Latest / eCommerce Podcast / Stop Using Generic AI In Your Ecommerce Business

Transcript

Stop Using Generic AI In Your Ecommerce Business · eCommerce Podcast

0:05
Well, hello and welcome to the eCommerce Podcast. My name is Matt Edmundson, and
0:09
it is great to be with you on this beautiful day wherever you are in the world.
0:13
Thank you so much for joining us. On this show, we talk about all things to do
0:17
with ecommerce, as you would expect given the name of the show. It's great to be
0:21
with you. I've been doing ecommerce for A long time. I call myself an ecommerce
0:25
dinosaur. I've been around ecommerce since 2002 was when I did my first ecom
0:29
website, uh, that we launched. Um, so I've been running ecommerce businesses
0:33
really since then and it's, it's amazing. I love it. I love ecommerce and I love
0:37
doing this show because I get to talk to experts and founders from all over the
0:41
world about how to run our ecommerce businesses better, learn from their
0:45
expertise, their stories, their insights. And today is no exception. So if
0:49
you're new to the show, make sure you like and subscribe and do all of that good
0:53
stuff. And of course, if you are a returning listener, which I know many of you,
0:57
in fact, the vast majority of you returning, let's be real, you're a legend.
1:01
Thanks for staying connected. Thanks for listening to the show. We've got a
1:05
great show coming up. No pressure to our guest because we've not had the
1:09
conversation yet, but we've definitely got a great show coming up. I'm really
1:13
looking forward to it since we did a pre-call actually. But yeah, thanks That's
1:17
it from me. Let's jump straight into it, shall we? Uh, Marissa, how are we
1:21
doing?
1:25
I'm great. Thanks for having me. Thanks.
1:28
Yeah, well, you're welcome. Now, first off, let's deal with the, uh, both the
1:32
location and the accent because they are, they're not the same, are they?
1:36
They're not the same. I am American and I am based in Athens, Greece.
1:42
Yeah, which is, is, uh, is, I, I think good on you. Why would you not want to go
1:46
live in Athens for a little bit?
1:52
You know, I was, I came to Greece on vacation by myself and I was working
1:56
remotely from here as one can do in ecommerce. And I just realised like, why am
2:00
I in New York City? Like really, why am I in New York City when I can be in
2:05
Greece and I can be in Athens and in the Greek islands? And that was in 2021 and
2:09
I never looked back.
2:14
Fantastic. Fantastic. And you're enjoying it, obviously, in Greece. What is it
2:18
about Greece that keeps you there?
2:22
Oh, I just love it. Well, first of all, I think there's something that I really
2:27
just love about being in this side of the world, waking up in the morning and
2:31
North America is sleeping, and I have like 8 or 9 hours of the day to myself
2:36
before, you know, everyone is blowing up my phone and my Slack. And that was
2:41
such a change from life in New York City where like my alarm would go off at
2:45
6:00 AM, I was at the gym by 6:45, rushing through my morning before like my
2:50
first, you know, 9:00 AM call. It was just a complete change of lifestyle. And
2:54
then I just think, you know, I'm from an island, I'm from Puerto Rico, born and
2:59
raised. I really gravitate towards islands. The Greek islands are just magical.
3:03
There's so many of them. There's so much to see that I really just feel like
3:08
it's a country that you could explore for the rest of your life and still not
3:12
fully understand all of it, see all of it. And I have zero connection to Greece,
3:17
by the way. Like, I speak about Greece as if, like, this is, like, a place that
3:21
I'm from. I have zero family, like, lineage in Greece. I knew no one in Greece.
3:26
Greece is not even a place that in Puerto Rico, like, growing up, it was not a
3:30
place that was on the radar, really. Puerto Ricans go to Spain, they go to
3:35
Italy. So, so yeah, it surprised me as much as it surprised everybody else in my
3:39
life when I You know, announced to everybody that I was moving here at 38 years
3:44
old. Yeah.
3:48
That's fantastic. I love it. I think that's brilliant. I, why not? Uh, my, I've
3:53
mentioned before on the show, my son, my youngest son, my middle child, he, he,
3:57
I think like a week or so ago moved to Bali and he's doing the digital nomad
4:02
thing, you know? And I'm like, well, good on you, bud. Why would you not? Right?
4:06
Part of me is like, that would be a nice thing to do.
4:10
Yeah. I never— the truth is, I never saw myself as the digital nomad type. Like,
4:15
full disclosure, I moved to Greece with like 14 luggages. I don't travel
4:20
lightly. I knew people that were nomads, like really, like, you know, 90 days or
4:25
3 months in one country, and then they would be in another country. And like,
4:30
that was not me. But looking back, if I could go back and talk to younger
4:35
Marissa, I would definitely tell her to to travel and work remotely a lot sooner
4:40
than I did it, right at the age of 38. Because knowing what I know now, I would
4:45
have had a lot more fun had I done it sooner. You know, anyways, everything
4:49
happens when it's meant to happen.
4:56
Well, yeah, I mean, you know, you're out there now, which is which is fantastic.
5:00
Now, for those of us who might not know about you, for those listening to the
5:04
show, just sort of give us the brief. Elevator pitch into all things Marissa.
5:08
Yeah. So, I started my career in wholesale and retail and working with some
5:13
ecommerce brands as early as 2006. I then decided in 2015 that I wanted to go
5:19
all in specifically to direct-to-consumer ecommerce. And then that's what I've
5:24
been doing ever since. I was launching and scaling brands as a fractional CMO
5:30
and COO. For you know over ten years, and then I started creating digital
5:35
products, courses, programmes to be able to reach more people and help more people
5:41
with e-commerce. And and now I do a little bit of everything from you know
5:46
mentorship to coaching to programmes to building out AI solutions for e-commerce.
