:18
>> James, thank you so much for having me.
1:20
Very excited for this.
1:22
>> Excellent. Now, uh let's start with the
1:24
basics. retail media network and
1:27
commerce media network. In some cases,
1:29
the terms get used interchangeably. So,
1:33
how do you define a commerce media
1:35
network versus a retail media network?
1:38
And in particular, what are the key
1:40
differences in the business models and
1:41
and where do you see them each having
1:43
the right to win?
1:44
>> Yeah. Yeah. I I love to start there. So,
1:46
we start with definition. So I I think
1:48
of retail media networks as basically
1:50
media monetization that's fundament
1:52
fundamentally anchored inside of the
1:55
retailers shopping experience. So it's
1:58
basically digital shelf if you think of
2:00
it. So you have like a sponsor products,
2:02
sponsor brands, on-site placements. It's
2:05
very close to conversion. So the
2:07
business model is relatively
2:08
straightforward if you want. So you have
2:10
auction dynamics, high intent and
2:12
measurable outcomes. I think when you
2:14
think of commerce media networks to me
2:17
they're broader because basically they
2:18
monetize commerce signals and commerce
2:21
experiences often I would say beyond
2:24
just the shelf and sometimes even beyond
2:26
a single retailer. So this is where
2:28
basically gets harder because basically
2:30
if your your commerce media network
2:32
pitch is just hey we have first party
2:34
data that's not enough anymore because
2:36
basically the differentiation has to
2:38
come from consumer experience that feel
2:41
I don't know native useful not
2:44
interrupted banners basically so they
2:46
really need to be integrated with
2:48
moments that actually add value and
2:50
connect what the consumer is trying to
2:51
do. So basically you you ask about the
2:54
the right to win and I think RMN's they
2:57
win when they have you know the right
2:59
assortment traffic conversion etc. So
3:02
it's basically the the store is the
3:04
media. I think CMN commerce media they
3:06
win when they can really create new
3:09
higherforming
3:10
ad experiences which basically can prove
3:13
outcomes across context not just on
3:16
site. I would say
3:17
>> I want to double click in this because I
3:20
have used a different definition which
3:22
is not to say yours isn't correct but
3:24
but help talk me through this because
3:26
when I've thought of retail media
3:28
networks I've almost thought of it as a
3:30
subset of commerce in the sense that
3:33
retail for me is more directly people
3:35
who sell usually physical goods. So if
3:39
uh Amazon or a Walmart or a Safeway
3:43
Albertson's, Target, those to me are
3:47
companies who are retailers who've
3:48
extended extended into having retail
3:51
media networks. Whereas I've thought
3:53
about other uh commerce media networks
3:57
as just being sort of a broader category
4:00
of other honestly more serviceoriented
4:03
companies, your Marriotts, your Uniteds,
4:06
Expedias, etc. and thought of them as as
4:09
being the companies that be more focused
4:11
on commerce media networks. So that's
4:13
how I've thought about it. How can you
4:16
match up what how I'm describing it
4:18
versus the definition that you've used?
4:20
Yeah, I I I think the way you describe
4:22
it is 100% correct and I like also the
4:25
the the cluster that you built because
4:28
ultimately you have commerce media
4:29
network which is this biggest bigger
4:31
bucket in a way then within commerce
4:33
media network you have a smaller one
4:35
which actually is making more money
4:37
though which is the retail media network
4:39
which is exactly you have you have a
4:41
marketplace so you have the buyer that
4:43
enters the marketplace with a high
4:45
intent to buy something and so this is
4:47
exactly where it's I would say easier to
4:50
monitor the type of intent because
4:52
basically you know merchants want to be
4:55
on top when there is a a keyword search
4:57
and then they're willing to pay. So
4:59
there is the CPC. So the search business
5:01
model with the CPC. So I agree with you.
5:03
Retail media network you may say is a
5:05
subset of commerce media network. But
5:07
then that subset I would say it's easier
5:09
to monetize because all of all the
5:11
reasons that we discussed.
5:13
>> Okay. Brilliant.
5:14
>> The the marketplace is there a
5:16
marketplace? Yes or no? If yes it's a
5:18
retail media network and it's easier. If
5:20
it's not the marketplace then it becomes
5:21
a bit more icky to monetize because
5:23
again video display it's it's a more
5:26
competitive space. So the absence of
5:28
that marketplace in the broader sense of
5:30
commerce media network that's what makes
5:32
the the the overall business model
5:34
different.
5:35
>> Okay. Both retail media networks and
5:38
commerce media networks are growing fast
5:40
but obviously without you know not
5:42
without friction and challenges from
5:45
from your perspective what are the
5:48
biggest opportunities and biggest
5:50
challenges that that they both face
5:51
right now
5:53
>> yeah so I would split it. So that's why
5:55
I I love to start from definitions. So I
5:57
think if we start with a with a broader
5:59
category, right? So commerce media
6:01
network. So I think the biggest
6:02
opportunity is really to unlock budgets
6:05
exactly beyond the shelf and these are
6:07
like video native sponsorship offsite
6:12
even creators I would say. And so then
6:14
the the the real challenge is how you
6:15
get off the ground with real
6:17
differentiation in this space because as
6:19
we just said display and video are
6:22
brutal categories, right?
