Search
FRANK•LY
ecommerce
Meetups
Workshops
Archive
Subscribe
FRANKLY ECOMMERCE
Oct 1, 2026
Massive Creative Variation Powers BFCM Acquisition
Massive Creative Variation Powers BFCM Acquisition
00:00
54:34
1×
Transcript
0:00
There's all kinds of AI tools this year promising results, trying to find ways to make it generative and scale.
0:06
But how does it do it in a way that locks into your brand identity and messaging, stays true to what you're all about? You've gotta feed the machine. How do we row together? We're in the boat with Meta.
0:19
How do we row with Meta, not against Meta? And if we don't give them enough creative, it's like, "Hey, you got one oar. Stop s- stop rowing in circles." Be two to f- five K a month.
0:31
Like, what quality talent do you think is gonna be working on that account? And that's, yeah, and that's the biggest issue is, like, I, I...
0:38
Any, for any agency that are servicing that tier are just not ones you can really trust, because they're just trying to give you a very basic setup.
0:46
You're listening to the Frankly Ecommerce Show, candid conversations about business with host Patrick Scott Pittman. Let's talk about ads.
0:56
Black Friday beckons, the day's getting closer, and there's no shortage of ideas for how to make your creative better this year than last year.
1:05
But there's all kinds of AI tools this year promising results, trying to find ways to make it generative and scale.
1:12
But how does it do it in a way that locks into your brand identity and messaging, stays true to what you're all about? Well, there's a, a hope for a solution in how we can make sense of whether it's gonna work or not.
1:25
I've brought in someone who can talk about how and why, someone who's been doing this for their own brands and now has developed some tools to make it easier and more accessible to you.
1:37
All the way from Phoenix, now recently relocated to Austin, Texas, Rob Green, welcome to the Frankly Ecommerce Show. Thanks, Patrick. Appreciate the, uh, invitation, and the welcome.
1:47
I'm excited to talk about ads and creative. You're probably gonna have to rein me in, buddy. Yeah. [laughs] Well, I know you've been at this a while.
1:54
Maybe you started with the Google Ads thing, trying to do dropshipping products. Eventually, you found your way to Amazon. I think, you know, you told me you've sold just about everything except beauty.
2:04
So we'll get into some of the stories and the products that you've created over the years.
2:08
But I think for the moment, to get started, for someone who's thinking about the Black Friday prep, there is a lot of lead up to making sure that your creative is dialed in and effective for that peak period of time.
2:23
And so I wonder if you could start by just telling us what you've been working on most recently, trying to solve that problem for yourself, and then finding ways to help others with that.
2:32
What is the state of, I wanna say the industry or the state of the art with respect to using AI to generate effective performant ad creative? How do you see the landscape right now?
2:46
And then lead us into what you think you're trying to do about that. Yeah. I'd like to take a step back in time a little bit.
2:52
It, it all started last year when, uh, Meta updated their Andromeda algorithm, and that algorithm allows Meta to determine the audience from the ad. So it can visually assess the ad.
3:06
B- back when I started doing e-com a long time ago, you had to go put in the affinity, and the demographics, and the a- everything about the person you wanted to target.
3:14
Their algorithms have gotten so good, they can look a static image or they can look at a, a video and they can determine who should we show this ad to. That's amazing.
3:23
So a lot of people are getting a lot better returns on their ad spend over the past year that know how to do this. The problem with that is then you burn through your ads a lot faster.
3:34
Because the audience that they're determining is more specific. It's not this broad, um, generalization of the American public or women. [laughs] Some people say women over 35, right?
3:44
That's a pretty broad generalist statement.
3:46
And so when you narrow it down, by definition you narrow down the bubble that you're going after, which means that depending on your spend, you can burn through that target demographic a lot faster.
4:00
So I, I like to say that in the old days, we used to find a trophy. It'd be this one amazing ad, and you could just power money behind it, and it would work for a long period of time. Now it's a conveyor belt.
4:11
How do we... I, I said this to, uh, someone the other day. Uh, you ever been to those sushi, uh, Japanese sushi restaurants where the food comes out on a conveyor? [laughs] Yes. Okay?
4:19
So you gotta, you've gotta put the ad on the conveyor belt, and as it rolls along, I take it off. It's great. I eat it. I need something else to be put on that conveyor belt, right?
4:29
I need new ads to be put on the conveyor belt as opposed to a trophy case.
4:35
That's how we gotta find the winners, and we have to give that algorithm and Meta the opportunity to decipher what are the best creatives based upon what we want, what they want, and everybody's a win-win for their customers included.
4:50
Yeah. Now, I wanna be able to forecast that. We're gonna get into this, some of the how, more specifically in a workshop in Austin, Texas coming up on the 2nd of October.
5:01
So if there's an opportunity to get your hands into it and you're in the area, that's when you should save that date, and you can contact me about how to participate in that.
5:10
But for the sake of what you've been trying to do most recently, I know you describe here on to your right and to your left, you've got some machines cranking 24 hours a day, building, building, building.
5:22
What have you found is effective to try and address this conveyor belt problem of having this constant stream of personalized ads? Yeah, so I mean, anybody that's ever run ads at, at, at any scale really,
5:36
the old method, and let's go back even, you know, before you and I were probably born, Patrick.
5:39
In the old days, they would just come up with a creative and they'd throw it out on a TV or radio before that or a, a billboard, and they had no feedback loop.
5:47
Or if they did have a feedback loop, it was really, really poor.
5:50
Obviously with digital marketing, we've shortened that feedback loop and we've got a lot better attribution over time, especially over the last five to 10 years.
5:57
And so the idea is, first of all, are we, do we, can we determine what the angle is or what the idea and the concept based upon our ICPs, right? So what's our ideal customer profile?
6:08
That's not a women o- woman over 35 It's more clear of what is their pain point, where are they in their life, what is happening with them that we wanna talk to them about the product that we're trying to sell them.
6:20
Can we meet them where they are in their experience, and hopefully solve the problem? Obviously, you have solution aware, problem aware, different pieces of the funnel.
6:29
But what we really wanna understand, this is what I find most people don't do, uh, very thoroughly, is they don't think about who are my ICPs, and they r- they create ads, especially with AI, that are just creatives, and it's a broad creative.
