SPEAKER_01 0:00

In this episode, Dylan is going to be diving deep together, Brit walking me through the process that he used with one of our clients to take them from a $3, sometimes $4 to $6 cost per click on their meta ads using this AI creative process to dropping them to about a dollar cost per click and doubling the return on ad spend in a short window of time. So in this episode, it's pretty tactical. We're gonna be walking through some really specific things on how you can get better outputs with your AI creative, allowing you to create more volume and get better results. Enjoy.

SPEAKER_00 0:37

In this episode, I'm going to be walking you through a process that I went through with a client that we were actually able to literally double the row as just by doing this. So I'm really excited. But before we get into this episode, I did want to just let you know we are on a quest to become the number one e-commerce podcast, and every rating helps. So if you've been listening for a long time, or maybe just a few weeks, it would mean the world to us if you would leave us a rating. By the way, I'm Dylan, and this is Josh, founder of the ECOmerce Alley, and we're gonna get into this. You ready? Heck yeah, I'm ready.

SPEAKER_01 1:12

This is if you're listening to this, don't let your designers or your outsourced creative team listen to this because this will replace the need for them, truthfully. I mean, you could build creative in-house if you're not a designy person, which I think is like one of the largest uh thoughts that people have. Like they believe this myth and this idea that, hey, I have to have an eye for design and I have to know how to build good creative, which is true, but how you build it is changing so drastically with AI that what Dylan's gonna walk through is something that we teach all of our clients. And we rolled this out almost two months ago with a huge program update with for all of our coaching programs and all the training material. And it was so cool that when we released this process, now the process we teach is Dylan's adding a layer to the process that we teach. So if you're a client listening right now, please keep listening. But taking that the the process that we taught was so cool because I just remember, I mean, for the last two months, every week people are posting in the community, holy cow, I followed the AI process and I could never build good creatives in Canva. And suddenly these things are cranking at two, three times the performance I've ever had before. And I could pump out two to three times more volume of creative. And so I say that because if you're listening and you think, oh, AI prompt, this is scary stuff. I don't know how to do it. I work with the designer, or I I I I'm not a designing person. This will solve that. And it did it for a client who's in person here. Yep. Uh we held a mastermind two and a half, yeah, how a month. Actually, it was a month ago.

SPEAKER_00 2:47

Yeah, about a month.

SPEAKER_01 2:48

A month ago, we had a client, we had uh 25 clients fly in and they spent three days with us here at at e-commerce alley headquarters, and we got into a lot of stuff. And one of these clients, would you say he would classify himself as like a non-designy client?

SPEAKER_00 3:00

Oh, 100%. They're their biggest thing, and I kept telling them this like guys, the thing that's holding you back the most is the fact that you're coming from Amazon, which they were doing great on Amazon, but they wanted to try and get out of the Amazon cesspool, I think is how they look at it. And the problem for them was they're coming from a marketplace where people are searching for they sell napkins. So they're searching for napkins. Okay. Well, if you sell really good napkins, which they do, well, it makes it really easy to place high on Amazon. So they don't really have to worry about convincing people to buy napkins because people will just buy the napkins when they search napkins on Amazon. But now that they're moving to Meta and D2C, it's kind of a hurdle for them to get over on like, okay, how do I convince people that they need to actually buy this product? And then even more so is like, okay, now they have to learn the whole creative process, which has really been a struggle for them over a few months until we had this conversation.

SPEAKER_01 4:03

So and on a note, by the way, when anyone who's listening, when you hear napkins, this is not like the cheap napkins you get at like a fast food restaurant or something. These are like nice essentially cloth napkins. They're not cloth, but they're essentially cloth. They're disposable, but they feel like cloth. Like these are for like if you want to impress people and uh your house guests that you have coming over, or you're hosting a party, or even like weddings and stuff like that, or like restaurants that really want to stand out, these are the type of napkins, which by the way, I love that. Yeah, we've never worked with someone who sells luxury disposable napkins. And I love working in these like abstract industries. So it's like you, I mean, I don't know what the multiplier is on how much more they charge that versus a regular napkin, but it's gotta be a lot of pretty substantial. It's gotta be like 10x a regular napkin cost. So I just love when people come with unique products like that that you're like, well, you can't, these would be considered a luxury napkin, I would say.

