I Found a BETTER Way to Test Facebook Ad Creatives transcript

If you're testing Facebook ad creatives, there's a good chance you're doing it wrong. And I don't mean just a little wrong. I mean fundamentally, structurally, completely wrong, wasting hundreds, if not thousands of dollars every single day on your ads. Across the brands we work with at The Moonlight I have spent millions of dollars across thousands of creatives. And I was testing the way that basically everybody teaches.

The way that seems logical, but still leaves so much money on the table and never actually maximizes your most powerful ads in your account. So, in this video, I'm going to show you exactly why the most common creative testing [music] methods are fundamentally broken. And then I'm going to show you step-by-step the best new method that we are rolling out across all the brands that we work with at The Moonlight right now.

Now, there are basically three very common testing methods that basically everyone is using right now. And if you're using one of these three, don't feel bad. I did this for years and I use these methods myself. It's not that they're just bad, though. They're specifically bad with the latest gem and Andromeda updates that shifted the landscape completely. The first and definitely [clears throat] most popular method is to test using better known as ad set budget optimization.

This is where you put all your new testing ads into an ad set budget optimization campaign. You force manual amounts of spend onto those creatives to then satisfy your creative itch, making sure that every single ad gets a set amount of spend before you either pause it down or bring it somewhere else. The second method that's grown in popularity, but is still not working at all, is at the ad level. It is the new creative testing function and it still forces manual intervention by making you select exact budgets you want to test per ad in the ad creative testing function.

The idea here is you could test a bunch of ads in one ad unit, but the actual reality is you can't actually do much once those ads are tested and you wind up self-selecting how much budget you want to put towards each new ad. And then the third common method, one of which I was on board with for a very long time, is to just introduce your new creatives into a CBO system, and then just let them go. Allow the algorithm to take full control.

Don't increase spend, don't decrease spend, just let them rip. This is still functional, but completely different than what we're doing right now. Now, all of these methods are logical. They seem like the right way to do it, but there's a fundamental problem with each one of these methods that most people don't realize. Let's start with the biggest flaw of the setup. Now, most people like the setup because you could force spend against any ad set that you want.

We could basically guarantee that certain ads or certain ad sets get spend because we spent money producing the creative. It sounds nice, but the reality is we are not smart compared to this algorithm. We just need to be the guiding factor for this algorithm. So, when it comes to the main issue is that certain ad sets outperform other ad sets, and we as human operators simply cannot A, react fast enough, or B, aren't actually spending the money in the right spot.

Facebook attribution doesn't show you every touch, every interaction, every engagement point of every single person. All it's doing in here is showing you the most critical touch according to Facebook. And in most cases, that's the last touch. So, it doesn't necessarily care about the first interaction they had with your ad if that person clicked on another ad later. That means impressions, people, views can get shared through every single ad set in your platform.

And because of that, what winds up happening is you start to overspend in the wrong places and underspend in the wrong places. So, for example, this is a client that before they started to work with us were using all I've narrowed down to the two ad sets that are now paused because we've changed to a full CBO M3 system, but I just want you to take a look at the state that it used to be in. Their number one spending ad set had a 1.1 return on ad spend.

Number two had a 1.5. Number four had a 1.8. Number five had a 2.5. Why on earth are we spending the most amount of money in number one when it is literally driving the worst return on ad spend of anything that has scale. So, the problem with an system is not that you're going to spend on the new creatives. You are. You're going to test those creatives. They're going to successfully be tested, but you will eat so much of your spend.

For example, let's say you were spending a thousand dollars per day and you launched a new round of creative. You're going to have to spend a dedicated amount to that creative in an fashion. That's going to force you to pull spend away from those winners and push them onto the new ads. And even if you're not pulling, say, away from those top ads, you're still adding budget to support those new ads. It has to come from somewhere.

And in most cases, new ads have a lower hit rate than already scaled ads. So, fundamentally moves your resources to the wrong place. Now, you might say, "But what if I find a winner? Isn't that worth it?" It absolutely is worth it, but your hit rate will not be as high as the algorithm doing it on its own, and you're paying a premium to find winning ads. And this brings us to the second method that has gained popularity in just the last few months.

It is the creative testing feature inside the ad level. So, if you go to your ad, you're going to see at the very bottom, just before tracking, creative testing. It's going to allow you to compare five different versions of your creative in a test that ensures delivery to new test ads. We do not want insured delivery. This is a negative, not a positive. If we click set up this test, what we're basically telling it is how many test ads we want to create, anywhere from two to five.

Let's just put in five here. How much budget we should go to these test ads per day, how long this test should run, and then how do we want to actually measure our performance. If you're ever going to do this, please at least put purchase here. Don't measure by anything else. Now, I'm just using a very low example here. We can get as aggressive as we want as possible on this, but what's going to happen is the spend is going to evenly distribute.

And just like even distribution is bad. You are creating cost centers in your business. You do yourself operating the ad account like this are literally wasting money. The algorithm is smarter. Don't force creative testing. You don't want to pull cash from the top or inject new cash when that cash can go to the better performing ads that Meta identifies as true winners. And the third method that is no longer optimal is testing in the classic CBO structure without any control.

Now, this is a structure that I myself have taught for the last 2 years, and things change, I let everyone know. And by the way, if you want to be the first to know when things actually change in terms of structure, strategy, what's actually going on in Facebook, then go to school.com/facebook or click the link down in the description to join our school community. Right now, I answer questions every single day from over 350 different brands.

