My Exact Creative Testing Method For Meta Ads transcript

[music] What is the ideal way to test new creative in your ad account? That is the subject of today's episode of the Andrew Ferris podcast. Thanks so much for joining me again and I want to tell you what I think about the best way to organize your creative testing in your Facebook ads account is from a structural media buying perspective. This is something people debate a lot or actually sometimes don't debate. They have a bunch of assumptions about the right way to do this and uh and they they just kind of go through the same method over and over again.

But I'm going to argue today that you should not have separate creative testing campaigns. You should not worry about making sure that your new creative runs and gets some spend. Instead, you should build your creative from the outset so that new creative launches against your existing best performing creative. And you should not worry whether it spends or not. I'm going to tell you why. [music] All right, this is uh maybe odd timing for this episode because maybe you're thinking, Andrew, it's in the first week of November.

Isn't it time for some Black Friday content? And my answer is maybe. But there's lots of Black Friday content out there. It's all really good. And if you want Black Friday content from me, my opinion has changed 0% this year compared to last about the best way to run Black Friday for your brand. And therefore, what you ought to do is go back and listen to older episodes of this show from last year. I did five episodes in one week instead of doing one Black Friday episode.

Each one of them is about five minutes long. It is very very short interviews with the likes of Sean Frank from Ridge, Taylor Holiday, Shireen Alber, like Jess Bachmann was on there. It's a very smart people. I asked each of them, give me your number one Black Friday tip. And so, you can go find those BFCM episodes. If you just go to my show and search BFCM, you'll find them. There's five of them. It will not take you long to listen to and they'll be helpful for you.

Go listen to what they said then. Go do what they said then. There's my Black Friday content. All right. So, there we go. Got that out of the way. All right. So, let's talk instead about creative testing, which is relevant to Black Friday. It's relevant in the sense that you are going to want to launch ads for Black Friday. You're going to launch new ads, and you're going to want to see how those perform relative to your other ads.

And the crucial thing here that is the key distinction that I want to make about how to test new creative in an ad account, how to launch new creative in an ad account is this. Do not have a separate creative testing campaign in your ad account. There's such a strange thing where people have this idea that they have to have a creative testing campaign and a scaling campaign and they think about or a bunch of scaling campaigns and they think about these as two very different things.

And there's all kinds of reasons why I think that's a bad idea. And let me just tell you though why people think it's a good idea first and then I'll and then I'll pick apart why I think it's wrong. Okay. The main thing people are concerned about is that they actually spend money on their new creative testing ads to give those ads a quote unquote fair shot. Right? So the idea is you want to launch ads not against your current winners because your current winners will get sort of an unfair advantage against your new ads.

All right? And that is considered a problem for many people. I'll just flag that now and tell you I don't think that's a problem at all actually. But many people think that's a problem. And so what they want to do is they they run their new ads in a creative testing campaign separate from their current best performing ads. And the idea is if an ad gets some spend and gets some, you know, some performance that you think is at or above your target, then you can then go launch that ad against your current winners.

And let's critique the best version of this. Let's not make a straw man here. Part of the consideration there is that your current winners have a whole bunch of social proof and data built into the creative. basically according to sort of Meta's read on your creative and therefore the you want to launch new ads separate from that in a separate campaign because uh you want to give them you want to give those ads a chance to generate some of that uh social proof and some of those things so that Meta kind of has some information before you go launch it against your best ads.

And that to me is the best argument in favor of this strategy. the idea that that ads will have actually ads that some ads have a serious potential to generate value for your business, but that you will actually never give them a chance to generate that value. And the reason you won't give them a chance is because they they just don't have the initial run that your your ads want, right? So, they actually would outperform or perform alongside your best performing ads, create some increased spend for you, create some increased scale for you, except that um they're just not generating the social proof that you need.

That's probably the best argument in favor of this that I have heard. I actually I mean this is one of those things where people sort of say here's how the auction works. Here's how the algorithm works that I just don't really know for sure, right? I've heard some of those things people say, "Oh, Meta says this happens." I've never looked seriously at it. And and so there you go. All right. So, so there's that. Okay.