SPEAKER_00 4:56

For sure, for sure. So yeah, they were here for the event about a month ago. And if by the way, if you haven't heard us talk about this event, we believe it was probably one of our best events to date. We had great sessions, we had great people here. I think there were a lot of breakthroughs even in the room. And then I think there were a lot of new people that attended this event that we haven't really gotten to meet in person, which was really cool. Getting to meet I think they were all, yeah, they were all clients. So getting to meet all of our clients in person, you know, when we're living in a Zoom world. So we had sessions on Wednesday and Thursday, uh, but then we invited everyone to stay on Friday for like a work remote day, which was probably my favorite day. I don't know about you, but everyone.

SPEAKER_01 5:37

I was surprised almost everybody stayed.

SPEAKER_00 5:38

Yeah, I mean, it was it was so fun just to have everyone in the office working together. You ended up basically leading another session impromptu.

SPEAKER_01 5:46

I was in the boardroom just with like 15 people, and we're just talking hiring, firing, teams, scaling. It became like a third or a fourth day or something. Basically, basically amazing.

SPEAKER_00 5:54

And while you were doing that, I was having this conversation with this client. So I was kind of actually hoping to have this conversation while he was here because I felt like just some dedicated time going really, really deep on this, we could, we could experience a breakthrough. And and I think we did. So the first thing that I ended up doing with them was trying to understand how they are using AI right now. And by the way, like Josh said, if this scares you, you hear AI prompting or whatever, this is a this is a client who kind of felt the same way, where like, I don't really know how to use AI. I had already somewhat had a conversation with them on how to use AI to prompt and and build these creatives and and come up with ideas. Um, and it just wasn't quite clicking yet. And and it wasn't until this conversation that I figured out why. So initially I was like, okay, what is how are you using AI? How are you prompting the AI to actually output something? And they're like, well, I'll give it my website and I'll give them, give it maybe some reviews, I'll give it a few things. And and and essentially what it came down to is they weren't giving it enough information, which I think is something that a lot of people stumble into. They think, well, I have AI, I don't have to give it that much information. It's ultra smart. But essentially, what the way you have to think about it is like, yes, the AI is is very smart, but it's only as smart as you allow it to be in the thing you're asking it to do. So if you ask it to be an expert in disposable cloth napkins, well, it's gonna generalize disposable cloth napkins and not necessarily know your disposable cloth napkins or your brand.

SPEAKER_01 7:37

So I would say I would say that that's a really common mistake people make with AI in general. It's like you let AI do all of your thinking for you, but if you just do 10% more thinking than everybody else, you're gonna get a hundred percent better performance or better outputs. I kind of think like like first of all, AI is trained to give everybody it's trained in generics. Yes, if you just go in and say, hey, I need here's my website, we sell napkins, I need you to build me 10 creatives. It's gonna literally go to the most generic things, and then every other person that's trying to build similar products is gonna say the exact same thing. But when you go in, you think crazy concept in an AI world now. You think for 10 to 20 percent more with specifics and you hash out the thinking first, use it to brainstorm. But think first, and then you give it these inputs, and the outputs are not just like one or two times better. I mean, you're talking these things are like 10 times better output than if you don't.

