Scale Your Facebook Ads With AI Creatives

Nicolas Korvessis shows how to scale Facebook ads cheaply by generating multiple creative variations with ChatGPT, then testing them across micro-audiences at low budgets. He uses Campaign Budget Optimization to move winners to $800/day, then pulls back to $150-200/day once Facebook learns the audience.
Chapters
In this video, I wanna talk about a few very simple ways that you can scale your winning ads without necessarily breaking the bank with a few new methods that we've been trying this year inside of some ad accounts that are spending between $100 to $1,000 a day. So if you get anything from this video, like and subscribe, let me know if you have any questions and let's get right into it. So overall, inside of Facebook, especially in 2025 and moving forward, creative is gonna be one of the most important things across the board. When it comes to scaling your ad campaigns, getting lower cost per conversions, lower cost per leads, whatever it is that you're chasing, having more creative is one of the best and the easiest ways to do that. Even if it's testing some simple variations of the hooks, the copy, the headline, or even the lead-in of the video itself. That's one of the easiest things you can do in your business is create a system where you can take a video, you can take an example of anything that you're using, and slightly tweak it to make multiple variations. I usually try to get at least 3 variations out of 1 creative. So every 1 creative to me is actually worth about 3 because I can change the hook, the lead-in, I can change the thumbnail, I can change the first half of the video to something totally different, et cetera, et cetera. So that's one thing is just kind of like a precursor to this because I'm gonna show some media buying tricks that you can do with existing creative, but It really does come down to having consistent creative across the board and making sure that you're leveraging it to actually get the best results by just pumping out creative. And before I show some of these media buying tricks, if you are still kind of confused on what creative to make, how to make creative consistently, maybe you don't really have a team, well, ChatGPT gives you some really powerful tools, and if you prompt things the right way, you can make some killer creatives through ChatGPT. I posted this in the Predictable Profits Mastermind, which is a free community that you can join below, that showcased some creatives that we made with ChatGPT that worked really, really well. What I'll do is I'll include the prompt that I used to make some of these creatives in the description of this video. So check it out. And if you want more prompts like this, more AI tools, we're working on a lot of them and we're posting them first inside the Mastermind community. Like I said, it's totally free. You should see the link below in the description to join. With that being said, let's get into some of the media buying tricks. Okay, so the first thing that you can do is something that we've done for a very long time, and it is micro audience tests. Micro audience tests are extremely easy, very low leverage, or I'm sorry, very high leverage with a lot of low effort into it. And it's super simple to find some quick winners with just some simple audience tests. So I'll show you kind of how this works. What we did with an ad set in here. So you'll see in this one right here, we've got this ad set that's performing well over the month of April, right? 40, 43 sales, $35 a purchase. It spent $1,500. It's working. It's doing exactly what I want it to do. Now, yes, the first thing I'm going to do is start testing maybe some different copy, some audiences with it, and then seeing whatever that creative is and figuring out how can I make more like that creative. But after that point, and while the team and I'm working on that, what's something high leverage that's low effort that I can do today to just get more results? So the easiest thing you can do, and this is exactly what we did with all these different duplicates here, is we duplicated it. I don't know, I, it's up to you based on budgets, how much you want to be spending, but we duplicated it about 10 to 12 times. Not all of these are exact duplicates, But most of them are. And what you end up doing, and what I end up finding, is if you duplicate them at a fairly low budget— most of these started off between $11 to $15 a day— and the only thing that was tested or that was changed was the audience. So what we'll usually do is we'll jump into the ad set, I'll duplicate it 10 times, for example, at a low budget, and I'll just change one, maybe two minor elements inside the audience specifically. So for example, if the winner was a lookalike, I'd maybe open the age range more, or I'd do women only, or I would remove the lookalike and keep the age the same. Or you could do something else where you're taking the lookalike out of it and just adding one single interest. This is probably the most effective one that I've seen is doing multiple changes like this, duplicating it 10 times, have 3 of those changes be a different size of a lookalike audience if you had a lookalike, have another 3 be different interests, whether it's grouped by a couple interests per each one or just one single interest, and so on and so forth. And you could have a broad one, et cetera. These are really easy. And I mean, out of those 10 or so that we duplicated, and we really didn't spend, some of these are still spending to be honest, but we really didn't spend much to find these. We found one, 2, 3, maybe 4, but we found 3 that were really solid. So about 3 out of 10, if that, were really good. I think originally we really just duplicated these right here, these 7, maybe these 8. So about 3 out of 7, 3 out of 8 were winners, and, or at least potential to be a good winner. 