6:23
It's difficult to differentiate. So the
6:25
question becomes what experience can you
6:28
create that people don't tune out? what
6:31
you know outcomes can you prove that
6:32
others can't and also to the buyers to
6:34
the brands how can you differentiate
6:36
because a banner is a partner a video is
6:38
a video yes first party data but again
6:40
how do you differentiate what experience
6:42
can you create I think when we move to
6:44
retail media network so again much more
6:46
targeted and again based on the
6:47
marketplace and all the good stuff that
6:49
we said so I think the early growth is
6:52
almost mechanical in a way right so you
6:54
drive merchant penetration you increase
6:57
ad spend as a percentage of basically
7:00
gross merchandise value, you reduce the
7:02
churn. But my point is once you hit the
7:04
plateau, once you're at basically, you
7:07
know, penetration and decent yield,
7:09
where does the next wave come from? And
7:11
so this is where basically I see the
7:13
biggest opportunities. So this is like,
7:14
I don't know, new formats, sponsorships,
7:18
offsite, if you think CTV and CTV, you
7:22
know, there are players that went very
7:23
hard on CTV. Take Amazon, you know, they
7:25
started with as a a traditional maybe
7:28
the first retail media network, but then
7:30
Amazon has, I would say, structural
7:32
advantages. They have Fire TV, they have
7:34
Prime Video. So for for them, it was
7:37
very natural to go to CTV. If you look
7:39
at Walmart, you know, they kind of
7:41
bought Visio, so the CTV operating
7:44
system. So I would say similar to Amazon
7:46
Fire TV, but I think the challenge is if
7:48
you don't have those, let's call them
7:50
unfair advantages. So, how do you
7:52
partner smartly and still create a
7:55
credible next chapter of growth beyond
7:57
traditional retail media network?
7:59
>> So, so let let's stay on this on the
8:01
sort of the growth opportunities because
8:04
I get asked a lot and therefore I'm sure
8:07
you get asked a lot. Uh there are so
8:09
many retail media networks and commerce
8:12
media networks. Are they all going to
8:14
survive? How do they keep growing? and
8:16
and I've talked about the fact that when
8:18
the original ad networks sort of started
8:21
around 2000 through 2005,
8:24
many of them reached a sort of plateau
8:27
around 2003 $300 million and very few
8:31
broke through to the next level. You
8:33
could argue that ad.com was a was an
8:36
outlier. And so I think there is this
8:39
challenge for retail media networks to
8:42
to the point you're making. they
8:44
monetize their own site, they do a
8:47
reasonable job and they start to reach
8:49
um some ceilings and they either have to
8:51
create new inventory or to your point go
8:55
off network. So as you think about those
8:58
particular areas, what if you were
9:00
running a retail media network, let me
9:02
ask a specific question. What would you
9:04
do when you start to reach those
9:06
thresholds and and when if the answer is
9:08
something like go off network, what does
9:10
that mean in your world? Yeah, and I
9:14
think off-site is one area and I think
9:17
you know I love this question because it
9:19
kind of builds on the on the previous
9:20
one and I think there are different
9:22
growth vectors. So I think of site again
9:24
is one of them and you know you can
9:25
think CTV etc. but then I think what
9:28
about data products for example so how
9:31
can you monetize your data better? How
9:33
how do you think about sponsorships and
9:35
I like sponsorships because sponsorships
9:37
could be something that you do digitally
9:39
or you also you know think of out of
9:42
home digital out of home physical etc
9:44
etc which is something that marketers
9:46
are asking more and more so brands are
9:48
asking for this blend of they call it
9:50
physical which I don't know if it's a
9:52
word that I like but it's really
9:54
physical but digital so again that's a
9:56
thought as well or if you think also the
9:58
the influencer space you know we know
10:00
many brands are doubling down on the
10:01
influencer creator type of commerce you
10:04
know digital commerce so I think that's
10:06
another vector if you think of we we
10:08
mentioned experiences so think of richer
10:11
on-site experiences so I think those are
10:14
all things all vectors that they can
10:16
explore and again I think these are kind
10:18
of underappreciated
10:21
you know uh areas that if you start with
10:24
merchant centricity then I think it
10:26
becomes much easier to understand what
10:28
are the experiences what are the formats
10:31
that can add value to the merchants to
10:33
the consumers and so on. So that's why
10:35
I'm what I'm trying to say is like most
10:37
merchants also they don't want to be
10:39
also media buyers right so they don't
10:41
want to understand all the details about
10:43
how this thing work but they want to
10:45
build good experiences for the consumers
10:47
they want to convert so I think this is
10:50
almost like a new mantra which is how do
10:53
I think of the merchants at the center
10:55
and all the experiences I can build for
10:57
them so that's a bit I would think about
10:59
the next wave of growth
11:02
>> you threw so much at me I'm going to
11:03
have at least A couple of different
11:05
follow-up questions just on this. The
11:07
first is when you use sponsorship in
11:10
this context. Sponsorship can mean
11:11
different things in different different
11:13
people. In some context, it just means
11:16
like a takeover, a take over the site or
11:18
take over the app for a day or it can
11:21
can mean something broader as you're
11:24
using the term sponsorship in this
11:25
context. What does it mean?