6:44
Instead of saying, "I've got health-conscious Hannah, here's a full dossier of who she is and what she likes, and let's create ads for her." Then we test, then we take the winners, we scale them up and do more of them.
6:59
We cut the losers. So that's the process I think of. And the bi- I'm gonna say this probably a lot, I think of it as a loop, right? It's identification analysis, ICP building, creatives, launch, analyze, feedback.
7:15
Just keep building a loop. It's a, it's a, everyone's, if you read Ray Dalio, uh, book Principles, I think of it like a feedback loop.
7:22
It's a recursive feedback loop that we can build that's positive so that we get smarter over time, and that's the whole way I think in, uh, models and systems all the time.
7:34
As the conveyor belt gets faster or the, the lifespan of an ad gets shorter, does that require an AI generative approach? Unless you want to employ a lot of people. I mean, I, I just don't think, uh...
7:51
Well, there's two aspects to that, right? You've got your, what I would call creative creatives, right? These are the aesthetic people. These are the people that want the prettiest ad.
8:01
Uh, they wanna win awards, and those are great.
8:04
I don't find those perform the best, and most of us are trying to, um, you know, meet a customer where they are in their place and give them something that inspires them to connect with us and purchase our product, and that's our goal.
8:17
Um, I don't find that the aesthetic, unless you have a brand that is very, very fashion-oriented or aesthetic, really matters.
8:25
But when you have humans doing that, and I still think humans are better than AI at that really perfect aesthetic, I think they're better, but that's not a majority of the ads.
8:35
In fact, most of the time the ugly ads, ugly, are the ones that outperform the aesthetic ads, right?
8:41
So I've had- Are you saying that, like for example, the sort of the TikTok aesthetic of casual, quick, it's real, authentic? Is that what you mean by ugly?
8:49
Yeah, but not e- but yes, I do mean that, but I also mean, like, ugly. Like, we've seen ads outperform that they don't look that appealing, right?
8:59
It's like, you've probably heard this old adage that, uh, if you wanna get Americans to understand the ad, you write for a fifth-grade level. Well,
9:08
uh, I don't, I'm not sure that everybody has amazing aesthetic taste either in the US, right? [laughs] I'm, I'm not saying... I j- I just don't believe that, right?
9:17
I mean, if you look at people's houses or just aesthetically what they like, you know, not everybody can be an above average, uh, artisan, right?
9:26
They're just naturally reversion to the mean, so it doesn't have to always be beautiful. I think it's more about connecting.
9:32
It's more about connecting to where they are, what are their pain point, what are we showing them and delivering to them at whatever they are in the funnel, and that's how I think about it.
9:42
Because all we're trying to do is find out how do we meet them where they are, and then can we give you the more varieties of that at pace, and that's where AI comes in. Because you can't scale- Meet them where they are.
9:53
Yeah. You meet them where they are. For someone who's thinking, "Well, but my brand has a certain point of view.
9:59
I have cultivated this taste that my brand represents," talk to me about the, the connection of those two things. I mean, it, it doesn't mean that you can't do that with AI.
10:09
I just think you have to have clarity on what you mean by aesthetic. Um, anybody that's ever built a brand or a, a logo, let's use a logo, this is a great example. I've seen people fill out questionnaires on logos.
10:23
They're like, "Make me a cool logo. I like the color blue." Uh, there is no chance whatever that designer is can hit the mark, and some people treat AI the same way.
10:35
So they say then to AI, "Make me a good-looking converting ad on Meta," and there's no direction there, right? So it's not that you can't do it.
10:46
It requires more input, context, design-oriented parameters around the AI to build things that meet your as- aesthetic, and that's not hard. I mean, that's doable if you're good at AI. It's definitely doable. Yeah.
11:02
Now, you, you have some things you've been working on that are specifically for TikTok- Mm-hmm... and I know we're talking about some Meta things here as well, and, and you started with Google. So
11:11
how would you distinguish between the TikTok sort of authentic aesthetic, maybe that, maybe that's getting to the video discussion. We're talking really more about static ads here. Mm-hmm. I guess,
11:23
is there a reason to make these kinds of distinctions by ad platform? This is a, a TikTok sort of- Mm. Mm... framing. This is the Meta, Instagram kind of framing. This is where we think about Snapchat.
11:33
Do you, do you go to that level of the platform optimization as it concerns this idea of ugly and beautiful?
11:41
I mean, I, I, honestly, I don't know if I have enough aesthetic ab- ability to determine what is perfect for TikTok versus Snap versus, um, Meta. I think it's more about the person. Like, I'll use my wife as an example.
11:54
I mean, you want the ads to be, to look like they belong on the platform, right? You don't wanna repurpose an ad that just doesn't match the platform at all, right?
12:02
But I do think, like my wife, if it res- if the, if the ad or the cont- I'll call it content, if the content resonates with my wife, I don't know if it really matters where it resonates.
12:14
I think those are just distribution channels to get in front of the people that you can hopefully connect with.
12:19
And I think s- the reason I, I talk a lot about static ads is I feel like a lot of times the video, the juice isn't worth the squeeze. The amount of energy, uh, literally physical- Yeah...
12:32
and electricity energy [laughs] and compute and time to do the creative for a video is awesome for a few videos. At scale it becomes far more expensive, and it has to pay off. What we've found with static ads, and we...
12:48
Just to give you an idea, we make about 100,000 images a day. Every day 100,000. That's why you hear a hum in the background. The GPUs are running. We use cloud GPUs. We make a lot of images.
12:58
And so a lot of times we stitch those images together into a photo carousel video or a slideshow, and that's for the TikTok stuff.
13:05
Additionally, then you could take those meta ads, and you can turn them out with statics really, really fast, and maybe use the statics for testing. And you're like, "I've hit the home run. I know my angle.
13:17
Let's invest and make more videos that meet this person where they are with that an- with that angle." Right?
13:24
But I think it's challenging to run enough video or sp- or unless you wanna spend the money on a lot of videos to do the testing to find the winners. Find the winners and then make more of the variations of the winners.
13:38
Yeah. That's my strategy.
13:40
I mean, at some point then that's really all we're caring about is the winner as defined by some performance metric, and the aesthetic is support of a brand ideal, but you're trying to sell product, and you approach it pragmatically.
13:52
Is that fair to say? I think that's correct. That's how I think about it. And you've thought about this because you've had your own brands. So before we get too much- Mm-hmm...