5:52
I'm I'm doing a bit of everything, to be honest.
6:00
Yeah. What's floating your boat at the moment?
6:01
Yeah. I don't know if there's anything that's slowing my boat at the moment. I
6:07
think that I'm really— I think I'm— I don't think there's anything slowing it,
6:12
actually. I think that I'm in a moment in time right now where I'm—
6:18
Sorry, I said floating, not slowing.
6:19
Sorry, I said—
6:23
I think there's a bit of a delay, but I said, Uh, what floats your boat, not
6:26
what slows your boat.
6:29
I was like, that's so interesting. Is something slowing my boat? Nothing's
6:34
slowing my boat. That's why I was like, I was like, wow, is this like a good
6:39
question? Are we getting deep here? I was like, what is slowing my boat?
6:44
Um, I tell you what, I tell you what, Marissa, you are the quintessential
6:48
professional because you just carried on talking. You were like, what are you—
6:51
I was really trying to think about it. I was like, this is, this is a deep
6:56
question. What is slowing me down right now in life, you know? Business and
7:00
otherwise. Honestly, what's floating my boat is I have a very— I feel very
7:05
excited right now. Like, I haven't felt this excited since like 2015, when I
7:09
first went all into direct-to-consumer, because I feel that we're like, in a
7:14
very similar moment in time, right? Like, in 2014, 2015, I was really excited
7:18
because I was like, oh my god, This is it. As someone coming from wholesale and
7:23
retail, right? I was like, this is a moment in time where small businesses have
7:27
leverage that they never had before, right? In terms of distribution, you don't
7:32
need a retailer to give you permission to build a big business. And in a way, I
7:36
feel that the same thing's happening now, except the new leverage is
7:41
capabilities. I think that what AI, you know, what it promises and what it could
7:45
be is groundbreaking for small businesses. It's groundbreaking for your
7:50
solopreneur, your small teams. And like, that's who I get really excited about.
7:54
So, I think that that's, that's what, that's the energy that I'm in right now is
7:59
I feel like we're at another, like, pivotal moment where how business is
8:04
conducted online for, you know, the everyday person is once again shifting and
8:08
giving them more leverage and more optionality and more power and more access
8:13
than ever before. And I think that's like a very exciting, you know, thing to be
8:17
a part of.
8:22
Yeah, I mean, that makes a lot of sense, doesn't it? And I, I think I share your
8:26
excitement. Uh, I was talking to, um, a group of individuals this weekend,
8:31
actually, um, at an event in the south of France, which was a wonderful place to
8:36
speak to people. I have to be honest with you, I got to enjoy the weather. But
8:41
we're, um, we're chatting away about AI and the impact of AI, um, and what that
8:45
means And there's both on one hand, this sort of existential crisis that seems
8:50
to be going around, like, is AI going to kill us in 3 years versus the man
8:55
alive, if we could just harness the power of AI, the impact it could have on our
9:00
businesses is extraordinary, really. So I think you're right. I think it's a
9:04
really wonderful time to be alive and to have that technology. With the— let me
9:09
jump straight into this question. Then, Marissa, this is a question I do like to
9:14
ask people. Um, given what you've seen, what you know, what you experience with
9:18
AI and the excitement that you have about it, when you are talking to people,
9:23
mentoring people, et cetera, what's the biggest single mistake we're all making
9:28
where AI is concerned in your opinion right now?
9:33
Well, that one's easy for me to answer because I think that AI First of all,
9:37
when I talk about what I get excited about AI, I want an AI— what I'm building
9:42
right now and what I'm working on is taking my entire, all my systems, my
9:47
frameworks, my methodology, everything that I know and feeding it into an AI,
9:52
right? What I get scared about, and also, I've seen it in practice, is when
9:57
we're talking about generic AI or prompts or out-of-the-box chat or Claude, or
10:01
even when you see people that have AI businesses for AI, that are teaching
10:06
solutions that are industry agnostic, that they work for everybody. That I'm
10:11
just like, what are we doing? What do you mean? No. Like, no, there's things
10:16
that are very industry-specific. And having a generic AI working with you, I
10:21
think you're gonna end up in a bad place. I mean, I've seen it from, you know,
10:25
people putting together financial models and P&Ls with ChatGPT. And it's like
10:30
ChatGPT is making these crazy assumptions. It's not asking you any qualifying
10:35
questions. It doesn't understand why it needs to know the weight of an item
10:40
before it can, you know, make an assumption on what your shipping costs are
10:45
going to be. So, if you're a first-time founder, which most of the people in
10:49
ecommerce are first-time founders that have, you know, no domain or subject
10:54
matter expertise, potentially, you're going to be blindly following the advice
10:59
of an AI that really doesn't know what it's doing. And so, I think that's the
11:04
biggest issue that I see is that The power of AI is to harness it with that
11:09
expertise and embed that inside. And then, I think that it's a game changer for
11:13
small businesses, right? Because you can afford sort of like the brain of people
11:18
that really know what they're doing via an AI. You can access that judgment. You
11:23
can access that knowledge. That's what I think changes the game as opposed to a
11:28
generic prompt. Prompt. And some of the stuff that's being sold right now for
11:33
ecommerce is really bad. It's generic. It doesn't have any context to it. It's
11:37
like you can't give advice. You can't tell someone what to do if you're missing
11:42
context or you don't even know how to ask for it. So, I think that's the main
11:47
issue that I see with AI right now is that. And you're even seeing like the AI
11:52
that's being produced by some of the biggest SaaS companies in the world. It's
11:57
like, you know, they're technologists. They're not operators. And so, to an
12:01
operator, the minute that they start working with the AI, they're like, whoa,
12:06
this isn't right. Why is it saying this? This doesn't look good. This isn't the
12:11
way that I would do it. And so, that's fundamentally, I think, where the
12:16
opportunity is, right? To get an AI that is trained and built and designed by
12:21
people that actually have done that job, as opposed to not.