11:27
>> Yeah. So spot on. So I think the
11:30
takeover which is you know you open the
11:32
app there is a big ad and maybe connects
11:34
to a specific I don't know Black Friday
11:37
Cyber Monday or maybe it's the you know
11:40
taco day or the the donut day. My wife
11:42
is Polish you know I learned there is a
11:44
donut day. So what I'm trying to say is
11:46
like definitely there is the takeover
11:48
type of sponsorship but that's why again
11:50
you can think beyond which is how do you
11:53
think of sponsorship outside of your
11:55
app. Think of you know if uh people that
11:58
follow the sport for example is there a
12:00
specific sport event you can connect you
12:03
know you'll be in front of your couch
12:05
you want to get for example you know
12:06
online food delivery so as you're
12:08
looking at your your game your match you
12:11
know there is a banner there is a QR
12:12
code so my point there is so much that
12:15
can be taught in terms of um uh
12:18
sponsorship in the real world in the
12:20
digital world and in the real world so I
12:22
think I would really invite retaliator
12:24
to think outside of the box. So not just
12:27
the single takeover, the simple
12:29
straightforward takeover, but everything
12:31
that again has to do with the experience
12:32
of the consumer in a way.
12:34
>> All right. The second follow-up was was
12:37
one of the ideas you talked about which
12:38
was data product sale and again that can
12:42
also mean different things. Uh I think
12:44
it is a a natural thing for retail media
12:47
networks and commerce media networks to
12:49
think about because often they have very
12:52
good data but their opportunity for
12:55
advertising the space that they have
12:57
available on their own site can be very
12:59
limited or or in their app and so when
13:02
you think about that um talk to me about
13:06
what that means because I can think of
13:07
at least three different flavors uh but
13:10
I'd rather hear your flavors than mine
13:12
in terms of what a data sale looks like,
13:14
who might they work with? And does it
13:16
involve actually selling the data or
13:18
does it just mean creating different
13:19
segments that their sales team would
13:21
sell? Yeah. Yeah. I I love this
13:24
question. I was having a similar
13:25
discussion with the CMO of one of the
13:28
largest luxury brands a few weeks ago in
13:31
in Orlando. There was the ANA masters.
13:33
So, we were discussing exactly this
13:35
point. And I think there is the
13:37
programmatic way to think of data which
13:40
to be honest for me is a bit outdated
13:41
which is you know programmatic started
13:43
20 years ago. And I think the question
13:45
is how do we evolve that thinking what I
13:47
mean with that is in my mind it's not
13:49
really about selling the data. So the
13:51
discussion I was having with the CMO is
13:53
if you think of first party data for
13:55
example your own first party data are
13:58
important but they become much more
14:01
powerful when you can match them with
14:03
other people first party data. So now we
14:05
enter the you know the space of privacy
14:08
etc etc and the point is and that's why
14:10
I I love a lot the space of clean rooms.
14:12
So what I'm trying to say is like for me
14:14
it's not about oh let's sell the data
14:16
and do targeting again that's 20 years
14:18
ago for me the thinking is okay I'm a
14:20
brand or I'm a retail media network I
14:23
have first party data can I build
14:25
alliances where I know for example
14:27
whoever is watching a specific uh movie
14:30
whoever is buying food from a specific
14:33
restaurant whoever is purchasing
14:35
specific items on on on retail media so
14:38
can I understand a purchase behavior can
14:41
I then matchmake make and build
14:43
experiences could be joint experiences.
14:46
So that's why long story short what I'm
14:47
trying to say is like the thought is how
14:50
can I think of first party data
14:53
partnerships which get then get
14:55
activated in a compliant way for example
14:57
via data clean rooms that's what I meant
15:00
a lot when I talked about data so it's
15:01
more like innovative ways to monetize
15:04
data in a way that ultimately also add
15:06
value to the final consumer so it's not
15:08
just the basic retargeting
15:10
>> interesting so so let me let me talk
15:12
through it you know I do a lot of these
15:14
interviews so I learn more. I have
15:17
thoughts. Sometimes they're wrong. In
15:19
the instance you're talking about, uh,
15:22
the clean room would be you would be set
15:24
up by the retail media network with
15:27
primarily their own first party data.