SPEAKER_00 8:32

Yep. And and really what it comes down to is kind of like you said, it's it's trained in generics, but it won't use generics if you don't, if you give it more than generics. But the problem is people don't do that. And especially like if you give them if you give it your website, it's like, okay, I will see what I can scrape across this multi-page website. Like it's not going to get the deep information that you think it's going to get. So I call this the blah blah blah prompt. That's where we're gonna start. This is basically going to be the, you know, you could call it the way to think, but really it's not even that. It's like, you know, the elevator pitch, it's the opposite of that. I want you to give your AI, which I used Claude for this, as many pieces of information that you can about your brand. It's going to feel asinine. It's gonna feel so stupid because you're just gonna sit there and type. And I'm pretty sure I typed paragraphs to Claude just over and over and over and over. And I had the the client standing right next to me, and I was like, more, give me more, give me more, give me more. Who do you sell? What do you so I here's here's some of the things that I that I said what you sell, who they sell it to, what the person buying it will use it for. I gave 10 reasons for that. Deeply explain the product over 10 pieces of information, what it is, the USP, such as the ink doesn't run, non-toxic, eco-label certified, durable, custom designs, what they feel like. Like describe it to the AI, what this napkin feels like, what they smell like. Maybe not the napkins for that. Maybe if you're it's a candle, what does it smell like? Where can they be used? Kitchen, office, bedroom, bathroom, etc. And how they could be used in those places. You know, in the kitchen, you could hang it over your stove, in you could hang it over your sink. In the office, it could be a nice touch for your employees. They get to see, they get to feel these nice napkins instead of a z thing in the bathroom. Hey, you talking about us? Well, I just like that we have them in our bathroom now. I think they're I think they're nice. They are nice. In the in the bathroom, you could use them to impress your house. Like there's so many different ways that you could use them. Put it all in this blah, blah, blah prompt. Then I explained why the customer would buy this product because it's also like, okay, I could I can I can tell you what the product is, but that doesn't mean that the AI will just be able to know why someone would actually buy it. So, what is the problem that they've experienced with other products? What is the problem they've experienced with maybe even cloth napkins? Like not paper napkins to these napkins, but like cloth napkins to these napkins. You can explain multiple different problems they've experienced. What are they trying to solve for these napkins? It's it's maybe the way their house feels. We've learned now after doing this that maybe it's just they don't like doing laundry, they don't like washing the towel in their bathroom. Specific situations that the customer has been in where they wish they had this product, which another example for that was like an impromptu dinner party. You know, people are just like, hey, let's go over to your house. You're like, oh, okay, now I need to go wash the towel, or you're gonna use my dirty towel that I used this morning. It's kind of wet, you know. Yeah, like yeah, the bathroom, yeah. So all these tangible things you're putting into this prompt, I ended up giving it 14 reasons why a customer would use the product, all very different. I feel like people might hear that. So so far I've said I've given 14 reasons the customer would use it. I gave 10 reasons the person would buy it, and 10 pieces of deep information about the product. And you could probably even go deeper than this. But this took probably 30 minutes to to throw this prompt together. Um, now I was kind of prompting the client to tell me the things to prompt the AI. So it was a little bit longer than just spewing it out of my brain. Or if you whisper flowed it. Or you could whisper flow it, yes. But it it this is like the the foundation that you need to build your AI prompt on. And then I gave it proof how many have been sold, how old the company is, large brands that use the same product because they do have some enterprise customers and some customer reviews that we were able to pull off of Amazon. And so I'm gonna read you this prompt really fast. And this is just a piece of it. I I so sorry, all of that was was the first part. You got to give it foundation on your brand. Then I gave it some education on meta because another, I think, something that people get kind of caught up on is assuming that the AI is a meta ads expert, which it is not. It is not a meta ads expert. So you say blah, blah, blah. Even if you do all of that correctly, then you say, make me ads. It's it's still not going to be good. You have to educate the AI on what you know about ads. And if you've been listening to this podcast for any length of time, you probably already know more about ads than than Claude does. So I would here's what I just put in the end of this prompt. After I gave it all of that, here's what I put in for the ad education. I'm gonna create three ad sets, all around, and then an ICP, all advertising a sample pack. People will choose four sets at a discounted price and also get a free gift. So I let it know what offer we're running to. I want you to create a prompt that I can give to Chat GPT to generate an image for the ad. Here are some things to keep in mind for the ad. Here's the key. The ad needs to be very clearly, very clearly tell a tangible story that the user reads and says, Oh my gosh, this is me. I need this. The headline needs to talk specifically to the user. They should feel like they're reading the creative, that it's having a conversation with them. It should answer their questions, explain the product clearly, make it obvious what's being sold, and create a no-brainer situation to click on the ad. The headline on the ad should be the hook, the thing that catches their attention. Then additional text on the ad should essentially be objection handling on why they need this. Keep in mind when people see this, they're not actively thinking about buying my product. They need to be convinced that this will change their life and help them attain their dream outcome. I want to create three ad prompts for each of these themes, ensure that the creatives are very diverse, and Meta's algorithm sees each of them as completely different ads, talking to different people about different situations. So I literally tried to educate the AI. So all of this is one prompt, like I said, multiple paragraphs in this one prompt just to say, I'm gonna give you everything that I can. And then I also will add at the very end, ask me any questions before moving on. Because it will typically still ask if you put that, it just will ask you a question. But it makes it think about what question it could ask you after you've given it all of this stuff. Maybe it's just clarification, like, hey, you mentioned some enterprise brands. Are you allowed to use this? And the answer was no, but it kind of helps you frame maybe how you could position the product.