12:25
It's a really interesting point. I mean, I'm a big fan that AI needs a couple of
12:30
things to function well, right? It needs clear instructions, it needs context,
12:35
and it needs to know what good looks like. And so what you're talking about is
12:39
both the context of knowing what good looks like. And I was intrigued by the
12:44
fact that Alex Hormozi started advertising a few weeks ago. I don't know what
12:48
it's called, but there's like, you know, for however many thousand pounds a
12:53
year, I can, I can sort of connect to his AI system, which has got, you know,
12:57
the whole Alex Hormozi sort of system in the AI, all of his transcripts, all of
13:02
his meetings, all of his notes, all of his frameworks, his books, everything. is
13:06
all in that AI. And so when you talk to it, you're kind of talking to Alex.
13:11
You've got that context. And people seem to be doing rather well with it. And so
13:15
you're in— I think what you're saying is really interesting because the frontier
13:20
models have got so good. The problem is it's this middle bit that you're talking
13:24
about. It's like giving it all the context that people need to make the right
13:29
calls, the right decisions. So how do you— If I was a founder, how do I get
13:34
around that? How do I deal with that?
13:38
Well, I think you have to be really, you know, I think in a way, it's the same
13:42
advice that I've always given to people. You surround yourself with someone
13:46
that's actually done what you want to do or been where you want to be. And so,
13:51
someone like an Alex Hormozi is super smart and has data at this point now of,
13:55
like, thousands upon thousands of businesses that have willingly just given him
13:59
the data. But he's not an expert in ecommerce. He doesn't do ecommerce, right?
14:03
And so, you're gonna get, I'm sure, good advice on some level, but not the best
14:08
advice you could for ecommerce. It's like I always say to people, I'm like, if
14:12
you want to be a good tennis player, you're not going to go to a basketball
14:16
coach. You got— you're going to go to a tennis coach who's going to help you
14:20
with your tennis game, right? It's like, but it's the same sort of like
14:24
principle. I think the first thing is that. I think the second thing is, you
14:29
know, whoever you're getting AI solutions from. Again, try to do a little bit of
14:33
research to make sure that they are tailored for your industry as smart as they
14:37
are, right? Email marketing if you are a law firm is not the same thing as email
14:41
marketing if you are an e-com business. You know what I mean? Like, yes, they
14:46
both might have email marketing but it's just going to look different and it
14:50
needs to do different things. So, that's what I would— that's my biggest thing.
14:54
And it takes time to build into AI and train the AI and give it decades' worth
14:58
of experience. It's also when you start doing this, which I've been doing this,
15:02
which is why I think about it so much. It's amazing how much of your experience
15:07
maybe you've never thought about because they're exceptions to rules. They're
15:11
the moments when you actually go against what you would normally do because the
15:15
context makes sense or because you feel like you should be heading in a
15:19
different direction. And when you really start to think about all the different
15:24
layers of training, judgment, and context, and experience into the AI, you
15:28
realise it's a big job, which is why, you know, most of what's out there is
15:32
generic and, you know, oversimplified to what it needs to be in order to really
15:36
be helpful for a business.
15:40
Yeah, I mean, I've done a similar thing to you. We've taken all our frameworks
15:45
and our understanding, and we've thrown them into— I throw them into it, but
15:50
maybe not the right technical term, but we've put them into AI. Um, and so when
15:54
the team here question, um, we use Claude, but when a team questions Claude,
15:59
we've got this layer called SAM, um, which is all our frameworks, it's all our
16:03
understanding, it's that whole context, and it's always perpetually learning as
16:08
we, as we're going along. And we found that as we've done that, the output has
16:12
got much better. I guess I'm just, I'm just kind of, I'm aware that Like you
16:17
said, it wasn't pretty, it wasn't as straightforward as I thought it was gonna
16:21
be just to do a brain dump from my head into AI. That took a fair bit of time. I
16:26
don't know if you found that.
16:31
Yeah, because even, you know, in my case, I began with already mapped out
16:35
curriculum because I had digital products and, you know, info products. So I was
16:40
like, how difficult could this be? Right. But let me give you an example. I took
16:44
an email marketing, I had an email marketing module that was originally like 6
16:49
videos and 45 minutes. And I was like, okay, if I really need to teach AI
16:53
everything about email marketing, let me break this up. And it went from 6
16:58
videos to 50 videos because there was so much more than I anticipated that I
17:02
actually had to sit down and give it the context for understanding. Like, you
17:07
know, when you do this versus when you do that. When does a business need this
17:11
versus when does a business need that? You know why? The answer isn't always
17:16
just black and white and the same for everybody. You know, because you have
17:20
businesses with different AOVs that might have just different customer paths to
17:25
purchase. Just to, you know, what are all the ways that I drive revenue via
17:29
email? Well, this is what I would do for one business. But this is something
17:33
else I would do for another business. And this is something. So, it definitely
17:38
took longer than I thought. And then, you have the whole layer of like, that's
17:42
the first thing of the AI knowing and being able to give advice and being able
17:47
to To tell you what to do and maybe even quantify, like, the revenue loss for
17:51
you. But then, there's the other layer of the AI actually doing part of the work
17:56
for you, you know, filling out the reports, pulling the information that
18:00
historically we've needed to pull from multiple sources to understand. And then,
18:05
obviously, you know, actually creating output. So, yeah. But it's been very fun.