15:29
You're not, it doesn't sound like you
15:31
think of the use case of sharing across
15:33
lots of retail media networks as being
15:36
the primary one in the short term. And
15:38
it would be set up for each advertiser
15:40
to activate against the the data set up
15:44
in that clean room. Is that how you
15:45
you're thinking about it?
15:47
>> For example, yes, 100%.
15:49
>> And then it could be the Exactly. So you
15:51
have a clean room which is the the
15:52
Switzerland if you will and then it
15:55
could be a clean room which is uh owned
15:57
by the retail media network or could be
15:58
a third party clean room where basically
16:00
advertisers and retail media network put
16:02
data in and then you can query the data,
16:04
you get insights, you drive the
16:06
activation. So that's the type of
16:08
business model that I think could be
16:09
very successful because ultimately again
16:11
as we started and this was the
16:13
discussion with the CMO your own first
16:15
party data are important up to a certain
16:17
point when you can have data
16:19
partnerships this is when it becomes
16:21
very very powerful and again to to to
16:23
give a very specific examples right
16:25
imagine you have an airline company and
16:28
you know who is flying a business of
16:29
first class and the other side you have
16:31
you know a luxury company and you know
16:33
who is buying you know very expensive
16:35
jewelry. So now imagine the 1 plus 1
16:37
equals 3 that can be done there, right?
16:40
>> And and talk to me more about the use
16:42
case. So I'm a retail media network. I
16:44
put my data in the clean room. An
16:47
advertiser, let's say Proctor and Gamble
16:49
wants to use it. Are they using it
16:52
primarily in my network or are they
16:54
using their DSP to then use that data to
16:59
buy across different publisher sites?
17:03
And if so, what might the economics look
17:06
like? Because when I put that data into
17:09
the clean room, I'm giving them an
17:11
enormous boon, right? They I'm giving
17:13
them something very valuable. Do you
17:16
think about it as being a one-time fee?
17:18
Do you think about it as being the
17:19
number of segments that they use? Give
17:21
me a sense for how the economics might
17:23
work.
17:24
>> Yeah, and I think the economics are very
17:26
interesting because ultimately with the
17:27
data clean room, you're no longer paying
17:29
for the data, right? because you're
17:31
you're more paying for the the queries
17:33
etc. You can have a flat fee. You can
17:36
have a data sharing where you're saying,
17:38
"Hey, I'm as you said, putting my own
17:40
data in your clean room as a value in
17:42
itself for you." Right? So, I think
17:44
there are different types of agreements
17:46
that can be done and those agreements,
17:47
let's be very transparent, will depend
17:49
also on the size of the retailer versus
17:52
the brands and how premium those
17:54
audiences are, how scarce those
17:56
audiences are. So, I think it's
17:58
difficult one to to categorize exactly
18:01
how that agreement should be structured.
18:03
But my point is uh uh so one there are
18:06
many different dimensions that should be
18:08
analyzed but then the second thought for
18:10
me is also it's not just about the
18:13
targeting in you know on site of site
18:15
again I would go also broader than that
18:17
which is if we think about sampling for
18:20
example if we think about emails so
18:22
that's why I keep pushing for what are
18:24
all the experiences that can be
18:25
activated when you can monetize or
18:28
leverage better your data
18:30
>> so I don't know which question I want
18:32
ask you first whether I ask you about
18:34
partnerships. No, I want to stay on this
18:37
uh this this question of the growth
18:39
limits of the retail media network. Do
18:43
you see a world where as different
18:46
retail media networks hit this cap that
18:49
I'm asserting will be there that we will
18:51
see a network of ad networks, a network
18:54
of retail media networks start to grow
18:56
up around specific verticals or do you
18:59
think that historically that's just been
19:02
too hard to sort of make the pieces work
19:04
with all the different incentives in
19:06
place?
19:08
>> So love the question and I think yes. So
19:10
if we look at the the dynamics, I think
19:13
retail media network I think they're
19:15
better positioned because ultimately
19:16
they will get the adoption and the the
19:18
the spend based on the fact that they
19:20
have merchants on the platform. These
19:21
are endemic advertisers etc. The
19:24
challenge is okay what happens with
19:25
commerce media networks right but again
19:28
I think do I believe they will create
19:30
networks ad networks I don't know if
19:32
necessarily ad networks but what I
19:34
believe is a trend that is starting is
19:36
the idea of uh if you look how media
19:40
commerce media I mean it's today there
19:42
is almost a silo because you have
19:45
programmatic spent robmatic media bank
19:48
where you buy display and video through
19:51
a DSP etc and then on the other side you
19:53
have for example sponsor listing and
19:55
sponsor listing you have you know
19:56
specific companies where you can
19:58
aggregate retail media networks and you
20:00
buy on keyword CPC versus programmatic
20:03
you buy audience CPM so my point is
20:06
there are silos so what I'm trying to
20:08
say is like to accelerate that growth is
20:12
there a way to get new demand so as a
20:14
written media network you can access
20:16
programmatic demand which is not
20:18
possible today because again one is CPM
20:21
and audience based the other one is CPC
20:24
and keyword based but my point is can we
20:26
create a space where we have a
20:28
programmatic buying of retail media
20:30
network so I think this is a space where
20:32
I think we will hear a lot in the next
20:34
future and I've heard a few companies
20:36
are taking some steps building like a
20:38
RNN kind of RTB protocol and I think
20:42
this is a space that I'm personally also
20:44
investigating a bit
20:45
>> excellent so so let's stay then on
20:48
partnership strategies because you know
20:51
often A retail media network is not the
20:54
primary business of the ret I would say
20:56
always. It's the secondary business.