SPEAKER_01 15:07

I would add to this too, because like essentially what you did is I mean, you just created or built you just built a creative brief. Exactly. Right? So this is like the like this is part of the game. Uh this is where people also get lazy. Like if you were working with any kind of creator, any kind of designer, any kind of team that's writing scripts or whatever, you would have to give them something. You'd have to brief them on what you want, and you could even say four to five ratio dimensions, like you could give them all all the things that you would have to give to a creative team to do the thing, just make sure that this is part of the process. And by the way, once you've done this once, you templatize it. And then it's like great, you could build a cloud skill on this, and then it'll know to pull that in and know that hey, when you do this, make it diverse, you know, whatever. Yeah. And so I just think that this is where most people get super duper lazy and they're just like, okay, make me the ads. It's like, no, no, no, no. Like, literally brief this as if it's a it's a creative or creator or like a design team. Yeah, you have to do a little bit more work.

SPEAKER_00 16:11

Yeah, because if you feel like you go to Chat GPT, or maybe this is the way you feel is like you've gone to Chat GPT, you try to generate images, you're like, these suck. It's like everyone's talking about how great AI creatives are, and I try to go do it and they suck. It's probably because you didn't do the legwork first to actually get it to not suck.

SPEAKER_01 16:31

You see, like, uh, I don't know if that you've seen this in I've seen this in my feed a whole ton, like people in our community, we're I'm friends with a ton of teachers, and they always like they joke about like, oh, every teacher builds a poster or whatever, and everybody's flyers or so AI flyers. And it's like, yeah, that's because people that don't know AI, obviously, we easily breed this stuff. But people go into ChatGPT and they learn I could say, build me a flyer, and here's the information. And they're like, Oh, it looks amazing. But you know what? Because you literally just gave it the most generic information ever, it just builds it like everybody else. It's so like you see all these like I saw this one yesterday, a police department had this thing going. So they did it, and it's just yeah, people are just hammering the AI-ness in it because they don't know what they're doing. Yeah, they were a police department.