18:09
And it's been, I think, really— I keep getting very excited, even when, like,
18:14
it's like, oh, we pulled back a layer and we're like, oh God, we have to go
18:18
further. Or why is it doing this? It's going rogue. And like, this is nowhere in
18:23
any of the information it got. So, it just blatantly made this up. I'm still
18:27
very excited by what it can do for small businesses because I have been there.
18:32
When you don't have a budget, when you don't know how you're going to get
18:36
something done, it's just, it changes the game for your small to mid-sized
18:41
businesses. It just, you know, but yes, it was, it's been more than I thought it
18:45
would be.
18:50
I, I, I thought it was quite cathartic actually trying to, when we sat down with
18:54
AI and we're distilling our frameworks like you, I, like you, I've, I, I kind of
18:59
understood what our frameworks were because we taught them to others. But
19:03
there's, like you say, there's a lot of nuance where when you're chatting to AI
19:07
and you're like, let me tell you what I think. I think it's almost like therapy
19:12
for the e-com founder, right? Because you're like, ah, so this is why I think,
19:16
and when AI says to you, well, hang on a minute, why do you think that way? And
19:20
you're like, well, this is why I think that way because of this story. And you
19:25
kind of explain it. It's one, I think it's empowering to relive your past
19:29
victories. If that, because sometimes I think we've forgotten things. 2, it's
19:33
good to understand how much you actually do know. But 3, it highlights what you
19:38
don't know. Yeah. And, and shining a light on what you don't know, which is what
19:42
we did. And again, I've mentioned this before on the show, and I don't want to
19:47
say the same story more than, you know, a few times, but probably the most
19:51
powerful thing that we've done recently was understanding that our frameworks
19:55
around mobile optimisation weren't up to scratch. And so we had Adam Pearce on
20:00
the show, talked to him, used the show as a foundation Did a whole bunch more
20:04
research, threw that in there, and we created a playbook which became something
20:08
that we could feed into the AI model. The AI model then enabled us to create
20:13
something called Moby, which analyses the mobile site. Well, on our own personal
20:17
site that we did that on, our mobile conversion went up 400%. Right now, that's
20:21
a lot.
20:26
Yeah.
20:27
When you're turning over millions on a website. That's a significant amount of
20:31
sales increase. And so it's just the power of this is if you get the right
20:36
information to AI, it can really help you. So know what you know and put that
20:41
into AI. Know what you don't know, um, and let AI help you. Um, I think it's a,
20:46
it's a, my, my story is, it's extraordinary. Um, and I, I, I'm enjoying the fact
20:51
that you're finding similar things, Marissa.
20:55
Yeah, I am. I also think that there's, like, a human training that's also
21:00
required on how to work with AI efficiently, you know? Yeah. And I found how
21:04
many people are, like, afraid to push back or question. How many people will
21:08
take what AI does at face value and just be like, okay, yep, this is good, and
21:13
not actually look through it. You know, it's kind of the way that you would work
21:17
with a human, right? You wouldn't just accept whatever a human would tell you.
21:21
You would fact-check them, or you would, you know, probe, or you would look to
21:26
go deeper. And so, I think that that part of it will be really interesting to
21:30
see how that develops over time, and people get, I think, more confidence in
21:34
working with AI, because I think that's also a really big part of it. The
21:39
success that you're going to have with the output is also driven by how you're
21:43
able to question and drive the process with the AI as well.
21:47
Yeah. Yeah, again, I would wholeheartedly agree. I think where I'm at with AI
21:52
and where the majority of the team are at with AI are 2 different things, right?
21:56
Because I spend so much of my life in here trying to figure it out. Um, so yeah,
22:00
I, I think part of my job is to get them to use AI better. I mean, we've spent a
22:05
lot of time, money, and energy trying to get the AI working well enough so that
22:09
somebody who doesn't know how to use it necessarily can come in and hit the
22:13
ground running with it. And that's been super powerful. I think we had a guest
22:18
on the show recently talking about this, about how AI should be treated as a
22:22
staff member. And you don't have somebody come into your building and you sit
22:26
them down in front of a desk and they do some work and you go, thanks for that.
22:31
And you just accept everything on face value. You train the new member of staff,
22:35
you give them the playbooks, you double-check their work, you want to onboard
22:39
them well, you spend time with them. And I think you're right with AI, you've
22:44
got a similar thing. You've got to do that. pushback. How do you— I guess people
22:48
listening to the show will be listening to that and going, great, this is all
22:53
great. And I, I get that we need to, you know, get this knowledge and, and so on
22:57
and so forth. And we can research that. How do you push back? How do you get the
23:01
best out of AI? What have you found that works?