20:59
They're often resources are in short
21:02
supply. You have to work out what you're
21:04
going to build yourself versus partner.
21:08
When you think about sort of some of the
21:10
key partnership decisions for retail
21:12
media networks on the and you know build
21:15
versus buy, how do you how do you think
21:17
about the partnership decisions?
21:19
>> Yeah. Yeah. So I I would say I'm very
21:22
opinionated here. So and my point is
21:24
I'll take a step back because I would
21:26
say in a word that's basically you know
21:28
this vaua which is a word that was
21:29
invented maybe 20 years ago but I think
21:31
is still very compelling today you know.
21:33
So vaua volatile uncertain complex and
21:36
ambiguous right so optionality is a
21:39
strategy actually fun fact my my PhD is
21:42
actually in real options valuation which
21:45
is in a world where there is a lot of
21:47
uncertainty building flexibility as a
21:50
value but then that flexibility as a
21:52
cost so how much are you willing to pay
21:54
for that flexibility but anyway that's a
21:55
longer discussion but I I'm a big fan of
21:58
optionality so optionality as a strategy
22:01
and so my point is you don't want to
22:03
kind of overbuild the wrong thing and
22:06
then you get stuck. So my rule of thumb
22:08
is build what creates you know durable
22:12
differentiation but then also partner
22:14
for capabilities where specialists are
22:16
already ahead and basically they will
22:18
stay ahead and it's going to be
22:19
difficult to you know it's going to be
22:21
expensive to to build yourself right so
22:23
if you think of stuff like I don't know
22:25
measurement clean rooms you know
22:28
creative formats now if you think all
22:30
the geni etc etc you know this offsite
22:33
do you really need to build a DSP yes if
22:35
you're Amazon you have the scale to do
22:37
it but think of again CTV integration.
22:40
So there is a lot where I think you can
22:42
focus on the core and partner for the
22:44
rest. So that's why I'm saying the the I
22:46
would say the key is really designing a
22:49
a modular stack. So basically you can
22:52
you know plug and play you can swap
22:53
components as the market evolves because
22:55
again there is so much uncertainty that
22:58
you better have that flexibility and so
23:00
then when you select partners I think
23:02
the what you should look at is really
23:03
pragmatic stuff. It's like okay what job
23:06
are they best in the world at right do
23:08
they have I don't know proof of outcomes
23:10
can they integrate cleanly what are the
23:12
APIs and you know what are the
23:14
incentives you know we talked about the
23:16
the business model right before right so
23:18
how do they monetize how do we align
23:21
basically adoption to performance so
23:23
it's not just you're buying software but
23:25
how do you create a strong long-term
23:27
partnerships and then you have that
23:28
flexibility and maybe the last thought I
23:31
would add is it's also basically in my
23:33
mind this builder versus buy. It's
23:35
really like how do you take an ecosystem
23:37
approach, right? Because again, you need
23:39
to find the right vendors. So you need
23:42
to vet basically what partners really
23:44
fit the job to be done for your
23:45
strategy. But then you need to
23:47
understand, you know, what constraints
23:48
you have, what timeline are they
23:50
committed? How do you build
23:52
differentiation if you're partnering? So
23:54
those are a bit the type of questions.
23:56
But I think yes I like the idea of
23:58
partner for anything that is not core or
24:01
partner on the new bets and then you can
24:03
always in inhouse later in a way.
24:05
>> No 100% agree with you. I have been in
24:08
position of having to previously talk
24:11
companies out of trying to build their
24:13
own ad server or build capabilities that
24:16
already exist in the market and really
24:19
push on do you really need to
24:21
differentiate in this area? Um, but
24:25
staying on that and again another area I
24:27
hadn't necessarily intended to go as you
24:29
think about sequencing as a retail media
24:32
network grows over time. Often we see
24:35
they start out with a sales team and
24:39
maybe no programmatic. It's pretty small
24:42
operation and they're working with a
24:44
couple of advertisers and as they grow
24:46
they start to do some programmatic and
24:48
then they have to start to think about
24:50
offsite. What if if you were starting in
24:53
an uh retail media network from scratch,
24:56
what would be the top two or three
24:58
technology decisions that you would make
25:01
at the start and then what are the the
25:03
subsequent decisions that you think that
25:05
you have to make?
