Transcription
We use Claude to help. Over 300 clothing brands generate more than $100 million in revenue from Facebook ads over ten years in this industry. And I want to talk to you about what's actually happening, like right now, because the gap between the brains that are winning on Facebook ads and the ones that are bleeding money has never been wider.
The brain is losing. They're still doing it the old way: into stacking big caps, 12 different ad sets, manually trying to outsmart an algorithm that has more data than every person on their team combined. And here's the thing. It's not that they're bad at this. It's that the algorithm has changed. Meta has shifted more in the last 18 months than the previous five years, and most brands didn't get the memo.
The brains that figure out how to feed the algorithm better inputs right now are going to own their category. The ones that don't are going to keep wondering why their cost per purchase keeps climbing. Now, if you're spending money on Facebook ads and not seeing the return you expect, I can tell you right now it's not the inputs, it's not your budget.
So in this video, I'm going to show you exactly how we can use Claude across our entire Facebook ad process, from knowing who we're selling to, to script writing, to the creative, to building the campaign brief before we ever even open Ads Manager. And we used to do this before, but it was like a bunch of research, and it took a lot more time and a lot more effort than what we're doing now, which is great because you can move quicker. But I'm going to show you the exact prompts that we use live so that you can copy them today.
Now, stay to the end. This campaign build is worth it. Let me show you the end result before you get into how we built it. So this is a campaign right now, $4,800 in daily spend. Spent $136,000 in the last 30 days, which I know for some people isn't crazy, but I had a 5X ROAS. That's not too bad.
So this left-hand side over here, this is this is Claude. This right-hand side over here, this is Meta Ads, right? This is Ads Manager. Here's what happens when they work together, okay? We go back to Claude. A customer avatar document Claude built for a clothing brand in under two minutes. Now we have three documents that we use and have created. They are foundational. I don't care how much you use Claude or AI, if you don't have these inputs to create a foundation for your brand, it's going to be very difficult. Three ad sets, broad targeting, and the cost per purchase is down significantly. Okay, we've got a $25 cost per purchase. Average order value is about $8,590, so pretty decent. The average order value is about 101.25 ish, and the cost per purchase is down 40% in the last 30 days. That connection between what Claude produces and what performs inside of Ads Manager is exactly what this video is about and what is possible for you.
Now let me show you how to build it. Most clothing brands who are running Facebook ads are working too hard on the wrong things. Again, they're stacking the interests. They're testing bid caps. They're building 12 ad sets, hoping one sticks. They're watching videos, which I know this is an oxymoron, watching a YouTube video, they're watching a video on somebody who creates ads for generic brands versus specifically about apparel with actual results. The brains that are scaling the fastest right now are doing less of that, and using Claude to do most of what actually moves this needle.
Here is the three-part system. Part one: Know exactly who you're selling to. Claude builds the avatar doc that makes every other input sharper. So we have the struggling avatar document. You need to define who your struggling avatar is. You'll grab all this information. Okay, I would do this. So I copy this, I put it into Claude. I'm going to use this tool. So it's called Whisper. If you're not familiar, just WhisperFlow AI. And I'm going to talk, and then I'm going to give it the script. Okay. So I need you to act as a marketing expert, and I'm going to give you information about my struggling avatar document. I want you to take all of this information and the parts that are blank that have questions. I want you to ask those questions to me, and then I'm going to answer them for you so that I can get this input or get this output. Okay. So I post that, and then I'm going to paste. Okay. So I just pasted the avatar document. It says, "Before I start asking these questions, let me get one piece of context so my questions are properly targeted. Are you filling this out for yourself or for a client?" I said, "I'm filling it out for a made-up brand." Okay. So I said, "Let's build this from scratch for a made-up brand. Tell me about the brand. What do they sell?" So, "All right, so this is a high-end clothing brand for athletes, specifically runners. It is made from a very high-end cotton. It's really breathable. It's got a lot less issues with sweat and odor. And what do they sell?" Okay. So we sell running shorts, we sell tank tops, we sell hats. "And what transformation do these people help with?" So what our brand helps people with are those who are chafing, those who care about materials, those who have problems with wearing long shorts that are usually marathoners, maybe ultramarathon or trail runners who get so sweaty that their shorts are drenched. This actually is so moisture-wicking, it doesn't hold as much sweat. It keeps everything away from their legs so they don't chafe as much, just much better quality product. There we go. Put that information in here. And then again, there's a struggling avatar, a unique solution. I'll explain what my unique solution is, and then future pace here. We won't go into all of that. I just want to hit on the struggling avatar here. So it says, "Love it. This is a really clear niche, serious endurance runners who suffered through gear that fails them on the long efforts." Not now. Let me walk you through the document questions one section at a time. So then it would answer these questions. Okay. So again, this document is extremely important. But now I'm putting it into Claude and giving the answer faster, more efficiently. And I can use Whisper. So I'm going to do this last one for you as well. "What physical problems were they dealing with on runs?" All right. Yeah. So they were chafing. Their shorts were soaked. They had odor discomfort. They were the smelly guy or girl. "How did they feel emotionally about the gear?" They were more than anything just irritated. Yeah. Frustrated. Probably upset. "What were the deeper frustrations like?" Feeling like serious gear didn't exist for them or their level of running, or that they had to choose between performance and looking feeling good? That's the biggest one. They want gear that actually looks good. They have a higher income, they want something to look good, and the stuff that looks good is either made of plastic and just junk, or it is really low quality. And so it's great for going on a three-mile run, not for the marathon or the ultra. And so it was just missing out on that. Okay, give us some more information. You can see now how fast and how powerful Whisper AI is. Okay. So Claude builds the avatar doc that makes every other input sharper. Okay. So I'm just teaching this to all of it.