So, that might be the best reason to do it. But let me tell you from the outset why I think that's a bad idea. There's there's a couple of problems here. And and so let's just take them in order. Number one, the number one problem here is that spending money on new ads is spending money. Okay, so this is part of it. There is cost here. And if your new ads are not actually as good as your other ads, spending money on those ads probably, by the way, detracts from the value of your other ads uh of of your current winners because uh reaching additional incremental people costs money.

Uh and so so if you're trying if that ad takes up audience, eats up audience that eats up spend that would have otherwise gone to a better ad, there's that problem right away. But even besides that kind of more technical issue, there is a baseline thing here, which is that you don't know how your ad will perform. Even if you, you know, are not somebody who thinks you can never tell how an ad is going to do in advance, you know, whatever.

Like even if you're pretty confident that your ad is is decent, you don't know how it will do really in the grand scheme of things relative to your other ads. And if your new ads are not going to perform as well, then there's a problem here, right? Which is that that you are going to spend dollars and this is a cost center in your business to test that new creative. And so eliminating cost centers is one of the ways you generate profit, right?

There's two ways to generate profit. It's more sales and less costs. That's it. Those are the only two ways, right? Of course, you have to do those. Obviously, if you have if you generate more sales while generating even more costs, then you haven't generated any value. And that's part of the point here. Testing creative costs money. And so one of the great advantages of not running a creative testing campaign is that you are likely to spend less bad money.

This whole notion that you want to spend money on new ads, I think is fundamentally flawed. Okay, in that respect because I don't care about spending money on new ads unless those ads are going to perform. Uh now I'm assuming in this also that you are running your ads with cost controls and therefore you should assume that if Facebook spends on your ad then Facebook is going to be able to spend that money at or or better than the return that you want.

If that is what you are doing and every time you are launching new creative you are launching in a costcontrolled campaign. I have no real problem with you launching all of your new creative in new campaigns. In fact, one of the great arguments in favor of cost controls is that you could sort of launch endless campaigns and I've looked at ad accounts run by, you know, CTC and Kinship people who are big-time cost control folks um like me.

And while I take a more consolidated approach, this this element of things is sort of on the margins to me. I don't care that much like uh which one of these two you do. I I have a a leaning on this issue and I I build mine in a more consolidated way than this. I've looked at ad accounts from CTC and from kinship where like there are hundreds [snorts] of campaigns and they don't think about they don't worry at all about sort of creating testing creative testing campaigns versus scaling campaigns whatever all they worry about is just launching new ads at their cost cap or at their bid cap or at their minimum ROS target or whatever they just launch their new ads there and if it spends that's it and Meta can decide which of those campaigns to spend on and and there you go.

That actually could be boom turn it off end of episode if you want to launch creative like that. That is for sure a better way to do it than uh to launch creative with auto budgets, forcing spend etc. Like I have never actually seen this create a problem of not being able to get enough spend through including to new creative because you know you're sort of not giving new creative a chance. Okay. So eliminate the cost center in your business by making it so that meta cannot spend on the ad unless it's going to perform at or under your target including relative to your other ads.

So, cost center number one. Argument number one, cost center. Number two, and this is another thing that I just think is like so obvious when you think about it and and just nonetheless, the logic here is crazy to me. Okay, how much money should you spend on a creative testing ad before you decide it is a winner? Okay, if you want statistical significance, you need to spend so much money. Okay, so yeah, just like way more than you think.

I mean, this also depends on your average order value and all those things. lower AOV could matter I guess to some degree here but like if you are auto bidding creative new creative because you want to see like how it's like what happens if you get a two to one rowass on five purchases does that count does that is that a winner like that that's ridiculous and absurd and you know better than to think that statistical significance happens in such small samples when you're launching new creative you are constantly in small sample land you are never getting out of small sample land and therefore the idea of waiting to determine whether or not an ad is a winner and then moving it to a scaling campaign based on that is crazy to me because you are going to have to spend so much money to do that effectively.

Which means the whole idea undergurting this concept is sort of flawed from a testing perspective to begin with. You just you just have no idea if an ad is going to work at any scale because it generated uh a good return on $1,000 or 500 bucks or whatever it is relative to thing. And of course, the more you the higher your AOV, the more of a problem this is. Like, you absolutely cannot and should not try to guess which ads are going to win by running it in an auto bid uh in an auto bid creative testing campaign if you have to spend hundreds of dollars for every purchase.