13:59
into some of the, the te- techniques or examples you'll show, tell me a little more about what stuff you've sold. [laughs] What kind of products have you moved across the, into the checkout?
14:10
Uh, I feel like everything under the sun, but I, I haven't done beauty, but I did, uh...
14:14
I first started in drop shipping in 2011, um, and then moved on to my own private label brand, started building brands, creating products to 2015.
14:23
Did a lot of products from China, so pet category, um, coffee products, uh, supplements, uh, arts and crafts, um, exercise products. So I've done a lot of different, uh, products, a lot of different materials.
14:39
Um- What kind of price points are those products? Some of those CPG sort of packaged consumables are s- are low price point. Maybe the exercise equipment is higher. What, what is the sort of the range?
14:47
The cheap- cheapest price I've ever- I've sold, built a product for is for $14.95 I sold for, and then the highest was $200. So 15 hu- $15 to $200. Yeah, yeah. So I haven't done anything more than, um, $200.
15:03
I haven't done, like, appliances or anything, you know, larger than the $200. Yeah.
15:09
And so in that you've had the accountability for profit, and I'm curious as you think about this time of year and the Black Friday big discounts and offers- Yeah...
15:19
do you approach this latter part of the fourth quarter as changing the kinds of offers and ads you're, you're making at this point? Have you already dialed it in for the season? I mean, I don't...
15:30
I, I'm, I've [laughs]... Boy, I, I've got enough battle scars I think from Black Friday and Cyber Monday over the years.
15:36
It felt like, especially when Amazon, I don't remember if it was eight, nine years ago when they first started doing Prime Day and really honing in on that, I felt like my revenue just kept going up, my profit kept going down.
15:47
I kept feel, I felt like I just kept discou- my, my top line number would get huge, and then I'd be like, "Oh, and I made $37." I did 300,000 in revenue, and, and I joke.
15:56
It wasn't 37, but it was, it kept going larger on the top line and smaller on the bottom line. And so I have, have kind of gone the other way where I'm not a huge fan of discounting.
16:05
I've also had a few friends that, um, have I call them lifestyle brands that don't believe in discounting much. Uh, and I think they've swayed me more of that direction of not over-discounting, right?
16:18
Not degrading your brand with just discounts everywhere. But it does depend on the brand. Is it a truly commoditized product? Is it a unique product? Is it, is it a recurring product?
16:28
Is it a single use product where they buy it one time? I used to sell foam rollers, Patrick. If you ever use a foam roller- Yeah... they don't break ever. [laughs] So they buy one EVA foam roller, Patrick.
16:40
Uh, they're not coming back for another one unless they own a gym. So, uh, I learned that lesson a long time ago. Um, try not to sell things that, that never break. Uh, it's hard to get a- get them to buy again.
16:50
Um, so it just, it depends on the category. I think that's really understanding your category. I think lifetime value, so many people don't measure their lifetime value correctly.
17:01
And I mean, at the end of the day, it's LTV over CAC. I care about... I'm a analytical guy. I care about the math, right? I've gotta make enough money from this customer in the long run to afford to run the ad.
17:13
If I can do that, I can make money. And so many people don't really have the right data to understand what their LTV even is. They guess. They're like, "Oh, it's, it's two months. It's four months." Yeah.
17:26
And it's like, boy, that's a, that's a double, that's 100% difference. We gotta figure this out, guys. Yeah. So I think it's individualized is what I would say. Yeah.
17:34
Well, I can't help but say then that the, the essence of that lifetime value calculation and figuring out how to do that for the Black Friday offer is what we will cover on October 9th in Austin. Oh, there you go.
17:48
The next, the next workshop in this Frankly Ecommerce season we're gonna bring in Dan Hansfeld of DataSmith, who does this so well. And we'll be able to really get into the making sure that your offer is
18:01
yielding that profit across different segments of your audience. So that's gonna be a great follow on to what we're gonna talk about with you, Rob.
18:09
But for your sake, um, and I w- and I also wanna forecast, we're gonna bring in a- another guest here in just a minute, uh, who's helps us with the ad stuff as well. Before he does-
18:20
What in the essence then is, is your primary tool for generating ads today? What's it called? How does it work? Yeah. So I built a software called Da- DittoAds.ai, and I can show it if you'd like.
18:33
But the, the concept behind it- You're welcome. You're welcome. The, I'd love to show it.
18:37
But the, the, the concept was a friend of mine, I was hosting a call with some friends, and he's spending about three grand a day directly to Amazon, and his team couldn't keep up with enough quality creatives
18:49
before they would burn out the creatives. And so I said, "Well, give me some time. Let me build something." And that turned into Ditto Ads, and this was maybe three, four months ago.
18:58
And then I got some refi- I refined it, talked to a few more friends. One of my buddies does 3,000 creatives a week. Um, he spends about a million bucks a month on Meta,
19:09
and he, uh, I don't want to say fired, but he reduced his, uh, human team after running head-to-head tests, uh, significantly because the, the creatives were outperforming his human team in not only quantity, but in actual ROAS performance.
19:27
And is part of that, do you attribute because of the personalization you can bring to the more specific type of buyer that Meta wants to be showing it to? That's a big part of it.
19:36
I also think you can make it, um, you can, you can just crank out numbers, right? The algorithm is going to determine what works or what is working, what isn't working.
19:45
You've got to feed the machine, for lack of a better word. You got to feed the machine. If you don't feed the machine, the only person you're hurting is yourself. Like, they're not stupid.
19:56
They're not gonna serve the ads that are garbage 'cause you're not gonna run ads with them. So they're, you're trying to enable them, and it's like, um, how do we row together? We're in the boat with Meta.
20:08
How do we row with Meta, not against Meta? And if we don't give them enough creative, it's like, "Hey, you got one oar. Stop s- stop rowing in circles." Like, "Oh, well, that doesn't work.
20:19
I got to give you two oars, and the second oar is gonna be more creative for you to actually work with." I like the analogy of rowing together. There's a real synergy there, right?
20:29
And so much people think these ad platforms can be some sort of an antagonist, like, you know, that it's always taking their money. But, but you're dealing, you're framing this in a way that shows it can be,
20:38
you know, you're, you're, you're, you're working together, right? You're trying to get people to just to discover you. Yeah. And it's, it's the most scalable platform from this perspective, from a visual creative engine.