2 sales at $30, 1 sale at $15, 1 at $59 is okay, 2 at $27, great. Like that's exactly what I was looking for. I can turn off the losers, cut my losses on those, continue to drill in and bump up the budgets on these ones. It's a really simple, like I said, a very high leverage, low effort that you're putting into it, but it's a very easy way to get some quick wins. So that's the first big one I wanna show you. I wanna show you one more simple thing that you can do to really scale your ad campaigns without needing to make a ton of creative just yet. Let's get into that next. The next easy thing that you can do is simply duplicating your winning ads into a CBO campaign. Now you'll want to start these off at actually a little bit of a higher budget than you're comfortable with. And the reason why is because it gives Facebook some very quick wins and some initial conditions that it can learn off of quickly, and then you can lower it down to your budget. So we did this with these 2 ad sets. We duplicated 7 of our winners in here. Started the daily budget at about $800 a day, let it do its thing for the first day, didn't fully spend the $800, but once it learned and once it figured out exactly what it was doing, I dropped the budget back down to around $150, $200 a day. And you can see the results here, 134 purchases for this month, about almost $6,000 in spend. And this is something that worked really well for us. Now, obviously with a CBO, the one drawback is Facebook's gonna start to put most of the budget in just a handful of ad sets. So even though this one was pretty solid, it wasn't putting a lot of the budget here over since like March 30th when I launched this. It was putting most of them into here, which is okay. It's not about— when the CBO, you're not trying to test and figure out what's winning. You're just taking winners and letting Facebook do something different with them. Now, with all that being said, before I end off the video, I want to talk about creatives and specifically leveraging AI to make creatives. Once again, you can get access to everything that I'm about to show you. If you join the Predictable Profits Mastermind for free, this specific prompt and custom tools site will also be in the description. And I want to start with first off using this prompt here. I tried making a custom GPT with this, but unfortunately ChatGPT, the new way that it makes images in GPT is a lot better than using, I forget what it's called. Instead of using DALL-E. So that's unfortunate thing is that a custom GPT, you have to use DALL-E, whereas this uses GPT's new way to make creative. So first things first is dropping in this prompt, which will be in the description. And what I really like about this is you can use examples of creatives to make ones that you are wanting to make. So the first step then when you plug this prompt in, this is gonna ask you for an example creative. So I'll upload that now. Okay, so I got a super simple creative here that I'll, that I'll upload just as an example. I'll give it to it, and then what it does is it starts by giving a visual analysis of the creative. It looks at the hook, the promise, the CTA. It breaks down the colors. So if you are big into branding and want to make sure that your creative is nailing the branding element, it makes sure to clarify that, give some focus areas, and then it'll ask you what elements or messaging do you want in these variations. So I'll just tell it here, I'll say make it more premium and for sure keep the brand idea as well as the hooks. And then it'll go through and it'll actually give you 3 specific concepts. And then you can choose for this prompt to either make all of them Or just one of them or two of them or whatever you want. I'll go ahead and have it make three of them. So I'll say, please make all three. And the way to make these prompts is super simple. You just got to give it the right inputs and do some testing yourself. But a lot of the work's been done for you here. I'll go ahead and pause the video. And once these are done, I'll unpause and show you how easy it is to make a lot of creatives with AI. Okay, so it's almost done with the last image, but I'll at least show these two. And AI isn't gonna necessarily be perfect, but even if it gives you more ideas, or you can take these, drop it in into Canva, make a couple of edits, like the saying 30% twice is just weird. Like I'll just block that out and put something else there or nothing. This one's cool. It's got some different colors. And then this one is seeming pretty similar, but it gives some good ideas. And these are some very easy variations that I could take if this is a winner to kind of uplevel it and take it a step further. It also does a great job of making humans and making everything look realistic. So. So give it a shot. Even if it gives you one extra ad variation per one creative, you're making one creative worth 2, which is super valuable at this time, especially with paid advertising. So I hope you enjoyed this video. Once again, be sure to like and subscribe if you got something out of it. Hopefully I'll see you in the next one. And if you need more resources, information, et cetera, join this group. It's totally free. We've got 519 members as of right now. Um, and check out the description for all these prompts and custom tools.