25:06
>> Yeah. So I mean definitely I would say
25:08
if we are a retail media network then
25:10
the point is you know you want to have
25:13
sponsor listings that work. So if you
25:15
think of the you know auction mechanics
25:18
etc etc. So I think the ad server would
25:20
definitely be the top one decision by
25:23
far that I would make. If you're more
25:25
like a a commerce media network then
25:27
you're more in the kind of video display
25:29
programmatic space and then I would
25:31
start thinking okay what are my what is
25:33
my programmatic tech stack cell side and
25:36
then from there I would also say what is
25:38
my my creative tech stack in a way
25:41
because I think now there is a lot of
25:42
convergence of media and creative so I
25:44
would say retail media definitely ad
25:46
server for the sponsor listic logics
25:49
etc. So quality score, bidding and and
25:52
so on. I think if you're a commerce
25:54
media, I would really think the
25:55
programmatic tech stack and the created
25:57
tech stack.
25:58
>> Brilliant. Uh and and while you're on
26:01
sort of tech stack, how do you think
26:03
about the integration of the data layer?
26:06
Because you've got sort of your business
26:08
reporting, you've got the data segments
26:09
that we've talked about. Is that sort of
26:12
something that you worry about a year or
26:14
two down the road once you've actually
26:16
got some products up and running or is
26:18
that like a first class thing that you
26:19
actually have to be thinking about on
26:21
day one?
26:22
>> Yeah, I think it's a it's a very good
26:24
point and the reason why I didn't
26:25
mention it as you know one of the
26:27
critical decisions at the beginning is
26:29
because my thought would be there how do
26:32
you make synergies and what I'm trying
26:34
to say is like uh if you think of all
26:36
these retail media networks, commerce
26:37
media networks etc. They're usually
26:39
launched as a business within a
26:41
business. But again, there is a larger
26:43
business and likely it's a tech business
26:46
and they already have a data strategy.
26:49
They already have made some decisions
26:51
from a a data standpoint. So what I'm
26:53
trying to say is like yes, it needs to
26:54
be a business within a business because
26:56
you need the agility, you need to be
26:58
able to be nimble, etc., etc. At the
27:00
same time, you need to make sure that
27:02
you leverage the the broader
27:03
capabilities. So I think data is a a
27:06
space where I think you know you can
27:08
definitely also understand try to
27:10
understand what already exists and then
27:12
you can basically start by leveraging
27:14
what is already there.
27:15
>> Brilliant. when you think about the
27:18
metrics that matter the uh my experience
27:23
in retail media network and it was kind
27:25
of eyeopening to me is you have to have
27:27
a different relationship as a publisher
27:29
to your advertisers and frankly you have
27:31
to care more about their performance. I
27:34
I don't know if I used quite these words
27:36
at the time, but some publishers I've
27:37
worked for, if an advertiser fell away,
27:40
you knew that there'd be 10 more,
27:42
hundred more in the queue sitting behind
27:44
them. The advertisers for a retail media
27:48
network are often the same people you're
27:50
working with in your core business and
27:52
they matter more and their performance
27:54
matters more. When you think about the
27:56
metrics that matter for retail media
27:59
networks, what are the top things that
28:00
come to mind?
28:02
Yeah. So I would say when you think
28:03
specifically retail media network so I
28:05
think ROAS is the you know the key
28:08
metric that gets discussed etc. So my
28:10
point is um also here I'm very
28:12
opinionated and I think RMN's overall I
28:15
think they've outgrown a bit their early
28:17
days right where they had like untapped
28:20
demand and so it means that every dollar
28:21
spent was both basically high rorowas
28:24
high incrementality so today focusing
28:27
only on rorowas incentivizes I think
28:29
behaviors that may appear efficient but
28:32
in reality they are kind of harming
28:34
long-term profitability and growth so
28:36
what I'm trying to say is like
28:38
roas today can be gained and and this is
28:41
problematic, right? Because basically
28:43
you can for example overspend on
28:45
retargeting or branded keywords and
28:48
basically these tactics they drive high
28:50
ROAS but at the same time they basically
28:52
focus on customers who are likely to
28:54
convert anyways and so they result in
28:56
low incremental growth or for example
28:58
you know you can drive discountdriven
29:01
sales. So you you can have discounting.