20:48
I'm happy to show a couple things- Yeah... if you've got a few minutes. Yeah. If you can, if you can make it work. Yeah. Let me, uh, yeah. Let, well, I'm sure we'll be able to make this thing work here.
20:56
Let's get this going here. At Facebook Ad B- Ad Library up there right now, but there, there's two things. I'll, I'll go this one real quick. I just built this last week.
21:05
This is a, uh, an app that allows people to ter- determine, um, how they're positioned. And the idea, my idea behind this, and this is a friend of mine's brand. It's a, it's a multi-eight figure, uh, clothing brand.
21:17
And so I said, "Hey, Brian, I'm gonna run this and just run your product and see how it does." And the whole idea is I created a tournament, so it runs Brian's brand and his ads that exist against his competitors,
21:32
and it can determine how it performs. Now, we're not spending money here. What we're doing is I built personas,
21:41
and so I built 30 personas here, and then the personas utilize the LLMs to determine how Brian, are they stopping on Brian's? Remember Brian's over here. Or are they stopping on Fresh Clean Threads? And so you can see
21:57
why they stopped, and then you can see the ad breakdown of hook, format, offer, things like that. So this was a tool I built last week, literally last week, [chuckles] to identify how you're doing with your ads.
22:09
Couple quick things on here. This is an aggregation of all the ads and ideas across his subvertical and his subcategory. Then I thought this was a fun one that it added.
22:19
It add the AI guess whether this ad was, it takes out your logo. Is this, who, what brand is this? And in this case, none of them guessed Brian's brand.
22:31
And even this one had his, his logo still in that one, and it still missed. It guessed Cuts. And is it perfect? No. Is it directionally perfect or directionally accurate? I'd say yes.
22:42
Last thing here that's really critical, so many people, we're not, we don't have the time to talk about this today, but so many people have an ad that then goes to a landing page that is not congruent to what the ad is.
22:54
And if you're not measuring this, and if you're not testing landing pages, there's a really good chance you're running, you're running great ads and getting a lot of clicks and a great click-through rate, but you're not delivering on the promise.
23:07
You're not delivering on the promise of the ad that got them interested over here on the landing page.
23:13
So this was something I built as a tool just to give them a quick idea, I did this for a few friends, of where you sit and how it's working.
23:21
So, like, I pulled 1,000 ad- ads across eight brands in the men's premium basics and then wr- gave him this report, and then he'll put some ad dollars behind...
23:30
Oh, and it also, it gave him some, uh, some ads to run, gave him a few ads to run. So this was something that I'm just trying to get to the essence of why things are working, but this is Ditto Ads that I built.
23:41
Um, the idea here is if somebody comes in and adds a new brand, this is just a test account, um, you can actually identify based off the, off of your URL and brand the color palette, anything you don't want, any generation constraints.
23:57
Then after you do that, you can go to the, um, audience, and what'll happen is the, the app will build out five ICPs for you. And these are not, I've seen some ICPs that are 30 words long. This is not the case.
24:10
These are legitimate ICP builds, right? These are buying triggers, voice samples, generative directive. And so that allows then the app to come in here and select and say, "I want to build for Alex Johnson," right?
24:23
So we can go in, and we can build ads for Alex in the platform And then we added the- Let's pause, let's pause there just for a moment. Yeah, please.
24:31
Um, I'm gonna let in Hal Smith, and we'll continue this conversation with this.
24:35
Take a break for a minute to consider all the shopper signals that are being sent when visitors browse your website, and you're capturing probably not all of them.
24:45
There's a opportunity to be a lot more intentional and aware of these signals using a Shopify app called Swim. Now they're installed on a bunch of Shopify Plus stores, and they got started being a wishlist app.
25:01
And that's good, but there's so much more than that. And I got excited about what Swim is doing when I happened to spend time with their CEO, Arvind.
25:09
I was together with him in Utah at the Ecom Elevated Conference put together by Commerce Catalyst and Matthew Holman there.
25:17
It was a great conference, and I had a chance to see and understand the bigger vision for how their product at Swim is evolving much beyond the wishlist idea.
25:27
All those shopper intent signals yield a whole wealth of data and opportunity to make things better. So Swim is gladly a sponsor of our Frankly Commerce season.
25:39
They're gonna be helping at a workshop coming up in Austin around conversion optimization, and I welcome you to check out what's going on at getswim.com. Now back to our show. Welcome.
25:54
So Hal is gonna be with us on October 2nd at the workshop in Austin, Texas. He runs H Street Digital and has some much to share.
26:05
You catch us at this moment going through what Rob's been building called Ditto Ads, and we're describing how this came to be.
26:13
His background in developing his own products led to this idea of how can you use AI more effectively to keep this conveyor belt filled with creative that's personalized for the kind of audience that Meta has targeted as appropriate.
26:26
So Rob, continue on with your demonstration of what's happening. Go ahead, Rob.
26:31
So, so one of the cha- So that's the brand and the audience, the setup portion of giving Ditto Ads the, the, the starting point, I would call it, of the feedback loop, right?
26:38
And then a lot of people struggle with, uh, I don't have any ideas. Where do I start? So what we did is we built a showcase. And what a showcase. These are 4- 40 ad concepts.
26:49
Um, actually this is the panel you and I sat on, uh, Patrick, at the last event. Remember, uh- Yeah... Kari? Right. So while Kari was presenting, I actually built this in Ditto Ads. I, I watched it.
27:01
[laughs] In our- I know. That's why I was like, "Oh my God." In our testica. So what we did is we took...
27:06
What we did is we took 40 ad concepts, headline, offer, promotional, testimonials, you know, your traditional concepts, anybody that's run ads before and after UGC native,
27:16
40 that we know tend to work for a lot of different people, right? Advertorial, editorial content ad card.
27:22
And then what we did is we run the information that we've done, we've, we've input with the brands and with the audience to make the five ICPs, and then we make this. And this is more just inspirational, right?
27:35
This is like, okay, um, I tend to like the UGC stuff. I'm just biased I guess. But I like this one. I like this one. Um, but then it, it's more like a Canva experience at that point, right?
27:47
They can come in, they can do new variations, they can refine it, new sizes, and they can make iterations of that as much as they want. And then we said, well, that's not enough.