You're just going to blow through money doing that. Again, part of the great advantage of cost controls is that it suppresses spend when Meta's algorithm thinks that when Meta's uh machine learning thinks it is not going to generate uh value at a return that is acceptable to you relative to your targets. Okay, so that's that's argument number two. So argument number one is it is a cost center in your business. Argument number two is you're not doing this scientifically anyway.

You are not being conscious of of the principles that you would need to be conscious of to know if an ad is a winner or not. Okay, those two are the top two reasons why. Now, those are arguments against running a creative testing campaign. Okay, now let me give you some arguments for putting everything together or just running everything, launching everything with cost caps into new creative. Okay, number one, and most importantly, you don't actually care how your ad does in isolation.

What you care about is how your ad does relative to the other available options. Okay. What I mean is if you autobid on a new ad and it spends a certain amount of money and and has some amount of performance, that is fine as far as it goes, but that's not actually what you're trying to do. What you're actually trying to do is make your whole account spend more money. It's you're trying to get the whole thing to generate a better return.

And therefore, by definition, what you want is Meta to distribute every dollar that you spend in the most efficient way possible because that's the actual pathway to the most scale. This is why cost controls are a scale mechanism because the more that your ad account can be built around efficiently distributing your dollars, the more marginal profit you have on every dollar you spend and therefore the more money you can spend because you're squeezing out as much profit as possible, right?

And and so that point is part of is part of the reason why I think it's so helpful to instead think about launching all of your creative into campaigns where it can actually run against other options. And therefore in turn as I sort of flagged earlier it is a value to your business that meta essentially gives uh a head start to ads that are already performing because that will make it so that you if that's even true by the way but meta meta will therefore make it so that your new ads actually have to outperform that creative enough to where it's really worth coming up in the algorithm along the way.

So or or in the in the dist distribution of spend against those other creatives. You you want creative you want Meta to keep spending on the stuff that is already winning. And you only want Meta to spend more if it's very confident that it can beat that other creative. You you you don't want to take spend away from winning creative if it's not very likely that that new creative will beat it. And therefore uh it's good it's good that Meta gives dollars to the stuff that's already performing at your best.

So that's argument number one. You don't care about the spend in isolation. You care about the spend against other possible options and that creates the most efficiency from there. Argument number two is it's a management problem. Like if you are launching stuff into a separate creative campaign, now you need a human guessing and wondering and this kind of goes along with the scientific testing method point that I was making earlier, right?

Uh suddenly you have to keep an eye on all of these creatives and figure out, you know, which one should go where and okay, is this one a winner, is this one a loser, etc. and then move things into different campaigns, all those kinds of things. Take your hands off the machine. Okay? Instead of doing that, don't try to find the winners in that respect. Uh, and move the winners. Instead, just let Meta spend and let Meta spend on the ads that are most likely to win.

Like, and so this is again combined with a cost control strategy in instead of trying to guess which winners there are, put a cost control on the adsets, put a cost control on the campaigns if it's ASC, and just let Meta spend. just let Meta spend if it's going to work and then walk away from it. You don't even have to worry about it anymore. Now, people will say, "Oh, well then how are you going to ever know which which is truly working and understand why things are working?" And the answer is every however many days go look at what ads are spending the most money and and observe what you can observe about those ads and learn from them what you can learn.

Don't worry about testing and learning in this really hyperspecific way. Now, I'll put an asterisk on that and say go back and listen to my episode with Jess Bachmann about his approach to creative testing strategy. I think there's really helpful thinking in there about uh how to approach some engagement related metrics. But the thing I would put on layer onto that conversation about how to use engagement related metrics to iterate new creative ideas is I would only care about those metrics at some scale.

So like if if your ad has spent 500 bucks, I don't really care what the engagement metrics are. I I only really care if your ad is spending meaningful amounts of money and then understanding what is making those ads win and seeing if I can tweak them or make new iterations or whatever it is. I don't think that it's that helpful to to have this sort of super clear hypothesis about why every ad is winning and why every ad is losing and okay like I just think there's a level at which you're there's too many variables and you're not really going to understand it all the way.