23:06
Well, I mean, assuming that you have an AI that has the content in it and has
23:10
the knowledge base in it, then, you know, I, I think that a lot of it is the
23:15
dialogue with the AI. I think a lot of it is, you know, letting— kind of the way
23:20
that you were saying earlier, you know, letting the AI understand what you're
23:25
trying to do, what you want to achieve, what the goal is, even with the content,
23:29
even with the knowledge base. I think that's really important. I also think that
23:34
who's using the AI is also important. Like, if you are a subject matter expert
23:39
that's using AI to, you know, increase your output or your capabilities, That's
23:44
really different than if you are a small business and you are using AI to help
23:48
your VA learn how to do something. You know, because maybe you don't want to
23:53
hire out that role. You've trained the AI and now you have a VA that you are,
23:58
you know, gonna put on that task. That's also different. So, I think it also
24:03
depends like who is using it and for what purpose are they using it. Across the
24:08
board, I think it's the same thing, though, that I would say to you before. I
24:12
think that you have to be clear on what you're trying to do. You have to push
24:17
back. You have to not accept the first thing that you get back. The more you can
24:22
give it directions or examples, the better. You know, but I think that you
24:27
always have to understand, like, for your team, who's using it, and then what
24:31
kind of additional coaching do they need in order to get the best out of it?
24:36
Because if they're not a subject matter expert, but they're using AI to
24:41
facilitate facilitate a job or a role for you, then I would be paying attention
24:46
to how they're working with the AI to try to give them coaching on the best way
24:51
to increase their confidence when they're doing something that they're not an
24:55
expert in. Does that make sense? That's like a whole other layer.
25:00
Yeah, no, absolutely. I'm thinking of someone that I know right now. I think
25:05
their overreliance on AI to do their job has been It's been quite fascinating to
25:10
see because the default assumption is AI knows more than them because it's AI.
25:15
And actually just having the confidence just to step back and push back. It's
25:20
not like AI's got a self-esteem problem. It's okay to push back. It's okay to
25:25
ask. One of the things that we've done, and I don't know if you've done
25:30
something similar, Marissa, I'd love to hear your thoughts on this if you have.
25:35
We set up, uh, an AI board of directors. Okay. This was, um, a thing of maybe
25:40
about, I guess, about 12 months ago. But it's been such a thing for us that it's
25:45
a practice that we've kept whereby you tell AI to, in effect, you have different
25:50
agents who represent different people. So for us on our AI board, we of course
25:55
have Jeff Bezos, right? Because why would you not? We have Steve Jobs, we have
26:00
Warren Buffett, um, and a few other people, you know, who make up this AI board.
26:05
And what's fascinating is when you present something to them, the pushback you
26:10
get is, is quite fun. And actually, when the agents start pushing against each
26:15
other, when Steve Jobs starts going to Jeff Bezos, no, that's not how I would do
26:20
it. And you're watching this conversation unfold. I mean, apart from just being
26:25
quite good fun and obviously very made up, um, you can start to see reasoning
26:30
and questions appear. And having that sort of multiple personalities, uh, in an
26:35
AI board as agents— I am very aware it's not Steve Jobs, I'm very aware it's not
26:40
Jeff Bezos— it— but as a thinking exercise for me, I find it's quite fascinating
26:44
because you get all these different opinions on something. All these different
26:49
questions, and it causes you to think and have to make a decision yourself,
26:54
which I think is really, really helpful and has been for me. I don't know if
26:59
you've done something similar.
27:04
I haven't done something similar, but, like, what a fun exercise to do. And the
27:09
implications of it are obviously greater than business, right? Like, you could—
27:15
life, you could literally create anyone that has, like, a profile where there's
27:20
enough information on the internet on them and on what they've said, what
27:25
they've read, what they've done, speeches they've given. I love that. No, I've
27:30
never done that. I've never thought about doing that. But I think it would be a
27:35
lot of fun to do that on a personal level.
27:40
Well, to be fair, I have different boards for different areas of life. So, we've
27:44
got the AI board for our e-com business. And then I've got an AI board, not
27:49
necessarily for life, but for some of these things over here, you know, these
27:53
different projects that I'm involved with. And then you get different people
27:57
involved. Like if you need a storyteller, go get Malcolm Gladwell, you know? Um,
28:02
and it's, again, you have to take it with a pinch of salt in many ways, but just
28:06
enjoy it and have some fun with it. I remember the first time I sort of, I put
28:11
the board together. It doesn't take long. It's just, you know, simple stuff to
28:15
do. Um, we put the board together and I must have spent 6 hours just locked in a
28:20
room, just going backwards and forwards, asking questions, answering questions,
28:24
thinking about the company in ways that I'd never done before, you know, because
28:28
it was the first meeting and they were grilling me about the business. And I
28:33
think in the prompt I said, grill me, don't be brutal, you know, don't— I don't
28:37
be timid, don't be shy, don't try and please me, just We want to get the best
28:42
outcome. And so, I found it both quite difficult, but also quite liberating all
28:46
at the same time.
28:50
Yeah. Yeah. No, I love that. That's a nice one. I'm going to do this exercise.
28:55
It's a nice one. I'm going to do it. I don't know that I would want all of the
29:00
founders that I've worked with over the last 20 years to do it, because quite my
29:05
experience with most founders is that there's too many opinions. They have a
29:10
hard time focusing, right? Like, they go to dinner with an investor who has a
29:14
friend who has a cousin who did this. And, you know, a lot of the time, my job
29:19
was to sort of like get them to have blinders on and just focus on what needs to
29:24
get done. So, I don't know that I would want— I would give them that exercise,
29:29
but I'm dying to take on that exercise. I think it would be a great one for me.
29:33
How would you— going back to your previous point, too much decision, too much
29:37
choice. How could you use, I suppose, AI to help you then as a founder there?