29:04
This will boost ROAS by generating
29:06
short-term revenue but lower margins as
29:08
well. Right? So basically what you're
29:10
doing you're attracting low LTV lifetime
29:13
value customers. So what I'm trying to
29:15
say is like there are ways to to to gain
29:18
ROAS and I think the key question is uh
29:20
you know is it good for the long term or
29:23
actually this behaviors lead to again
29:25
shrinking new customer cohorts. You you
29:28
increase the reliance on discounts and
29:30
reduce the margins. you low you lower
29:33
the the customer lifetime value. So I
29:35
think there are many negative effect. So
29:37
that's why long story short chasing
29:39
rorowas at all cost can actually lead to
29:41
slower growth and declining margin. And
29:44
so my point is efficiency metrics like
29:47
rorowas are necessary but also must be
29:50
balanced in a way with effectiveness
29:52
metrics that focuses really on long-term
29:55
outcomes. So if you think for example of
29:57
uh uh again we we mention lifetime
30:00
value. So can you measure the total
30:02
revenue contribution from the full price
30:05
customers basically they get acquired
30:06
over six months. Can you measure again
30:09
incremental so incremental revenues from
30:12
non-brand keywords basically. So you
30:14
track the revenues that are generated
30:16
from new demand sources. So that's I'm
30:18
saying again rorowass is an excellent
30:21
efficiency metric but then it's a bit of
30:23
a poor northstar. So it's I'm saying
30:26
striking the right balance between
30:28
efficiency and effectiveness which is
30:29
what you know brands always try to do
30:32
will basically ensure that the business
30:33
scales sustainably but also maintains
30:35
the margins. I I love how opinionated
30:38
you are. So I'm going to ask you another
30:40
one where you may have a strong opinion
30:43
given what you just said about the
30:45
importance of rorowaz to the key
30:46
customers. Most retailers have the the
30:50
partners that are absolutely critical to
30:52
them. Should they be prioritizing them
30:55
higher in sort of their ad serving stack
30:59
giving them access to unique inventory
31:02
on their site or do you think that at
31:04
the end of the day all partners should
31:07
be treated equally within a retail media
31:09
network?
31:10
>> Yeah. So I think there are two types of
31:14
partners, right? So we have the the
31:15
endemic partners which are also the the
31:17
the merchants and so these are people
31:20
that actually drive the the core
31:21
business in a way right and then you
31:24
have the the non-endemic partners they
31:26
are not merchants so then the question
31:28
is do they compete for the same
31:30
attention I would say yes definitely you
31:33
know but then u do you need to give more
31:35
importance to one or the other I think
31:37
ultimately reality is merchants at least
31:40
based on you know my experience and what
31:42
I've seen around merchants should always
31:45
be anyway the biggest part of your
31:47
revenues and I think when people go to
31:49
non-endemic that's more like a an add-on
31:51
so they can be very important again but
31:53
it's more like an add-on so if you think
31:55
of how you build your strategy for kind
31:57
of endemic versus non-endemic I think
31:59
the first thing is make sure that you
32:01
take care of your merchants because
32:03
ultimately they are both advertisers and
32:06
again they drive the core business so I
32:08
think definitely there is a more
32:10
attention there in a way
32:12
>> to talk to me about sort of the tension
32:15
between advertisers and merchants and
32:19
the sort of overall site experience of
32:21
selling product when you're setting up a
32:23
retail media network because clearly
32:26
there is a lot of margin associated with
32:29
the retail media network but there is a
32:31
balancing act. How have you seen that
32:33
balancing act work in practice?
32:36
>> Yeah. Yeah. So ultimately you know as a
32:38
as a a technical person you know I'm
32:41
sure you like all the concept of
32:43
portfolio optimization right this is in
32:45
my mind the perfect case of portfolio
32:47
optimization right so and I think there
32:49
are two tensions here so from one side
32:51
it's it's what we touched upon which is
32:54
basically merchants are also advertisers
32:57
and uh this is a delicate balance
32:59
because basically do you want to
33:01
maximize ad revenues and what if at the
33:03
same time basically again the merchant
33:05
is not performing things. What I mean is
33:08
you have late deliveries, you have the
33:10
the wrong offers, etc., etc. Are you
33:13
still willing to to you know get them to
33:16
advertise and win the next order if you
33:18
know they're you're making advertising
33:20
revenues but you're also hurting the
33:22
business in a way right so my point is
33:24
if you think the Google quality score
33:27
which basically you know says okay if
33:28
your quality score your page relevance
33:30
is lower you need to bid much higher so
33:33
how do you set up your auctions that's
33:34
why it goes back to the the the ad
33:37
server the capabilities etc how do you
33:39
set up the the auctions where it's not
33:41
just the highest bid wins. But in
33:43
commerce media, basically in retail
33:44
media, the platform should also kind of
33:46
score the value to the business and the
33:48
customer experience. So again, if a
33:50
merchant is unreliable, late
33:52
deliverance, late deliveries, bad
33:55
service, high returns. So do you really
33:57
want to amplify them because they bid
33:59
aggressively? Probably not, right? So
34:01
then you want to beat and optimize
34:03
basically against the broader value
34:05
function. So merchant quality, margin
34:08
impact, customer satisfaction, long-term
34:10
retention and so on and so forth. The on
34:13
the other side you have I think what you
34:15
were hinting at James which is you have
34:18
this duopoly right there are the the
34:20
merchants the endemic advertisers and
34:23
the non-endemic right and this is also
34:24
delicate because ultimately they they
34:27
compete for the same attention right
34:28
they compete for the same inventory
34:30
besides sponsor listing for if you think
34:32
on-site and offsite that's the same
34:34
right so the question then is are they
34:37
complimementaryary how do you architect
34:39
the marketplace in in a way that is well
34:42
balanced Right? So that then you have
34:43
the I don't know the the large brands
34:46
they still benefit from the the
34:48
contextual uh space the data etc. but
34:51
then the merchants are more focused on
34:53
on conversion. So I think that's a bit
34:55
the balance that need to be that need to
34:57
be built.