SPEAKER_00 17:12

It's actually funny. I was talking to this is a little tangent, but lunch a few days ago with my parents, I was talking to them about this, and we were at a restaurant, and every flyer in the restaurant was AI generated. And they were actually able to tell that they were AI generated. But then I started uh telling them about a post that I'd seen about how, and this is super tangent, but it's really funny, how people are like, okay, so it was you use m-dashes, which now means that you're writing with AI, but now the people have completely removed M-dashes from their writing. If you don't have M-dashes, now it seems like you were writing with AI because you told the AI to remove the M-dashes. So now we have to go back to using M-dashes to make it look like it's not AI that is AI. Does that make sense? And they were like, What's an M-dash? was like the first thing they asked. And I was like, Okay, so they still don't even know some of the so there is like a little bit of that where you don't like if you're not deep in AI, you won't know. But I think that also goes to show if you're not deep in AI in your business, then you won't know how to actually get this output. So that's the first thing. Here's the second thing, something that you may have noticed in what I just said is I told Claude, I was using Claude because that's my favorite thinker AI. I told Claude to generate the prompts for Chat GPT because there's there's like it's kind of like they speak their own language, right? AI speaks their own language, and they speak to each other really well because they understand how they think. And we have only just started to learn AI language. Now, this is all hypothetical, but you can think about it that way because an AI will do a better job talking to an AI than you will. Because if you tell it what you're trying to do, Claude, give me a prompt for Chat GPT, you are going to get a better prompt than if you were to just do it. 100x. 100x better. Especially specific. Especially if you start with all of this foundational stuff too about your business. Yeah. Because if you go to Chat GPT and you say, Can you make me some creative? Here's my product. You're you're not going to get a good output. And just like you said, if you did that with a creative agency, if you went and hired a creative agency and you just showed them your product and said, Make me some creative for this, you're probably not going to get a good creative. And that's essentially what you're doing with ChatGPT. So you can't assume that Claude and ChatGPT are all knowing. That's like the that's essentially what all of this has been about. You can't assume that the AI is all knowing. But if you give it the right starting point, it kind of can be, which is why I use this prompt because the more information you give it, the more the better outputs you'll get. And for me, it's just easier to just sit there and type and type and type and type and type and type and type. Instead of thinking about, okay, what do I have to say? How do I structure this? How do I say? Just type everything you can about your brand, your ICP, all that kind of stuff. So once Claude gives you the chat GPT outputs, if you stopped there, you copied, pasted them, and put them into Chat GPT, saw what they generated, they would already probably be better than anything you've ever generated from ChatGPT. But I want to take this a step further. And by the way, I did that. I did that with the client standing right beside me. Generated, I believe, nine, nine images. By the way, one one more thing here. We generated nine and I only used six. I think that's a big thing that people probably also do. They think, oh, well, if AI generated this ad, it must be good. I have to use it. I have to put it in my campaign. No, please don't do that. Because just because A, just because it was generated by AI doesn't mean it's instantly the best ad you could ever have. I generated nine of them and only six of them were actually good. You don't want to just run with stuff because AI generated, because you could run into the situation where Meta will put all your budget into the one that you knew was crappy, but for some reason it puts all the budget there, and then you're like, well, shoot, I shouldn't have ran this in the first place. So you need to still use your marketing brain in between the actual creative generation and the launching to say, maybe I shouldn't run this one and be okay with that. Maybe go generate a few more if you really want to run with nine. So that's the first thing. But then here's the thing. If you just stopped there, you'd be fine. I want to take it a step further and talk about really, really, really getting diversification. Because that prompting thing will give you diversification, but this will get you even further. Because here's the thing like Josh said, the AI is trained in generalities. So even with the best prompting, you're probably still going to get an aspect of generalness because you haven't given it any outside thinking on how to get outside of that generality. So here's what I did. With the With the client standing beside me, I went onto Atria, which is our favorite for this, and just started pulling random creatives. It wasn't in their space, it wasn't even necessarily e-commerce. It wasn't, I was just pulling creatives because I wasn't looking necessarily for, you know, oh, I should take this creative and make it say this for my brand. I was just looking for creative diversification. That's it.

SPEAKER_01 22:25

So and if anybody doesn't know what HRA is, by the way, it is an ad inspiration platform. It's kind of like meta-ads library, but a hundred times better. It's where you could filter by industry, by static, video, carousel. You could filter by length running, number of impressions. You can filter and find all kinds of inspiration on ads. So when Dylan's going to grab inspiration and reference images for what he's talking about, Atria is our number one go-to tool. If you want to check it out, go to ecommerceally.com slash atria, and we'll include a link in the description below just so we get a little credit for that. But it's our number one tool. We use it every single day. We recommend it to all of our clients. It's just, it is a paid version of MedAd's library, a hundred times better than that.

SPEAKER_00 23:08

Yeah, it really is. So, so I'm talking about visual diversity, but you may be asking, what is visual diversity? We've talked about it on the podcast before, but I want to give you some examples of just what to look at when you go to Atria. What do you need to pull? Because once again, you're not pulling, oh, this would be perfect. If I took this ad and I just changed this headline and changed these bullet points to my brand, this ad as it is would be perfect. That's not what we're looking for. We're looking for what just what can we give ChatGPT to actually create diversity? Here's some ideas. Traditional advertising ads, you know, like a billboard or magazines and news posts, UGC ads that could be founder related, uh founder-based, customer-based, b-roll, yapping, before and after ads, vertical before and after, horizontal before and after, the after blurred out. You show the before, but the after is blurred out, and you have call outs, us versus them, unique messaging visuals like post-it notes, a cardboard sign, a notes doc, like in your in your Apple notes, or an iPhone message conversation, story images, like a creator posted it on Instagram. Like they use the Instagram editor feature to make this ad. It bold and bright with big colors, big headlines, lots of USPs in a list, or simple and subtle, soft colors, let the product shine itself, minimal sales language, call outs with bullets, lines pointing to features drawn on the image, ugly ads like Microsoft Paint, Crayon drawing, whiteboard drawing, diverse video hooks, hook with a face, hook in a car, doing a funny hook, a clickbait hook. So that's just like kind of the start of visual diversity. And the reason I just blew through that is because it's not about finding one of each of these. It's just about finding diversity. Just go find and scroll and say, oh, this one looks different than everything else. This one looks different than everything else. This one looks different than everything else. Pull all those together in a folder on your desktop, give it to Claude and say, Claude, I want you to look at these creatives and determine which of these creative styles would match the ads you've already generated. So you've already generated nine ads. Which of these creative styles do you think would match this? It doesn't, you don't have to figure that out. Allow Claude to do that and say, you pick. By the way, I'm gonna recommend a few brands on Atria. If you want to go start on Atria, here are a few brands that I'd recommend jumping on right away. Oppo Pop, Magic Spoon, Ridge, Mellow Sleep, Comfort, Nuzzle, Hydro Light, Calmy, True Classic, and Eight Sleep. Those are just where I would start, and those are a lot of the brands that I pulled for this exercise that I did with this client.