29:41
How would you distill choice down to what's important? Because you're right,
29:45
there are so many options. There's so many people telling you what to do. How do
29:49
you distill it down to what's important specifically for you? That's a really
29:53
big question. I think that's almost bigger than the question than what slows
29:57
your boat, Marissa, but have a go.
30:00
Um, you know, I've always said that my opinion about ecommerce and, you know,
30:05
probably life in general is that you don't have to be, you don't have to be good
30:10
at a lot of things. You have to be great at a few things. And so I have, you
30:15
know, talked extensively about shiny object syndrome and like how difficult it
30:20
is in ecommerce. It's just, you know, an echo chamber of like every day
30:25
someone's like, this is the new thing, the hot thing. And I'm like, let's not
30:30
forget that there's only like 3 actions a person can take on the internet that
30:35
matter to an ecommerce business: click, subscribe, and buy. That is it. So, you
30:40
know, we got to drive the traffic. How are we going to consistently drive the
30:45
traffic so that it clicks? Right. And so I would always be like, on certain
30:50
levels, you got to be great at, you know, paid ads, you got to be great at email
30:55
marketing, you got to be great at the content that fuels both of them. And then
31:00
obviously, like, you have to have the ecosystem, the ops to support it, you
31:05
know, from inventory to your reporting and analytics to, you know, your unit
31:10
economics. So, I'm more on the camp of less is more than— and I'm more on
31:15
there's a lot of things that you could do. But what are like the key things that
31:20
you have to do in order to really grow this business? So, that's always been my
31:25
perspective. My perspective has always been like you're either going to win at
31:29
something that's consistent, repeatable, scalable, Or you're going to be in a
31:34
business that is, you know, having peaks and valleys. Uh, uh, and I've never
31:39
wanted to be in one of those businesses that have peaks and valleys. So I've
31:44
always opted for the ones that I can, can scale predictably their growth.
31:49
Yeah. Yeah. I love that. Click subscribe and buy, um, become good at a few
31:55
things in, in those areas. And I think that's quite empowering, isn't it?
32:00
Because it enables you to say no to a lot of things and not worry about it. Um,
32:07
it's not easy, and it's hard when you have a content engine, right? 24/7. When,
32:11
you know, 24/7, whether they're on LinkedIn or whether, you know, you're, you're
32:16
hearing on Instagram, you're seeing things people are telling you, like, this is
32:21
how you build your business. No, this is how you do it. This is how you do it. I
32:25
think it's harder than ever to just focus on what I call, like, the very, like,
32:30
mundane you know, tasks that are required in ecommerce when you look around and
32:35
everyone else is doing these like fun things. It's really hard to narrow it
32:40
down. You know, you have this sense of like, am I missing out? You know, should
32:44
I be building a personal brand? Is my personal brand what's gonna get people to
32:49
buy my product online? Like, you know, there's just all sorts of noise.
32:55
Yeah, that's so true. That's so true. And it's— I think your first 5 years of
32:59
business are learning actually what is noise and what's relevant for you, isn't
33:03
it? I, I— you— I think you, you get clearer as you spend longer in it. But at
33:07
the same time, you're right, the fear of missing out is a really powerful
33:11
psychological thing, isn't it, that wants to draw you back into these things?
33:15
It's a fear of missing out. It's also the way that the content machine is, um,
33:20
The way that it works, right? People make claims without context attached to it.
33:25
So, you know, if I had a dollar for every time a founder came to me and they
33:30
were like, so-and-so said they're doing X. And I'm like, okay, but is so-and-so
33:35
in your same business model? Are they 100% direct-to-consumer like you are? Or
33:40
are they also doing wholesale? Or are they also selling on other? You know, like
33:45
there's content, in my opinion. Like oftentimes you're you're lacking context,
33:50
so it looks like a shiny thing, and you're like, "I want to be doing that. It's
33:56
working so well for them." And it's like, "Yeah, but that's not your business
34:01
model. That's their business model. If you want to be in their business model,
34:06
you can. There's trade-offs on both sides, right?" Yeah.
34:11
So true.
34:12
Yeah.
34:12
It is so true. I I'm I'm. We could talk about this all day, Marissa. I think we
34:16
could. But let me ask you, where should founders— what's the sort of checklist,
34:21
maybe is a better way to phrase the question, that founders could have of areas
34:25
where you think, right, if you are a founder, if you're new to AI, or you've
34:29
been sort of on the outskirts, you're not like a full-blown expert and your
34:33
whole system's already set up, but if you're sort of thinking about it and
34:38
playing about it, where would you start? What are the quick wins? What are the
34:42
areas that you see having the biggest impact? Like start here, then go here kind
34:46
of a thing. I don't know if you— I appreciate that's, you know, what kind of
34:50
business you have, but in general, where would you start and where would you go
34:54
next?
34:59
Are you asking for ecommerce in general, or is this for AI and ecommerce?
35:06
Oh, well, let's go with AI and ecommerce. And if there's difference between that
35:09
and ecommerce in general, I'd love to talk about that as well. But start with AI
35:13
and ecommerce.
35:17
Well, I think they're probably very similar. I think that the first thing that
35:22
everybody needs to get straight, and a lot of people don't have straight, is,
35:27
you know, the numbers. The amount of ecommerce businesses that don't have, you
35:31
know, a real plan in terms of what they're trying to do is staggering to me.
35:36
That's where I begin. We have to first have the numbers. We have to make sure
35:40
the business is capitalised properly. We have to know what success looks like.