27:57
We gotta give people more discovery, right? So we, uh, pulled in 1.3 million ads, um, into the database and said, "All right. Well, we love Allbirds, and Allbirds matches who we are."
28:10
Well, I mean, they didn't do, they haven't done very well, so maybe Allbirds is a ba- Oh, Farmer's Dog. Let's say we sell dog pro- dog food, right? Farmer's Dog.
28:16
And you're like, "Boy, I love Farmer's Dog, and I love the messaging, and I sell pet products," right?
28:21
Well, this is an analysis of kind of their media mix, uh, their formats, and their landing pages, and then here's some of their ads.
28:29
So if you need inspiration and you want a model or a copy or do whatever you'd like with that, this is inspirational, right? And then the person can come in and pick something, and then they can copy the link.
28:40
They can recreate for their brand. And then it'll build off of that. It'll build for their brand off of the information that, that they have already. And then, uh, Ad Studio is where all the ads can live.
28:52
And also, if you know what you wanna build, you can just create from scratch and do a startup ad. Uh, we are small approval with Meta, still waiting for multi- you know, full approval with Meta to be able to push.
29:04
But the idea will be that we'll be able to run campaigns to Meta, extract, build, and then unlimited, um, creatives. And the key is find something that works and then crank it up, right?
29:17
Do more of what's already working. And it doesn't...
29:20
The, the beauty of, I think, a lot of people that are not heavy into AI is that they don't understand, like, a lot of these problems that we had a year ago have been solved, right? They don't exist anymore.
29:31
And what I mean by that is we used to really struggle with text. We don't struggle with text anymore. You can look at the showcase. Maybe at a 40, maybe there'll be something a little off.
29:41
But rarely, rarely do we have issues with text anymore. And keep in mind, no human was in the loop to build this. Yeah. Right?
29:49
The one you're sharing there, an example, that came out of an event hosted by Kevin King, who's famous in sort of the Amazon space and involved brands that may have a strong Amazon presence, and they're wanting to move to their own Shopify direct-to-consumer kind of opportunity.
30:05
That's the case here with a, also a company in the UK that wants to move into the US market, a bigger opportunity.
30:11
So in that case, we know that that, I would say, let's just say that that brand is relatively unsophisticated- Yes... relatively small team. Yes.
30:21
But, you know, there are brands in Austin that are operating at a high level with a team of specialists. I'm thinking about, you know, how they may have already a paid performance manager full-time on staff
30:35
How do you rate things like this? How do you correlate these tools with the sophistication of the user or the brand? I haven't met many people yet that have an internal team that I would say is fully utilizing AI.
30:52
So I would think it... I don't-- I would think this is a tool in their tool belt. Someone who is more sophisticated, uh, let me, let me use a good analogy here. Great, I, I got one for you.
31:02
So, uh, give me a hammer and give a carpenter a hammer. There are two vastly different outputs with the hammer. I'm probably gonna have a bloody finger, and he's probably gonna have hammered in 100 nails.
31:17
It's a tool in their toolkit, and the more sophistic- We tried to make it as easy as possible and approachable for everybody, right?
31:25
But it-- the more sophisticated they are, the easier it is to, to build it out without having to deal with API calls, ChatGPT, concurrent limits, all the things that happen. How do I get the prompting right?
31:39
Can I continue the design guide? All those things that, honestly, most of the people that run high volume don't wanna deal with anyway, to be honest. Yeah.
31:47
Well, let's bring in Hal at this point, because let's say that Hal is a power user, or let's say that his team is operating at a high level of sophistication.
31:56
You're gonna be contributing to the workshop as well and have your own, um, input to that.
32:00
But as just a first introduction, as you see things like this, maybe you could give your perspective on how your team is doing something similar or different and how you kinda see the opportunity there. Yeah.
32:12
So first of all, I, I mean, this is great, Rob. Um, it's interesting. We're trying to build this ourselves right now, [laughs] so I, I know exactly the problem you're solving. I hear this a lot- Yeah... Hal. Yeah.
32:22
I hear this a lot from a lot of guys. But, you know, the key word there is trying. So, [laughs] um, yeah, so I think where do we see the application here and, and, like, the interesting things that come out of this?
32:32
So first of all is, is giving creative strategists and media buyers enough creative variety to consistently test concepts without burning out, right?
32:42
A big issue that media buyers have is, you know, they, especially with Meta today, they run into serious limits with Meta, you know, running spend behind certain creative, it dying pretty quickly, and then they don't really know what to do to go, you know, what the next step should be.
32:55
How do you go back to the drawing board and come up with more concepts?
32:58
The other issue, too, especially with, let's say, creative strategists, is it actually takes a lot of time to put together lots of different briefs, uh, and distinctly different concepts, and then to do the research, um, and then give that to the, the, you know, graphic designers, video editors to actually create the content.
33:16
Um, so I- the-- what I see here and, and the problem we're trying to solve, and I think you're solving is, is one, it's, it solves the problem of just coming up with new ideas quickly and, and h- being able to come up with a lot of different variations of that, which will continue to feed Meta.
33:30
Uh, but the other thing, too, is it, it cuts down a lot of the production time. I think, you know, you kinda mentioned this, too, AI has solved a lot of these problems.
33:38
I think the misconception that people have is the amount of time it takes to, to make ads like this today, like with these static ads, right?
33:46
Um, and then the amount of QA you need to do on, like, let's say, the, the hook or headline text.
33:51
Uh, and, and, you know, if you do this kind of manually with a team, if you have graphic designers cranking this out, it will take a lot, a lot of time, especially if you want this much variety.
34:01
Um, but this really does solve that problem. And I think the big takeaway I have at least, or what I would emphasize to our clients, is this allows you to rapidly test lots of different ideas.
34:12
You might not have the one, you know, killer ad that's really gonna scale the account, but at least you're gonna cover enough surface area of testing concepts to where you can take the things that are starting to work and then put that into higher production quality, or you can make tweaks to it and then start, you know, generating ads that are gonna take up a lot of spend and actually be, like, a true scaling ad.
34:30
Uh, is that something you agree with, or, uh, are, are you guys able to kind of further iterate?
34:34
I-- Yeah, you can further iter- So I-- Just to give you a little background, Hal, you just missed the beginning part, but we, we do this with- We-- I make 15,000 TikToks a day- Wow... for one product for one brand. Yeah.