Part two: Creative that converts. Claude turns that avatar into an ad script, hooks, and copy your team can shoot tomorrow. Now, if you don't have a team, you can shoot it tomorrow, right? It just literally requires an iPhone. I mean, we're using a fancy schmancy camera here, but you could technically shoot in 4K high quality right here with an iPhone. So I have all this information, gave me this good stuff. I can answer more questions, but I want to talk about the structure here. And I could say this. Okay, so I'm doing this for a YouTube video. So I'm not going to answer this last part of the questions, which these are good, but I just want you to take information, run with it, and then help me write a script for social media. I want you to be a top social media content strategist, and I want you to write a video for me that's about 20 to 30 seconds long that I can use for Instagram. Pull inspiration for what's trending right now, what would be relevant for my audience, and hitting on these pain points. What I didn't share here is that we have a messaging framework on how to talk to your audience, and I didn't give it here, but I would. Okay, it's I would give that. But it's grabbing all this stuff for us, and we'll come back to it.
Part three: Structure the campaign to scale. Claude builds the brief before you ever open Ads Manager. Okay, so one of the brands that we work with is Any Means Necessary. And it was running so many different assets, I'm going to say 14, but it's probably completely different if you're stacking interests, inconsistent ROAS, getting completely. That's fine. Right. That's most brands. That's normal. Right. But what we did, we just simplified everything. We plugged Claude into all three layers for ourselves to make things easier. Three assets. Actually, I think we have a little bit more than that now based off of some new drops, but it's all broad. So age, gender, location, the cost per purchase is down significantly. A lot more revenue is up. But the biggest one here is that profitability is up because the ads are spinning more efficiently.
So let's start with the foundation. Right. So part one, the avatar doc. Okay. So part one, before you write a single and ask yourself one question, do you actually know who you're selling to? Okay. And that's what we are getting into here. And it wrote out this script. Not a vague idea, not a streetwear fan's 18 to 35, a specific person, right? We want a specific person. What do they do on a Tuesday morning? What do they want people to think of them when they walk into a room? What would make them stop scrolling and actually read your ad? Your audience should feel like your ad was made specifically for them. Hopefully, you feel like this video was made specifically for you because, in my mind, this was created for clothing brand owners. So I hope this is helpful for you. If you're not a clothing brand, that's fine. Hopefully, it's helpful for you too, but that's what we're creating content. That is how you scale. And Claude builds this document faster and sharper than any process that I've used literally in the last decade.
Now watch the top of this screen here. This is where I type the prompt, okay? And I say type. I use Whisper, right? That's where I speak the prompt. Or you can type the prompt and watch what Claude returns below it. Okay. Like all of this information just right here. Here is a prompt that I want you to copy. Right? I set the context first. "You are a direct response marketing strategist." Which we can do this right now. I'll do this and I'll say it out loud. So, "You are a direct response marketing strategist for a clothing brand. Here's what we sell, right? Here's our price point. Here's how we describe the brand in three words. Here are our top three best-selling products." Okay, so it knows all this information. And obviously, we made up the brand earlier, okay? And it knows that stuff. But then I want to ask you, right, "Build me a detailed customer avatar for this brand. Include age range, income, location, what platforms they use daily, what they watch, what they aspire to become, what frustrates them." Okay, you could do all of that, or you could just use this struggling avatar document that we already created, and it will ask you these questions. But this all comes back right with specific details: what they watch, what they're aspiring to become, what frustrates them about every other brand in this space. This is the document you hand to your creative team. This is what you reference when you're writing hooks. And this is what tells the Facebook algorithm, through your creative, exactly who to find. Without this document, you're guessing at all of these strategies. With it, every step gets sharper.