SPEAKER_01 25:46

That's the beauty, by the way, of using AI is you could go gather inspiration. And I encourage you and gather inspiration from adjacent or completely irrelevant industries from yours. Yep. We do this all the time for our stuff. We'll go grab e-commerce inspiration for the e-commerce alley stuff. We're not selling e-commerce products, but go gather inspiration. Don't do what everybody else in your industry is doing. Yes, you could pull competitor stuff, but I would much rather go and find, like you mentioned, Ridge. Yep. Ridge is very different than eight sleep. Yep. You know, so like go gather inspiration that's very diverse as well, and you'll get different styles.

SPEAKER_00 26:19

Yep. So here's what I did. I just created a folder with all of these diverse creatives that I pulled together from Atria. I gave it to Claude with this exact prompt. Here are some ads that I want to use as inspiration. Create a new markdown file where each of these ads are paired with a message slash ICP slash angle that would best suit the style of the creative. I want a prompt to give to ChatGPT for each of these. I will also give Chat GPT the original creative. However, I don't want these to be an exact copy paste, only inspiration. So then what I do, Claude goes through, spits out all of the, like, okay, I looked at this ad and it would match really well with this ICP that I've already identified from the stuff that we did previously. So it'll go through and do all of that. Then you take that, you give it to ChatGPT, like I said, with the with the inspiration image that you already gave it. Because once again, even if you just use the Claude prompt that it spits out, and you don't give it that inspiration image, it will still revert back to generalities. If you give it the inspiration, you're going to get a little bit more diverse of creative than if you just use the prompt that Claude spit out. So that's kind of the next step that you can take. Now, here's the thing. Once again, some of these that I generated after pulling the inspiration, after pulling all of this, some of them still weren't good. And I didn't use them because you don't have to use them. Just because AI made a thing doesn't mean that thing is like the golden goose, golden egg thing. It's it's not. You are still smart enough to look at it and say, hmm, maybe I shouldn't run this one, or maybe I should change this one, or something like that. Instantly, though. So this brand that that we're talking about has been stuck in essentially the same creative with a headline and a subheadline at the bottom. And that was about it. Just by doing this, it it opened up the possibilities to create some of the most diverse creative I've ever seen this brand have. And it I'm gonna pull up their their breezeway and just talk through some of the the numbers because the the change that I've seen is is massive. Last thing, I just want to give a few more AI notes and additional thoughts here before before we do that, just to give you some of my some of the thoughts that I had, like technical on the AI. So first off, I am I prefer Chat GPT for image generation. That's my my preference. A, because I think a lot of people already have a chat GPT account. It's a really low barrier. I know Higgs field is a big, a big one as well, but ChatGPT, a lot of people already have and already use. It's the lowest barrier to entry. If you haven't started doing this, just start there because that's the easiest starting point. Uh start a new chat for each image you're generating. If if you are trying to generate an image and then you're like, okay, great, and then you give it the other prompt in that same chat, and then you generate another image, that is going to also lead to generalities. Starting new chat for every image you generate. You can also open multiple chat GPT windows and generate multiple at a time if you want to do that. Keep in mind, this is a big thing. AI is not an ads expert. You probably, if if you've been listening to this podcast, you probably know more than it does. So you need to distill your knowledge of meta ads, the algorithm, your brand, all of that into a prompt so that the output thinks about it the way that you would think about it. Because, like I said, you probably already know more than the AI generally. Once you actually give it a direction, it can probably get smarter, but you already know more, you need to use that. And then AI cannot do all the work for you. A lot of people think they can, like I said, throw their website in, say, make me 10 creatives, and that's how you end up with a bunch, a batch of AI