34:46
And the agent every day analyzes every single input to those 15,000 slideshows. And so what happens is then we're able to differentiate upon different aspects of the ad. Let's say it's the hook.
34:58
Let's say it's the headline, whatever it might be. The-- I actually think the software's better beyond testing than even just the testing because we can go make 100 variations of the ad that's already the winner,
35:12
and now we have them tomorrow. Them today, technically. [laughs] You'll have them today, and then you can run them tomorrow, right?
35:18
And I think what you talked about and what I talked about a little earlier with Patrick is that the ad fatigue is real. Andromeda burns it through. So once you can test, find the winner, get the data,
35:30
come back, do more of the winners, it allows you to do a lot more variations within the platform than just the testing, right?
35:37
So you can take one that wins and now pick every single parameter or variable in that image and change it. I--
35:45
A woman who spends, let's just say she spends almost a million dollars a month on Meta, reached out to me last month, last week, and we had a conversation, and I ran an analysis.
35:54
I ran one of these, uh, arena ads on hers. The 85 of the new creatives, which is not enough for her, that she did static ads this month, 82 had the exact same words. Hmm. Yeah. No variation. None.
36:11
And she's spending almost a million dollars a month. Yeah. And so she's using a team of, of humans, as far as you understand it. She's paying an agency- A lot... way more than she should, to give her
36:24
very simple variations- Mm-hmm... which is not feeding the algorithm new content because it's using all of the same words. And by the way, when I shared that with her, she didn't know that.
36:36
I wanna tie this back in, um, I
36:41
I had a conversation some time ago with, uh, Dan Hansfield, who's gonna be teaching the following workshop on the 9th of October about measurement and offers and testing, and we had a discussion about what was posted by Cody Sanchez in Austin a few weeks ago on LinkedIn with this idea that a brand under 5 million just should not or cannot effectively advertise on Meta, and how you, you, you called that out and we had a little back and forth about that.
37:10
As you see tools like this, as you think about the dilemma, and maybe you can fill in, Rob, what that was really about, if you remember, um, and maybe you can speak to that now, as you see the opportunities evolving with tools like this, is that still true?
37:22
Was it ever true? What would you say now? Well, yeah, so, so one important caveat.
37:25
I, I, I was not saying that brands shouldn't spend on Meta at less, less than 5 million, but they probably shouldn't be working with agencies at that level, because agencies that are specifically servicing that tier are not good agencies.
37:38
Um, the number one issue that I've had a long time in talking to prospects and brands is they've just been burned by agencies, and the reality is this, um, how much
37:49
it, you know, at the amount that a brand at that size is willing to pay an agency, which is typically 2 to 5K a month, like, what quality talent do you think is gonna be working on that account?
38:00
And that's, yeah, and that's the biggest issue is, like, I, I, any br- any agencies that are servicing that tier are just not ones you can really trust, because they're just trying to give you a very basic setup.
38:10
They're not giving you any attention, and the only way that agency's gonna be profitable is if they work across, like, 30 to 50 clients. Um, so that means, like, you're not getting the time of day.
38:19
Uh, and so that was my main point. And the other point I had was at that size, it's really risky because you are effectively pre-product market fit.
38:27
Um, and so it's really hard to successfully test and kind of validate your products, your offers, your angles at that size.
38:34
Um, it's, it's tough, but I, I personally have, have s- have launched and scaled multiple DTC e-com brands, um, that have, you know, I've scaled to a couple million in revenue, um, and, and it definitely works.
38:47
And I would 100% say Meta's the best channel, uh, a startup DTC brand should, should go to in order to start scaling up and really starting to figure out who their customers are and who the right messaging is and what products they should be advertising.
38:59
It's the best place to learn those things. Um, but you have to have a really good structure for it, and you have to find really efficient ways to do it. Um, and so, you know, obviously I like a tool like this.
39:10
I think this would be great for a brand of that size because you can produce really high volume that has a variety you need that can also just do the research and testing and learning that you need.
39:18
Because the biggest thing you gotta do at that stage is you really need to test a lot of variants at first, and then you need to kind of whittle down and focus on the things that are working.
39:28
But if you can't have the variants at the beginning, if you can't have distinctly different, you know, ICPs or personas you're going after, different formats, different messaging, you're gonna be stuck in this very narrow lane, and your likelihood of success is very low.
39:40
But if you spread it wide, uh, you'll quickly kind of figure out pockets that are working for you, customers that are messaging, that is, and then you can start doubling down in those areas.
39:49
But if you don't start with a good structure for testing and have that variance, um, it's really hard to figure out what's gonna scale for the brand. I wanna double-click on what Hal said there.
39:57
I also agree that you see brands at that level that are, the lack of sophistication sometimes, it, th- they don't have tagging right. Correct. They don't have landing pages that look good or work.
40:10
So they have all of these little things that even with a great structure, i- isn't gonna work, but someone who's more experienced and understands all those little things add up to success or failure, right?
40:24
And if you're sending people to a crappy landing page that doesn't have tags, I mean, go... I don't care who you are- Yeah... good luck. I mean, it's not gonna work.
40:33
And so I think that's what I see a lot of people that are inexperienced with it.
40:36
You see, I, I, I started a lot on, I did a lot on Amazon in the, in the beginning and, and for a long time, and a lot of what I call Amazon-only brands lack sophistication off- Mm...
40:46
of Amazon, and I find that that is where they make these critical mistakes that they don't really understand. They don't know what they don't know, which is all fairness.
40:56
But then, like Hal said, I mean, if you're gonna spend 2 to 5 grand a month on an agency, I mean, you're gonna get what you pay for.
41:03
So if I hear the example that Rob spoke to of the million-dollar-a-month brand with not very good variations, let's suppose they can be better about that.
41:13
It still sounds like all those variations and all those testings sound expensive. So going back to this question about for a brand 5 million or under,
41:23
suppose they do the things correctly you've just, Hal and, and Rob spoke to,
41:28
still, is there a minimum amount of ad spend to viably test and do all these things, or does some of the, the tools we're talking about here help us
41:36
be more efficient at a lower total ad spend cost for some of these smaller brands? W- we talked about this.
41:43
Hal, why don't- The tools definitely, yeah, the tool- the tools 100% help you get more creative to test at a more effective cost and a more efficient cost. There's no doubt about it.