35:45
And I think that also what what the founder views success at, right? I'm going
35:50
to give a founder that wants an exit a very different business strategy than a
35:54
founder that wants a lifestyle business that's going to pay them a certain
35:59
amount, you know, every single year. So, I think to me, like, figuring out
36:03
success, figuring out the financial model, the P&L, what it's going to actually
36:08
cost you. And I think that now, with AI, what's beautiful, right, is that we can
36:13
deploy resources. So, potentially there's savings with AI. If you have a good AI
36:17
and you've trained it and it knows what it's doing, you can save on some of
36:22
these line items. And you can redeploy that into either, you know, more paid
36:27
ads, more inventory, more money for you at the end of the year, whatever you
36:31
want. But that's where I would actually first start. Then, you know, for me, I
36:36
think one of the most— the hardest areas for ecommerce businesses to often do
36:40
really well, especially your small and mid-sized businesses, is actually email
36:45
marketing, where I think AI can do a huge benefit, you know, something that in
36:50
the past would cost you $3,000 to $10,000 a month for an agency or multiple, you
36:54
know, freelancers to support with. I think email marketing is an area that AI
36:59
can do a lot of really, like, fantastic work on. You know, and then I would move
37:04
through the other marketing areas. But I would start with, I would start with
37:08
the numbers. And, You know, like anything else, I think my advice is always
37:13
going to be not to use a generic AI, out-of-the-box AI, or an industry-agnostic
37:17
one, to really find one that has been, you know, trained by operators that
37:22
actually has knowledge and context and all that kind of stuff. But that's
37:27
probably the 2 big areas where I think it's like the most helpful immediately
37:31
are in the numbers side and really like wrapping your arms around that. The P&L
37:36
and the financial model. And then actually, like in email marketing, I think
37:40
it's a huge win for small businesses. Awesome. And then obviously, like in all
37:45
the other areas, I'm a little more apprehensive about AI when people are talking
37:50
about it, like replacing, you know, you're not going to need a media buyer. It's
37:54
just a button. Like, I'm still not there actually with something that's as
37:59
important as driving traffic to your site, you know, at a price that makes
38:04
sense. Um,
38:09
yeah,
38:09
if anything, it's maybe more distrusting of media buyers because I'm like, are
38:13
you just put— like, I need to actually see what's happening. And I'm probably
38:16
like a lot more on top of that role than ever before. Um, did I answer the
38:20
question or did I just go off on like a tangent? That, that's where I'd start
38:24
though. I would start with the numbers.
38:28
Yeah, that was a good, good place to start. I, I mean, I, I, I, I want to go
38:32
down that media buying thing, but we won't. I, I want to stay on track a little
38:37
bit. The, um, the numbers then. So coming back to the And sorry, just to come
38:42
back to the one point you did mention, which I thought was really important, and
38:46
I just want to emphasise, is this idea of AI replacing humans. Now, I get that
38:51
AI can, you know, when it comes to graphic design, for example, AI can do the
38:56
job of a pretty competent graphic designer. And like you were saying, you know,
39:01
with the media buying, AI could now do the role of a media buyer. Is it going to
39:05
do a great job? I think like you, I think it's dangerous to assume you still
39:10
don't need a person in that chain. The same with, I mean, we've talked about it
39:15
around content for months, haven't we? That how do you avoid the AI slop? Well,
39:19
you need somebody to check it before it goes out there. You need somebody to
39:24
read it and go, yes, I'm happy with that, or no, this needs to change. And I
39:29
think the same is true with everything that AI does. So I would agree. I think
39:33
probably the best named AI out there is Microsoft's Copilot because I see AI as
39:38
a copilot. Probably the worst AI out there, in my opinion, is Microsoft's AI
39:43
Copilot, but that's another story. I just think it's a mindset you have to get
39:47
into. Let's go back to the finances quickly if I can. What are maybe 2 or 3
39:52
things that you see there that founders aren't tracking, but could easily do
39:57
something with where AI is concerned? What are maybe 2 or 3 metrics that you
40:01
really like?
40:06
So, the question is, where are founders maybe not utilising AI to the level that
40:11
they should utilise it to? Well, again, I think that I come from a different— I
40:17
have, you know, 15 years of financial models that were built for me by third
40:22
parties, you know, everywhere from, you know, people in Boston Consulting Group
40:27
to a guy in India, like every type of financial model that you can imagine, you
40:32
know, from a subscription business to a subscription and regular ecommerce
40:37
business. And so now, for me, it's really easy. It's wonderful. I can plug all—
40:43
I've plugged all of that in. And I have, you know, a financial modeler who is
40:48
better than anyone that I've worked with. But it also has, like, a list of
40:53
questions where it understands it cannot proceed without getting those questions
40:58
answered, right? And in some cases, it also knows the questions where it can
41:03
find industry information online. You know, if you have a business that, you
41:09
know, wants to figure out what to assume for a return rate or what to assume for
41:14
a discount rate based on their product category, it knows where it can go, like,
41:19
you know, look out and where it needs information from the founder to actually
41:24
come back and say, hey, you know, back to the shipping analysis, have you
41:29
weighed this product? What is the average weight of it in the box? Do you know?
41:35
Do you know? You know, so I think that if a founder has this sort of template of
41:40
a P&L of every single line item, I think going back and forth with a trained AI
41:45
can be incredibly helpful in terms of understanding costs, in terms of knowing,
41:50
even for a human, right? What's the difference between hiring a graphic designer
41:55
that does direct-to-consumer that's based in the US versus hiring one that's
42:01
based in the Philippines versus finding talent that's based in Brazil? Like,
42:06
doing that kind of competitive analysis is so much easier, so much faster now.