creatives that look the same. Here's the thing, and this is not just regarding creative, by the way. This is for any time that you want to use AI. You can't just say, How should I, what offer should I do for my brand, or how should I change my website? Because you're just gonna get generalities. This is across the board. I'm getting a phone call here. This is across the board, just the better way to use AI. Don't assume that it's just smart. You have to give it that that prompt. And then one thing, this is the last thing, ethical dilemma of AI. I I've heard clients say, I don't know how I feel about using AI creative, right? It's uh my my product is home handmade. I don't want to use AI Creative to show my handmade product. Or you use AI to design something and you are a designer and you look at it and you're like, I could have done a lot better than that. But the thing there is if you don't want to use AI Creative, which I'm not saying you shouldn't, I do think that you should use AI Creative, but if for some reason you don't want to, you can still use this process to inspire you to go out and do your own photo shoot that's diverse. It could give you an idea to take into Photoshop and improve. You could even just you generate something and then you work with ChatGPT a little bit to improve the thing that it generated. I think everyone should do this. It doesn't mean it has to be your final output, and I think for most people it should, but it doesn't have to be your final output. But even if you just use AI as inspiration versus going out and just trying to think of it yourself, you're already going to have a huge upper hand on a majority of brands out there. So that's just kind of a final thing there on conversations that I've heard people say on I don't want to use AI. So I want to jump in and just show you what this could look like in an actual ad account. Because for me, this was a brand that I've been coaching for a few months now, and it was this conversation that turned the tide for them. And I'm just really excited to share what their account looks like. So a few campaigns that they've launched that they've spent spent thousands on had a.98 ROAS, a 0.98 ROAS that spent 4,000 and 3,000. And it was working through, okay, what message works, what ICP are we going for, what audience, what working through all these things. Now, I would say it was not just this creative process that I just outlined that caused now a 2.04 ROAS in their most recent campaign. It's not just that, it was all these other campaigns that they learned from that stacked on top, and then it was the cherry on top for the creative part of this. But their biggest thing was the cost per click, trying to get their cost per click. I'm looking at their their whole all their campaigns here. I see a $2.22. By the way, cost per click should be like $1.50.

SPEAKER_01 33:04

Or less, yeah.

SPEAKER_00 33:05

Or less. $2.22, $2.33, $6.91, $3.90. There are a few that are like $1.86, $1.27, $1.34, but the ROAS was all terrible for that.

SPEAKER_01 33:18

And these are these are past campaigns.

SPEAKER_00 33:20

Yeah, these are all past campaigns. And then the newest campaign that they launched with this process, I said, I don't want you to just take the creatives because I sent them the creatives that we generated for them. I was like, I don't want you to just take these either. I want you to go generate even more and push it even further than than what I just gave you. And the campaign that they launched was is now at a 2.04 ROS since its lifetime, which I believe is less. Yeah, it's just under two weeks old. And it's at a 2.08, sorry, 2.08 ROS at a dollar and five cent cost per click. And by the way, their break-even is a 1.69 ROS. So their profit.

SPEAKER_01 33:57

High uh repeat business, high LTV. So for them, breaking even if the break-even ROS is 1.69 or let's say 1.7, and they're at a 2.08, that's a huge win because their repeat business is so high.

SPEAKER_00 34:09

Yep. So this was just for me, it was really cool to see something finally clicked. So I'm hoping that for you, this podcast episode might help something click for you. If you've been struggling with creative diversity, maybe you hear people talking about that, but you don't actually know what it means. Hopefully, this episode will help you try to get closer to true creative diversity for Meta. And maybe these are some of those results you'll you'll be able to start to see.

SPEAKER_01 34:35

Cool. Hey, thanks, Dolan, for running that. Leave that. I appreciate that. Everybody, listen to this, go do this, and we will see you in the next episode.