41:53
I think so many people think of advertising as an expense instead of an investment, and maybe it's just my mindset. I've always thought about it as, uh, I launch a new product.
42:04
Let's say I launch a new product back in the day on Amazon. I spend a lot of money on ads, and I spend a lot of money on ads because I need data and I need data fast.
42:11
I need to understand, this is before a lot of the data was exposed by Amazon. I needed to understand what keywords worked.
42:16
I needed to know now so that I could refine my, my product, my detailed listing page, and then also my ads. So I'm investing to get information And that's how I think about running ads. And a lot of people don't.
42:28
They think of it like an expense. And you're like, I mean, yeah, it'd be great if you never had to run ads from a profit standpoint, but that's a very, very difficult road to, to, to go down. Yeah.
42:38
I think, um, I 100% agree with Rob. I, I think it's, I mean, brand owners are entrepreneurs, and I think they have to think like entrepreneurs.
42:45
I think everyone is just, you're, you're trying something, and you see if it works, and if it doesn't, you iterate and keep going from there.
42:50
But I think the kind of through line here is if you're gonna test, uh, you know, with Meta, if you're gonna go test a set of creative, you have to have a very clear, like, thesis for, you know, uh, answer you're trying to get, right?
43:02
And then you have to have a structure for how you're gonna get that answer. And so my recommendation for brands and, and how to test a certain amount of budget is th- think of how many things you're testing,
43:14
and then how many, you know, conversion actions you wanna have, right? Where you're gonna say with some degree of confidence, like, "Oh, this ad is a winner," right?
43:22
Think about, like, estimate your cost for that action, and then tie that into how many things you're testing, and then that'll give you a rough estimate of how much budget you should spend.
43:30
So for example, let's say you have a, and I'm using simple numbers here 'cause you're never gonna get $10 CACs on Meta these days, but let's say you have a $10 CPA, um, and you're testing five different ads, right?
43:42
Um, so basically you should spend at least, uh, 250 bucks, um, and that'll give you a, you know what, let's say a purchase across each ad. But you know that's not h- typically how Meta works.
43:53
Uh, typically Meta's gonna distribute bu- uh, budgets unevenly based on which ads are working. So you can just plus up a bit and maybe do, like, let's double it and say 500 bucks.
44:02
500 bucks should give you, like, let's say five or so conversions on your top-performing ad.
44:07
And then when you go run that test, you should be like, "Okay, I have one ad that got five conversions, another one got three, another one got one. These ones got none," right? And then it's gonna give you some clarity.
44:15
It's like, okay, this is a winning ad and I know, you know, to invest more into this, and I, and I'm gonna go study and analyze why that ad worked.
44:22
You know, preferably if you set it up at the beginning and you kinda have a case for each of those ads and why you think it's gonna work, right?
44:27
And if you do that and it runs and it, you know, uh, i- if it supports that argument, then it's great, right? But if it doesn't, then you have more to learn from that.
44:35
But I think if you go ahead and set it up based on what your target action cost is, as well as how many things you're testing, you get a rough estimate of how many actions you need per test, uh, that's gonna give you the best structure, right?
44:47
And, uh, the, the issue that I think a lot of brands have is, um, they don't actually understand how to then benchmark these things to set up the test.
44:54
And what I mean by that is you, you also wanna create a little bit of a media plan where you have your, your budget, your estimated cost to advertise or CPM, estimated click-through rates, estimated conversion rates, right?
45:06
And if you can set that up, then you have a lot more clarity as you run the test if anything else is wrong.
45:12
'Cause if your CPMs are radically higher than, uh, what you originally forecasted, you know there might be an issue with your creative.
45:19
Um, if your conversion rates are much lower than you forecasted, then you're like, it might be an issue with the landing page too. But those two lenses will give you a much better idea of what's working and what's not.
45:29
But I think, and then what you do is you just run small kinda sprints. You, you test a budget, you have a thesis, test a budget, and you keep going from there.
45:36
What most brands wanna do is they, they just go set up a campaign with a bunch of different creative that are just random ideas, and then they kind of just have this Hail Mary approach [laughs] where they just go run it, and they're sitting there and they're super stressed and super anxious about it.
45:50
And the reality is you shouldn't do that. You should just go for the short gains, um, and then just make sure you're learning every single time.
45:56
That will get you to success much better, more accurately, and, and frankly, faster than just trying to do a bunch of random things and doing a Hail Mary. So...
46:06
I wonder, I appreciate the way you're framing those kind of actions.
46:10
Rob, if you were, we can go into this deeper into the workshop itself, but for those who won't be participating in that, much more people will be listening to this and not there.
46:21
As you were approaching the Black Friday period, from what Hal just said and the tools that you're both developing and imagining you haven't yet developed, what do you suggest for the brand to sort of at this point in the year, going into the October period, almost November,
46:38
what are those couple of things that, that you would add on to what Hal just suggested or, or add or suggest yourself?
46:44
I mean, s- step one is does your product, is, does the seasonality of Q4 affect your product heavily or less? I mean, some, every product's different, right?
46:53
Like categories of supplements, it's more January than it is, you know, uh, November, right? So it depends on, I, I, this is how I think about everything. It's uh, what's the product? What's the market?
47:04
What's the affected timeframe in the seasonality? And maybe Black Friday and Cyber Monday is your setup for your Q4. And what I mean by that is, is your product giftable?
47:14
I used to sell, uh, whiskey products, whiskey decanters, things like that, right? So Black Friday was a great way to get awareness, and people came back and they bought it as a gift.
47:24
You know, 60, $70 product, they buy it as a gift for their father, their brother, their husband, you know, a month later, right? And that's part of the retargeting, right?
47:34
I think so many people don't under- I, I keep going back to math. People don't understand the math, and they don't do the math. They don't know LTV. They don't know CAC. Oh, and then they don't retarget.
47:44
So it's like until I ha- I mean, it's all developing a plan. This is probably what Hal does with his customers. I mean, this is just developing a plan, right?
47:50
It's a strategic plan, and then the tasks are part of the strategic plan. So step one is what are your goals? What are you trying to achieve?
47:59
Does your product, you know, uh, you can go watch a YouTube video about running Black Friday ads, but if you have a different category that doesn't match that, I'm not sure that's the right approach.
48:09
I- in fact, now that I was listening to Hal talk,
48:12
I think if I was starting a new brand and I didn't really know how to do this, I would pay somebody that does- can't afford a great agency an hourly wage for one hour of consulting.