Uh and so don't even try. So it's sort of like this combined approach. Learn from the data you have to some degree, but trust that spend is the ultimate piece of data in an ad account. And if the ad is spending there, that's the the strongest signal you can get and and go and learn from that uh as best as you can. Okay. So, so that's number two, management. You don't want somebody having to make decisions about how to do this and and inserting human decision-m into this process.

Let the machine learning decide it is smarter than you. It is way better than you predicting the future. [music] What in the world are you doing? hiring teammates, adding to your team and not looking in the Philippines for incredible e-commerce talent. People with deep experience in e-commerce in DTOC across every part [music] of the business like marketing, supply chain, logistics, customer service, whatever, all over the business at a much lower cost than you would have those same people cost you in the US and yet at a great price in the Philippines.

So, you're going to get the best of the best talent that you can get there. That is what is available to you with my friends at more staffing. You can go visit more staffing at more.co because more staffing will help you recruit, hire, train, onboard, everything, coach, give ongoing support to incredibly uh talented people from the Philippines to add to your e-commerce business across every element of the business. If you are thinking about adding team members in your business right now, stop what you are doing and think about just take a call with more staffing.

Go to morning now.co. Tell them that I sent you. You should consider doing this. If your hire that you make with more staffing doesn't make it for an entire year in your business. They will replace they will help you find a new uh hire at no cost. They'll help you find a new employee to take over that spot so that you have a guarantee that you'll find somebody that you like for a year. That is an incredible value. Again, it is all going to come at a much lower cost than doing this whole thing in the US.

If you are thinking like me about DTOC and maintaining a low opex as a percentage of your revenue, this is a core part of how you can do that. Well, do that with incredible talent in the Philippines by going to more.co, my friends at more staffing. Go do it now. Now, two counterarguments that I get. Number one, what about when the machine learning is wrong? Okay. Okay. So, this is a cost controls critique. This is a creative testing critique. uh you know how can and and we we'll come back to the the the biggest argument that people make in a second here which is how can Meta know if it doesn't spend.

Okay, so we'll leave that kind of part of it aside for a second. We'll come back to it. But counterargument number one, what happens if the machine learning is wrong about which ads to run next? And my answer to that question is what is your alternative? The machine learning is going to use a much better method to predict this first of all than you are going to ever be able to do ever. No chance. It has access to way more information.

And as I always say, right, its girlfriend never broke up with it that day. It never was too tired. It's never hung over. Like there's there's all these factors for why on top of just like human finitude. There's there's all of these factors as to why machine learning is better at this than you are. Okay. And secondly, and relatedly related to that point, you don't have an alternative. Okay. Yes, machine learning probably is wrong sometimes at predicting things.

Like I have no problem admitting the idea that m meta's machine learning is not perfect in its delivery of your ads. But the alternative is you. You're the alternative. And take the average there. The machine learning beats you. The machine learning is better than you. And therefore, the alternative to the machine learning is you. Okay? If that's the case, great. Get you out of the way. Let the machine learning win. Go from there.

Okay. So, just trust that it will work better. Especially the more complex the ad account gets, the more creative you introduce, the more things that you have to analyze. Okay. Secondly, related to that, there is this ma the number one piece of criticism that that my approach that I'm advocating for here suggest uh garers. The number one piece of criticism is what about when meta doesn't spend basically at all on my ads?

And here's what I will say about that. First of all, above all, if you if meta is not spending, it's because your ad probably doesn't work, including yes, when it is only spending a few cents per day. You say you will say, "How can Meta possibly know if it's only spending a few cents per day?" And my answer is I don't know, but it seems to. And I have never once, and I mean this, I've never once seen Meta spend so little on an ad like that only to have it uh over time by doing some different media buying method or just whatever else only to have it over over time become a true serious winning ad.

Like it might get a little more spend over time, might move from a few cents a day at some point to a couple hundred a day by doing some things. Fine. Okay. But that's not the kind of win I really care most about anyway. What I care most about is wins where there's a real meaningful difference made in my ad account. And I've never seen Meta suppress an ad like that that is really going to become a like top 30% of my spend ad.

Okay, it just doesn't happen. Meta I think because of the sheer number of user signals that Meta has access to, I believe Meta can figure this thing out really fast. I think Meta understands quickly whi which ads work and which ads don't work. In fact, even this whole notion about social proof, like Meta knows that a new ad doesn't have social proof. And if if it was best for the customer, right, they have a strong incentive here to do what's best for the customer so that we'll spend more money because they'll be more effective.