Now, real quick, if that avatar just changed how you think about your targeting, subscribe. Every week, I break down a new system and a new Claude workflow or a real brand result, just like this. You don't want to miss the next one, because the avatar doc is only valuable if it changes what you put in front of people. And in part two, I'm going to show you how Claude takes the document and turns it into ad scripts your team can shoot tomorrow.
So part two, you have an avatar doc. Right now, Claude turns it into ads. Most clothing brands' ad copy the same way. I mean, you've probably seen "elevated look," "premium." Someone takes a crack at it, right? It sounds like the brand, sort of. It gets approved, mostly, but it's, you know, it's first draft, and then it runs, and you wonder why the hook rate is super low. Basically, nobody watched past the first six seconds. When your copy comes from the avatar doc, the hook isn't a guess. It's a direct response to something your customer already feels like. Again, we went into the feeling part of this. This was the problem that they had. Watch the left of this screen, right? I'm pacing the Claude avatar doc in here first, then I add the prompt below it. Watch what Claude does with both.
Now, here's the prompt. Copy this. Okay, "Using this avatar, write me 30-second direct response video scripts for a Facebook ad. Each script needs a hook that speaks directly to the person's frustration in the first three seconds, a body that connects our product to what they're trying to become, and a call to action that creates urgency without being pushy. Writing hooks as if you're stopping someone mid-scroll." All right, hit enter. Three scripts, three different hook angles, all built from the same avatar. Okay. So it's pulling it up right now. This is awesome. "The fabric problem," "You're better," and "The smelly runner." These are good. This is your creative brief, okay? This hook is already written. The message is already dialed in. Hand this to whoever shoots your content and tell them to deliver it. And if it's you, and back to yourself, right? Read that back. Hand it to yourself. Read that back. That's not a template. That's not a swipe file. This is a line written for one specific person with one specific frustration, based on the avatar we built ten minutes ago.
So let's just read one of them here. So I like the "smelly" one because I think it's funny. "You know that runner? Maybe you've been that runner. It's not a hygiene thing. It's a fabric thing. Synthetic materials trap sweat and lock in odor. The longer the run, the worse it gets. Our premium cotton breathes differently. It pulls moisture away, resists odor buildup, and doesn't leave you drenched and embarrassed after a 20-miler." See, the 20-miler, good job like picking on an actual like terminology. "This is what gear looks like when it's actually built for runners who go the distance. Call to action: Limited stock. Tap below to grab yours before the next race build starts." Good. I'd probably make a couple of changes, but honestly, 7.5, 8 out of 10, okay? If you know the industry, probably make some changes. This is what stops the scroll, and your competitors are not building their hooks this way yet. It's good to be social proof, creative, real customers, real reaction. Claude writes the brief for what emotion to capture: product in context. Your product in the environment your customer actually lives in. And Claude tells you exactly what that environment looks like from the avatar: direct response. So someone on camera addressing the avatar's frustration directly, like right there, looking at them. And then Claude writes the script.
If you've made it this far, you're serious about this. So let me show you the part most people skip entirely because the avatar and the creative are only two-thirds of the system. The third piece is what actually determines whether you scale or stop or stall, and it happens before you even open up Ads Manager.
Part three: The campaign build. Most people open up Ads Manager and start building budget here, ad set there, paste in the copy, hope it works. We open Claude first. Now, here's prompt three. Watch this. I'm giving Claude everything we've built so far: the avatar doc, the scripts. I'm keeping it all inside of this one spot. And the offer, and all three go in together. Then I add this: "Based on the customer avatar, these ad scripts, and the offer, build me a Facebook ad campaign brief. And what would you include? The recommended campaign objective, sales you should do, ad set structure, how many creatives to test, what to look for in the first 72 hours, and what metrics tell us this is working or not working?"