34:58
>> Thank you. Uh just one or two more
35:01
questions and I hadn't intended to ask
35:04
it but I have to ask it about AI and the
35:06
impact that we're seeing. Everybody
35:09
talks about where they think it's going
35:11
to have impact. I'd be interested, you
35:13
and I are talking here at the end of
35:15
2025.
35:17
Where are you actually seeing impacts?
35:19
Not about where people think it may be
35:21
impacting, but where it's actually being
35:24
used within commerce and retail media
35:26
networks. And if the answer is you're
35:28
not seeing it, that's there's signal
35:30
there for me as well. So, let me know.
35:33
>> Yeah. So I think there are also here
35:34
kind of you know the the lowhanging
35:36
fruits and and the the big vision and to
35:39
be honest many of the discussions that
35:41
I'm having they they they include of
35:43
course AI discussions and what I'm
35:45
trying to say is like lowhanging fruit.
35:47
I mean we just talked before a bit about
35:50
uh you know the creative piece. So again
35:52
there were a couple of PRs in the last
35:54
couple of weeks and they talk about
35:56
large retailers. Again, I will not
35:58
mention any names, not to promote or
36:00
advertise any company, but there were
36:02
other large retail media networks that
36:05
stroke partnerships with these Gen AI uh
36:09
you know uh production companies. Genai
36:13
creative company. So they can do like
36:14
production, they can do adaptation, they
36:17
can do personalization and you put a
36:19
prompt, you get an image or you put a
36:21
prompt, you turn an image into a video
36:23
and you can do personalization. So I
36:25
think it's fantastic, right? And so you
36:27
think you're a retail media network, you
36:28
have a long tail or small advertisers
36:30
that actually don't have that capability
36:33
in house or cannot afford paying
36:34
agencies. So I think uh very low hanging
36:38
fruit is oh gen AI for creative. So I
36:42
think that's easy then you can actually
36:44
push it much more. So if we go up you
36:46
know to the long-term vision if you look
36:48
at how many retail millionaires are
36:50
structured today you have a lot of
36:52
silos. So you have uh advertising,
36:56
you have uh offers, you have sponsor
36:58
listing, you have sponsored brands, you
37:00
have off-site. So what I'm trying to say
37:02
is like merchants specifically, which
37:05
again contribute to, you know, the the
37:07
largest portion of revenues. They don't
37:10
want to be traders, right? They want to
37:12
drive their business. And uh ideally, if
37:15
you have a capability to say, okay, give
37:17
me your budget. Give me the metrics
37:19
you're optimizing for. Do you want again
37:21
incremental rows? Do you want new
37:23
customers? Do you want you have a
37:26
budget? You have a KPI and then the
37:28
retail media network can optimize
37:30
everything in the background. So it's
37:32
really you need to have the AI to
37:34
understand how to bring everything
37:36
together. So it's ads plus offers, it's
37:38
onsite, it's offsite, it's the right
37:41
creative, etc., etc. So this is where AI
37:44
again the the the creative piece can be
37:47
done today. I think this uh this broader
37:50
vision where you bring everything
37:51
together and you really meet the
37:53
merchant where they are. I think this is
37:55
going to be the big disruption for for
37:58
ret media networks.
37:59
>> I I love it. It partly you're butdding
38:02
up against something we used to call
38:04
holistic yield management or full page
38:06
monetization. It's what everybody tries
38:09
to do and it's been incredibly difficult
38:11
up until now. I agree. It's probably one
38:14
of those areas that I think AI is primed
38:17
to be able to help. I haven't seen that
38:19
occur yet. I think that we will start to
38:21
see that in the next year or two. And I
38:23
I think you're right on that. Danilla,
38:25
this has been so amazing. Thank you for
38:27
taking the time. What is the best way
38:30
for people to follow you and follow up
38:32
on all the amazing things that you're
38:34
doing?
38:35
>> Um, very active on LinkedIn. So, that's
38:37
my favorite social network. So,
38:39
definitely have to happy to connect with
38:41
people on on LinkedIn.
38:43
Excellent. Well, again, thank you for
38:45
taking the time and I'm sure you and I
38:47
will uh spend some time together in the
38:48
future. Have a good rest of your day.
38:51
>> Thank you for having me, James.