If that's the case, they would adjust the machine learning to give a little boost to something if if no machine learn if if no um engagement was going to really hurt your ad. just like realize that Meta has built this thing so that it can most efficiently distribute your dollars. And it doesn't mean it's perfect. I get that. I get that it's not perfect, but it's the best way to approach it overall. And again, it's I have never seen an ad that was like really really seriously suppressed in spend that actually turned out to be a huge massive winner if I'd have just pushed it spent.

And I've tried all kinds of other things like everybody says, right? I've launched things in separate campaigns. I've launched things with higher bid caps than my other ads that sort of disproportionately get some spend. I've tried all this different stuff and I've never found any of it matters. Okay, so in the end, here is the way that I would test your ads instead. Okay, what I would do is either launch every ad into a new campaign, okay, and run the sort of like I said, kinship CTC sort of approach where you just have no problem just campaign after campaign after campaign, run bidcaps, cost caps for all of them and just let it go.

That seems perfectly fine to me. Or what I tend to do is I tend to have one massive CBO, typically a BAU campaign still, so legacy targeting, non-ASC. I have not seen ASC make any positive difference yet for my for my accounts so far as I can tell. Not in any really meaningful way. And I just launch new adets into the same giant CBO at the same bid cap as long as the same products, the same AOVs as all of my other ads.

And over and over, what I see happen is if that ad is any kind of a winner, Facebook identifies it pretty quickly and rockets it to the top of my spend. It happens all the time when I launch ads that win and that are going to beat other ads that I'm spending on. I do not have a problem. I just haven't seen the problem really of ads not getting spend that otherwise would have been a winners or all of my new ads get suppressed all the time.

If your ad is not spending, it's probably not because of a media buying error or an auction error or a machine learning optimization error, whatever. If your ad is not spending, it's probably because your ad doesn't work and you need to go back to the drawing board and do more creative. It's not a media buying problem, it's a creative problem. And therefore, launch your creative against your winners, either in that same CBO, like I said, in new ASC campaigns if you need to launch them with cost controls.

Um, and let the best creative rise to the top. Meta is insanely good at it. It will take your decision-making process out of it. It will reduce cost of all the testing massively in your business and allow you to launch lots and lots and lots of creative with all kinds of freedom knowing that as you launch all that creative, the best ads will win. [music] All right, there we have it. That's my episode for today. I wonder if you agree, if you disagree.

I get a lot of push back when I bring this kind of thing up on Twitter. People have strong opinions about this issue. So there we go. Next week on the show, I've got Cody Blaer from Jones Road Beauty. Uh Cody told me he hadn't done a pod for a little while. He was eager, excited to do it. And it's because and we're going to talk more media buying, more metab buying because Cody is spending a whole lot of money. He's doing more of it now with bidcaps than he ever used to before.

And I said, "You got to come on Cody and you got to tell people what changed your mind and why it's working and how he has become a bid cap maxi like me." Uh I I don't know if he would use those words, but you'll have to listen next week [music] to to find out. It's going to be a really good episode. Make sure you um stay tuned. Make sure you subscribe so [music] that you get that episode. It's going to be really great.

He has spent a lot of money on Meta and knows what he's doing. Uh [music] so Cody's going to come on the show. Hey, if you want to reach out to me, do that and all the usual places, and JJ Ferris on Twitter or go to my website, ajfgrowth.com. There's now an email capture there where I'll give you some free resources that you should go get. There's good stuff there. Cohort forecasting, cash flow stuff, all kinds of things that [music] are going to help your business be better.

Just drop your email in there. I won't even send you an email probably, honestly. I don't even have like a newsletter or anything. Just just in case, just one day, who knows? Uh, afgrowth.com. Get all that stuff there. Also, email me at podcastfgrowth.com. And of course, rate, review, share with a friend. All those things are so useful, so helpful to me if you would be [music] willing to do those things. And don't forget to go visit my friends at more staffing to hire scale up your team with incredible talent from the Philippines at more.co.

Thanks so much for listening. I will see you next time. [music] All I know with my eyes closed and I [music] say that's all in the way they