Okay, now if you don't have background in ads, then this may be very helpful for you, but this prompt can give you a baseline here. I'll tell you right now, you should almost always go for sales, 99.999% of the time. I don't care about any other number except for purchases. I don't care about CPM. I don't care about cost per click. I don't care about any of those things, but as long as you got purchases. If you don't, then I start to look at those other metrics, okay? And so this hopefully gives us some context. All right, purchase, right? Thank goodness Claude backed me up here, right? So conversions, purchase event, always optimize for the first point down the funnel. Good job. Ad set: cold broad interest stack, right? It's telling you to do an interest again, retargeting. I'm not telling you to do those things. Claude doesn't know everything here. But that's why I wanted to share this with you in real time. Okay. Broad on each one of them. Your targeting is done by the creative. Having background or information about this is important. And that's why the video is here. Okay. So I would give it that information and say, "Hey, look, I'm going to create a campaign that's sales. I'm going to go broad targeting. And then I'm going to let the creative do the the targeting for me."
Now it has each one of these scripts. Static image, UGC, carousel. I love that. Love the dynamics. So does Andromeda, okay? But the campaign objective, the ad set structure, the creative testing planning, early signal metrics, what a good 72-hour read looks like, and what a bad one looks like. Now, 72 hours tells me what to watch. Don't make any decisions for 72 hours. That's just in general. Check the spend. Check for the purchases. Check which creative is pulling. Spend. Okay, now you open up Ads Manager with a plan and not a guess. And again, you have to run this through your own lens, okay? It is not perfect. They will tell you right here, AI can make mistakes, but that's why I'm telling you the answers here.
Here's what the structure looks like when Claude has informed every layer. Typically, we do broad. You can do an Advantage Plus, but typically we do broad. It's open targeting: age, gender, location. We suggest three ad sets, kind of max, depending on your budget, okay? Every budget is a little bit different. But if you've got $100, we can do an ad set or two. We can do three to five ads or so in each one of them. A good place to start. We don't do interest stacking. We do no bid caps. We don't want any of that fighting for delivery. Now, Claude messed up here and said, "Yes," and I need to kind of tell it, "Hey, don't suggest that." But your job is to not tell Facebook who your customer is through targeting options. Your job is to let Facebook know who your customer is through your creative. Claude, build that creative from a real avatar, and the algorithm finds that match. I mean, if any of you guys have been on YouTube or on Meta, and hopefully we're doing a decent job, we could always get better, you've seen one of our ads, and it's because we're targeting clothing brand owners.
Here's the part that nobody talks about. Structure isn't just about scaling. It's about protecting your margin. And when you fight the algorithm with bid caps, you're telling Facebook to find cheaper clicks, not better customers. Better customers spend more, buy in, leave reviews, they refer friends. And a slightly higher cost per purchase from the right customer is more profitable than a low cost per purchase from the wrong one. And I have to say this all the time. People always look at cost per purchase. Okay, I don't care about that nearly as much as acquisition cost to lifetime value. Am I going to make money back? To kind of Any Means Necessary, they had again, like 14 different ad sets, interest stacking, inconsistent ROAS. And we plugged Claude into all three layers: avatar, ad scripts, campaign brief, and restructured the whole campaign, and it just became much simpler so that we could actually scale up. Like I said, we're about three ad sets now. It's broad. Customer purchases down, right? Revenue is up. Not because we found a magic audience, because every input the algorithm received was sharper than anything they built manually. That's the algorithm shift I talked about at the start. The brands that are still doing it the old way are going to keep bleeding, and the brands that plug Claude into their inputs right now are going to own their category. 300 brands, more than $100 million in 40% lower cost per purchase, all from better input.
I put together a Claude Skill and Prompt Pack specifically for Facebook ads for clothing brands. If you want that, just comment the word Claude, and I will send it over to you. Most brands spend weeks testing ad copy before they land on something that converts. These three prompts compress that into an afternoon. Okay, so you have the avatar doc prompt, you have the ad script prompt, you have the campaign brief prompt, plus extras for writing hooks, testing angles, and reading your early data. Like I said, just comment the word Claude down below, and I'll send you the link.
And if you're running a clothing brand doing real revenue, and you want us to take a look at exactly where your Facebook ad system is breaking, we should talk. Just check out the link in the description down below. It's 45 minutes. It is specific to you, specific to clothing brands. It is free. However, you've got to go do it. Link in the description. The spots are limited. Go book one. Right now.
The brands that are still interest stacking and fighting the algorithm with bid caps, they're going to keep getting the same results. And the ones using Claude to build sharper inputs: the avatar, the scripts, the brief, are going to scale. YouTube knows what you need to go watch next. Go check out one of these two videos. I'll see you on the other side.