48:24
I actually just thought of that when Hal was talking. I was like, "Why wouldn't someone just pay-
48:29
500 bucks, 1,000 bucks for an hour or two hours to help develop structure, strategy, lay out some tasks, lay out the tools, and now they can then go and, and act on that. Yeah.
48:44
Well, Hal, there's a new link on your homepage. [laughs] I think you should drop it up. It'd be great. I thought of doing that before, yeah. It's a good idea, Rob. [laughs] Um, there's a platform for it too. Well, good.
48:54
Uh, one thing I was gonna a- Yeah...
48:55
add to that quickly too, for, for seasonality around brand, um, I think when brands, like, the, the kind of misconception is, like, throughout the year, your, your revenue just keeps going up.
49:05
It's always up and to the right. But that's not actually how growth happens for any D2C brand. And, and all brands are seasonal in some degree.
49:13
Uh, I would say the less seasonal brands are still heavily dependent on Black Friday, Cyber Monday, and holiday shopping.
49:18
But I think the better way of thinking about it is you have to think about where your brand indexes really well and where there are times to make a lot of money.
49:26
And it's always, like, it's always, like, these big kind of like, uh, kind of chunky movements in revenue.
49:31
And so, like, you'll have, like, a big kind of pop around, like, Labor Day, 4th of July, Black Friday, Cyber Monday, or Christmas, and your expectation should be you wanna scale and, and, and have a lot of attention on what you're doing in those kind of peak promotional periods where you can make a lot of money.
49:46
But then recognize there are times where you just kinda have to pull back. Like, um, let's say August is always kind of a flat month, right?
49:52
Uh, but if you kinda plan around those, those leaps and those jumps, uh, in revenue and, and overall just customer behavior, you're gonna be a lot better off.
50:00
And so planning around, uh, promotional periods or really studying the seasonality of your business is gonna help you a lot.
50:06
Well, coming up on the top of the hour here, I appreciate you guys could probably go back and forth, and I know we're gonna have a chance to do that in the workshop. But Rob, would you say anything else to close us out?
50:16
What else would you want someone to know who has heard this conversation so far? Maybe where they can find out more about what you're doing or, or what you would suggest. Yeah.
50:23
Uh, if you wanna reach out to me, robgreen.com easiest way to do it, um, or LinkedIn. Either one of those is great. I, I think too many people... It's the old Abraham Lincoln story, right?
50:36
He's got six hours to chop down a tree, and he spends five hours sharpening the ax. I think too often people just start chopping at the tree.
50:44
And so take some time to develop a plan and a strategy, not just a series of tactics, and not just the latest and greatest LLM that may or may not work for your creative idea. Then once you have that plan, execute.
50:58
Well, thank you. I appreciate you being able to talk with us on the Frankly Ecommerce Show. Hal, anything else you would say to close us out? No, Rob, I love that. I love that comment.
51:06
I'm 100% about that 'cause I, I think when you set a strategy, and you have a plan, and you have clear targets you're trying to hit, that, that really gives you a lot more focus, and it gives you a lot more clarity on the actions you actually need to be taking.
51:19
Uh, a big issue in this place, in this space, sorry, for D2C, media buying, et cetera, is, is LinkedIn and Twitter.
51:26
People are posting these random tactics all day based off of these very recent trends, but no one's ever contextualizing it and the actual impact of these tactics, right?
51:35
And, and no one's actually contextualizing it against a business and really trying to tie it into whatever, uh, you know, uh, agenda they have or whatever they're trying to get done for the business.
51:44
And I, it, it's, it is, it's a huge risk to the business just to go grab a bunch of random tactics and throw them in.
51:51
The best thing you can do is just really set up a strategy, have a clear plan of what you're trying to do over the year or next 90 days, and then start to, you know, tie it to specific goals for the business.
52:01
Uh, it's gonna give a lot more clarity to, like, what's working and what's not. But if you don't have that, you don't really know what's working and what's not.
52:06
And I think a lot of brands get stuck because they just keep throwing random tactics in, and they don't, they don't see things move.
52:12
And, and they get frustrated, but they're not really tying it to something they're actually trying to solve. Yeah.
52:17
Well, there's no shortage of tactics, but I think the context of having it in a workshop, in a small group setting, which we'll have throughout the month of October, is we can take the context of your particular business with the trade-offs or the constraints, and we can
52:32
sort out where the next most viable test is. And I love the way Rob was able to frame it earlier of that ad spend, you are investing for information, and you can use that then to dial and refine it in.
52:45
I also love the observation that we are, if we're gonna be chopping down trees or cutting logs, we're gonna spend a lot more time sharpening the ax, sharpening the saw before we just launch into it.
52:56
And so those are both, uh, great observations to take from this conversation. Thank you so much, Rob Green and Hal Smith, for being on the Frankly Ecommerce Show. Awesome. Thank you, Patrick. Thanks, Patrick.
53:06
[upbeat music] We hope you enjoyed this episode of Frankly Ecommerce with host Patrick Scott Pittman. Read episode notes and get links to learn more at franklyecommerce.com.
53:18
Next time you're in Austin, check out the Austin Shopify Meetup. It's open to all.
53:25
There's a good chance you'll meet just the right person, as it's one of the longest running meetups with 2,600 participants going on 15 years.
53:35
If you'd like to get more involved beyond the meetup, then Frankly Ecommerce pass holders participate in workshops and special invitation-only events. Subscribe to our newsletter at franklyecommerce.com to learn more.
53:48
This episode is sponsored in part by the friendly team at eBusiness PROS, a marketing firm in Austin, Texas. If you're a brand seeking to connect with an enthusiast audience and hold their attention,
54:01
if you want less discounting and more story, if you're curious about what we might mean by attuned marketing as contrast to behavioral trickery, then contact us at ebusinesspros.com. Ask for Abigail.
54:15
That's me.
54:34
[upbeat music]
Frankly eCommerce
Listen on
Apple Podcasts
Apple Podcasts
Spotify
Spotify
YouTube
YouTube
Recent episodes
Marketing leadership through story, data, and growth bets
Sep 28, 2026
What AI Didn't Solve about Data Analytics
Sep 19, 2026
Why Now is the Best Time to Launch a Brand with Chris Lang
Sep 4, 2026