42:11
And that can make a big difference when you start to, you know, you're really
42:16
massaging those line items to get to what your fiscal plan is going to be and
42:22
your people plan is going to be. I think it's wonderful to use it like that. Or
42:27
to also validate if you're in contract negotiations with a 3PL, putting that
42:32
contract into the AI and asking its opinion or asking for it to maybe do a
42:37
little bit of a comp landscape for you on other 3PLs that might be better
42:42
options for you. I don't know if I answered your question. But I think that
42:48
there's a lot of— in some cases, the AI is really, really great, any AI, at
42:53
giving you information it finds on the internet that makes sense for you. In
42:58
other cases, you need an AI that knows to ask you first before it makes
43:03
assumptions. Like this founder sent me a P&L that ChatGPT created for it. And
43:08
ChatGPT just made up the paid advertising line. ChatGPT just decided like 20% is
43:14
what you need for paid advertising. And I'm like, on what basis? You've never
43:19
launched. You're pre-revenue. Where did ChatGPT get that 20% was the right
43:24
number? You know what I mean? So, but I think it's incredibly helpful for
43:29
everybody. It's just, you just need to make sure that you're You're giving it
43:34
the right information. You know what you can ask it to do, and you know what it
43:40
needs guardrails before it like moves forward. So I don't know if I answered
43:45
your question, actually.
43:52
No, that's great. That's really. I think it was a great answer, and I I I I I
43:56
love it. I think it's such a powerful thing. You know, I was I'm I'm sort of
44:00
saying. Yeah, we've probably made that mistake. Yeah, we've probably made that
44:05
one. I, I, we got a contract, uh, last week, um, and I, I had to read through it
44:09
3 times because the numbers weren't making sense. And so just doing what you're
44:14
saying, just rather than just accepting ChatGPT's answer, we were like, hang on
44:18
a minute, I need you to show me the maths. I, I need you to walk me through the
44:22
maths like I'm a 10-year-old because I'm just not getting it right now. And it
44:27
was very, very helpful. In fact, one of my favourite things to say to AI is just
44:31
assume I'm thick, I'm stupid, I'm like a 10-year-old in— do you know what I
44:35
mean? I just explain it to me like I'm like that, and it usually does a pretty
44:40
good job of helping me get to where I need to be. Marissa, listen, thank you so
44:44
much for coming on the show. I'm aware of time, and it's flown by at 1,000 miles
44:49
an hour, as it always does. If people want to reach out to you, if they want to
44:53
connect with you, what's the best way
44:58
They should find me on LinkedIn for sure, Marissa Rodriguez, because I spend
45:04
quite a lot of time on LinkedIn. They can also find me at throughexperience.com.
45:09
They can also find me on Instagram, Marissa_ThroughExperience.
45:15
Fantastic. We will, of course, put all of those links in the show notes. So, if
45:18
you're listening on a podcast player, just scroll down through the show notes,
45:22
you'll find them. If you're on YouTube, just go to the description, the links
45:25
will be there. Of course, all the links will be on the website at
45:29
ecommercepodcast.net. Just find this episode's page on the website and you'll be
45:33
able to click through and connect with Marissa to your heart's content. But do
45:36
connect with her, connect with her on LinkedIn, go and say hello. I'm sure she
45:40
would appreciate it. It's always nice to connect with new people. But Marissa,
45:43
thank you for coming on the show. I like to do this thing called saving the best
45:47
till last, as I mentioned. Just in closing, for anyone that's sort of still
45:50
listening, Out of everything that we've talked about, what's the number one
45:54
thing you want founders to know where AI is concerned? I think I know the answer
45:57
already, but let's just bring it home. What is the key takeaway from your point
46:01
of view where you're just like, do this and it's gonna make such a massive
46:05
difference?
46:08
Yes, please, please do not use AI that is industry agnostic, that is designed
46:12
for any industry. You're not gonna get the greatest results. Please do not just
46:17
use, you know, ChatGPT or Claude out of the box. I mean, have you actually
46:21
looked at Reddit? There's a lot of misinformation on the internet. And you do
46:25
not want to be on the receiving side of that. Please do find AI solutions that
46:30
have been created and designed by operators that have done the actual job.
46:34
You're going to get there faster. It's going to have a better result for you and
46:38
for your business.
46:44
Fantastic. Fantastic. Marissa, thank you. I loved it. I feel like I could talk
46:48
another hour without too much trouble, but thank you for coming to us via Athens
46:52
and New York. It's been amazing.
46:55
Thank you so much for having me. And I've enjoyed it as well. I could definitely
46:58
talk to you for at least another hour.
47:01
Maybe we'll do part 2. We'll do part 2 soon. But thank you. Well, that's it this
47:05
week, ladies and gentlemen, from the eCommerce Podcast. Thank you so much for
47:09
joining us. As I mentioned, Everything you need to know is on the website at
47:13
ecommercepodcast.net. The transcript from today, the notes from today, the links
47:17
from today, a really detailed blog post. There's even prompts and stuff you can
47:21
put into AI from today's show if you also give it the transcript. There's all
47:25
kinds of cool stuff there. So just go ahead and check that out. But thank you so
47:29
much for joining us. I will see you next week. But from Marissa and myself, that's
47:33
it. Thank you so much. Bye for now. Bye.