If you join today, you could literally ask me a question right now and get a reply from me in less than a day. Also, you could join the next live Q&A which happens every single week. Again, link is down in the description if you want to join. The price is never going to be this low. So, the way that I used to teach this is every time you launch new creatives, you launch them into a new broad pack. Now, whether you had four creatives, two creatives, five creatives, it doesn't really matter.

You just launch new creatives into new packs, and you allow the algorithm to figure it out. Now, in practice, this worked well for the last 2 years. The prospecting campaign would normally distribute the budget to the best performing ads, and when you launched new ads, let's say this bottom pack was a new round of creative, it would start a little bit slow, but then eventually it would scale up as long as those ads were good enough.

However, we saw a change after Andromeda. We saw new ads hardly getting any spend at all, and this means our system needs to change. So, what's the better way? Here's how I manage every single account right now. We call this M1 of our M4 method. And the core reason this works is because it minimizes your wasted ad spend, maximizes your spend on your better performing ads, but still allows ads to be tested with just enough budget so that they can actually spend and actually get conversions.

It allows ads to ramp up very quickly if they're successful, and it still allows your bottom ads that are not going to be successful to get a little bit of run, a little bit of delivery, a little bit of spend to at least be tested. Overall, this minimizes your wasted spend and maximizes your upside. Let me show you exactly how to set this up in your own accounts. So, we're going to be setting up the M4 structure. Now, this structure is just going to focus on prospecting cuz that's where we test our ads.

So, the key difference between M3 and M4 is that we have set minimum budgets in place, and we are grouping things together by both avatar and concept. So, instead of just allowing all of our packs to run free and the creatives to be tested at will, we are now forcing spend to new tests. And the way that we're doing this is we're setting 7-day minimum budgets on all of our ad sets. I'll show you how to set that up in a moment.

The next thing we're also doing is our creative is being grouped by both concept and avatar. You always want to group by avatar and concept together. Without the grouping of avatar and concept, your targeting is a little off. This is something we've noticed since July of 2025. It's the biggest change, and the shift comes at the exact time of the meta Andromeda updates. So, let's jump into the ad platform. I'm going to show you exactly how to set this up.

First off, we're creating a prospecting CBO campaign. If you already have one, then you're not going to need to take this step. This is your core scaling prospecting and new acquisition campaign all built into one. Our bid strategy is going to be campaign budget and we're going to set our budget to whatever our daily spend is. In this case, I'm just going to set it to $1,000. Everything else can be left exactly the same.

We're then going to go to our ad set level. Let's make sure all of our settings are set up properly. Conversion location is website. We're going to use maximize the number of conversions for this brand. We're then going to make sure our data set is set up, our conversion event is set up properly. Our attribution model is going to be standard 7-day click, 1-day engage, 1-day view. In this case, we're leaving our audience broad.

We're just going to put in our exclusions that we would normally exclude. So, everything from our retargeting campaigns and everything from our retention campaigns. If you're not sure what to exclude, first make sure these are at least targeted in other campaigns before you exclude them. And second, I would generally exclude site visitors 30 days, add to cart 90 days, and anyone who's purchased from me in the past. These could easily be targeted in other campaigns, so we don't over index and over spend on them right here.

Now, after we set up this ad set, we're calling this pack number one. Now, I'm just going to ask you to imagine that this pack right here has already been running. It's been in your account. What do you do when you go to launch new ads? First, we're just going to duplicate all of the settings from our existing pack. Of course, if you have ads within here, you're deleting those ads and you're putting pack number two. So, you're getting the new ads into here.

Once we launch this new round of creative, we're then going to our ad set spending limits right here. We're clicking set a minimum. We're clicking use dollar value and we're setting our average daily minimum to 1x our cost per acquisition. In this case, I'm just going to imagine that it is $100. We are only keeping this ad set spending limit on for 7 days. The reason 7 days is so important is because it doesn't allow you to overspend on stacked concepts.

If you don't set the ad set minimum, what usually happens is an ad spends very, very little until it eventually starts to scale up and hit its maximum spend threshold. Now, what happens in the opposite when you do set a minimum is you shortcut how much you're spending right off the jump. So, what happens in these first 7 days and only 7 days is you're actually feeding the ads with more signals, more conversions. The ad then does one of two things.

If it is not successful and the end of the 7 days, marked by my line right here, occurs, it's going to die very quickly and spend very little. Or, on the flip side, what we want to happen is this ad is going to spend much quicker up to its threshold and then eventually plateau. Now, plateauing is completely fine. We're not solving for the plateau. What we are solving for is this giant surface area of untapped potential.

All of this area in here represents cash that you are missing out on when you are not testing properly. And you are trading the cash in here for the possible testing in here. This right here is your testing budget. This right here is your upside. The upside is significantly larger than the testing budget. We care about the upside. At this point, we've implemented this strategy for the majority of the accounts that we work with at the Moon Liters.

The M4 method is fully adapted to what's actually working today. But, here's the thing. Testing only matters if you actually have good creative. So, I've recently made a video on how you can create a ton of high-quality AI, UGC, and static images, literally hundreds at scale. This right now, I believe, is the fastest and cheapest way to actually create high-quality creative basically non-stop. So, if you want to see that video and actually tie these two together, then I'll link it right here.

And if not, I will see you in the next one. And I hope as always you got a ton of value out of this. See you soon.