The world of meta ads is sometimes changing, always changing, tactics change over time as the platform changes. And I've got updated opinions on some of the things that are most important to me as a media buyer based off of what I'm hearing and learning from a combination of reps, the performance marketing summit and just what's happening in my own accounts. We need to do seven things that have sort of changed my mind about or that are the most updated parts of my media buying strategy for meta ads. This is going to be an incredibly tactical actionable episode. Then tell you exactly how many ads you should be putting in an ad set right now, things like that. So let's talk about it right now. All right, but I want to hit is like the things that are sort of changes of mind, sort of updates to the platform, some combination of those things, but it's a state of the union on what I think is the key way to handle meta media buying right now. And so the first thing I want to actually do is tell you what's not changing for me. Okay. What's not changing for me is the basic assumption that machines are better at making decisions about the future than I am. That is to say machine learning is a better probabilistic port forecaster of the future than I am and my baseline strategy as a media buyer is to get my mind out of the way for decision making about how best to distribute my ads instead. Let me just machine learning make those kinds of decisions for me and instead, but my focus on how products and creative generate value for my business and giving me the tools it needs to go in return. Distribute my ads in a way that generates that value back for my business. So that's that's a basic approach here. Okay. So I am trying as always to run everything that I possibly can through the lens of how do I get met and make more of the decisions about optimization for me and get myself out of the way. That's the baseline principle because on average, meta is going to win at that. It only increases the reality of that more all the time as there are more placements as you added. You know, the increasing adoption of reals is all of these different tools come out AI Gen creative all of these different things are all coming out. So I'm going to give you seven things though within that framework that I think help leverage meta ads within that, you know, while standing on that basic principle even better than before. This is going to be really, really tactical. Like I said, okay. So number one. 15 to 20 ads per ad set. Okay. For a long time, I have run non-ASC campaigns. Okay. And I've run them with four to six ads per ad set. And that's based off of old meta documentation. It's actually something I was never that certain of and never, you know, it's not something you probably seem to make a lot of content about people ask me about it all the time. How many ads in an ad set. I've never really known exactly what to say about this. But there's some old meta documentation saying four to six. Then they came five to seven. And, and you know, an ASE. It was many, many more. Why that tested ASE against my setup might be a you set up was basically the same as ASE running CBO campaigns where you're letting you're doing, you know, campaign budget optimization. So you give the campaign the budget, let the ad sets get distributed as metacase fit and that ended up, you know, broad targeting that. And it ended up mimicking a C I tested them head to head found that my my setup performed about the same as a C. And so there was no reason really to switch over to a C which I also find clunky and annoying to build with that said there's a major change that is happening in the platform, which is that ASE and and the older style BAU will call it campaigns have have our our our melding. Okay. There's no longer distinction between them. In fact, most counts, I think in all of my accounts now that's already the case. There is no such thing as an ASE campaign anymore. There's no such thing as a BAU campaign. There's just a campaign. And in this new setup, where ASE has sort of taken over everything, but with some distinctions that look a little bit more like the BAU setup. There, there is now a question about sort of like how do you live within this framework and I'm perfectly fine with that. I was never opposed to ASE. In fact, I like the ASE. So I'm not bothered by this setup, but by this change at all. But in the midst of that, one of the things that has become clear from my number of conversations I've had in a lot of different places is that it looks like the optimal number of ads per ad set at this point is somewhere between 15 and 20. And so I've seen a few different ways in which meta has signaled this from various conversations with people we've begun running everything this way. And I've seen a meaningful increase in in the really I should say decrease in volatility increase in stability in terms of daily spend. And it seems to me that that in that new world, okay. Where you are running 15 to 20 ads per ad set, you can still run CBO with multiple ad sets 15 to 20 ads per ad set works just fine that you end up getting a good amount of learning occurring to each ad set. I'm going to talk more about learning in the second, okay. But because it's another big big issue for me. But, but you you're able to see this go really well. I love this 15 to 20 ads per ad sets is easier to build with. It's much easier to manage less ad sets. All the way around. So I'm I'm grateful for it. Like I think it's a great change. I'm quite happy with the Meldy VSC and BAU in an easier to build with environment. I think it just makes it a lot easier as an advertiser to again sort of have less things to manage and again get yourself out of the way and accrue more purchases more conversions to each ad set, which makes it easier to analyze as well. So yeah. So I like this 15 to 20 ads per ad set. Still I run campaigns separated by products. So it's not one monster campaign necessarily. So I'm actually fine with that if you can, but typically separated out campaigns by product, which makes it easier to manage and easier to see what's going on. If you need to make any specific changes at the product level as inventory changes, things like that. That's all normal for me. That's been the same for well 15 to 20 ads per ad set is is a big part of that again all reflecting the combined of ASC and BAU into this new environment. Okay. So there's that. Number two, very much related to this is I've begun to care a lot about getting ad sets out of the learning phase. This is something I used to not care about actually at all. I would actively say I don't care about the learning phase and even met as documentation about this was not so much about the learning phase having volatile performance. You know, met as machine learning is really quick at establishing baselines of conversion optimization and there's just the basic problem of human behavior being fundamentally unpredictable in such a way that that like, you know, small amounts of spends are just going to show a wide range of potential results because you're constantly dealing with small samples. So. So in the learning phase, what a lot of people thought of as volatility and performance. I just think of as like volatility of small samples and small behaviors. And and in any case, I didn't really think it mattered that much, especially relative to the trade off of launching everything with a manual bid, which I'm still doing, you know, big caps and TRO still doing that across the board. I've got you know, boat loads of content about exactly the stuffers have to go through my past content and see how I'm approaching these kinds of things if you want to see those kinds of details, but. But but the learning phase was something I just didn't really worry that much about that's really changed for me again that's from a few different places, including conversations with reps, including metasome documentation about this metas push on this I tend to think that if meta is pushing on something it's in your best interest to listen. And this is one of those things that they have for a long time pushed on and multiple reps of mine have talked about it different times. You've got too many ads in the learning phase and the way this has been explained to me and what I've seen reflected in my add accounts is essentially the learning phase problem is actually just a problem sort of machine learning giving up on assets more than it is a problem of actual volatility of performance that if you get the assets out of learning you're more likely to make it so that you. So that meta sort of you know quote unquote feels more confident about the delivery of your ads and that's what you want and and in order to again get more stability in the delivery of your ads and that combined with the 15 to 20 ads per ad set thing is really nice because now if I can get 15 to 20 ads per ad set and get more purchases accrued to less ads that so this works really really well to combine that these two principles together because if I have let's just take 20 ads let's call it five ads per ad set I would probably done four to sit you know up to six before but. It's called something like that right so now I can put all 20 ads in one ad set and so spreading them across four ads that's and that means if the goal is 50 purchases per week to stay out of learning something like that then this much easier to do with one ad set them with four and so that you combine those principles and it gets a lot easier I think whether you're testing scaling whatever. To get all those together so I now care a lot about this this also means that as I'm switching over old ad sets that had you know. Five or six ads per ad set if they're out of learning i'm not touching them if they're if they're just out of learning i'm not going and updating them with a whole bunch more ads per ad set throughout learning i'm just going to leave it in there and just building on top of those now with ads that's with more ads in them over more time and if they. Do kick back into learning at some point then I might combine but otherwise i'm quite happy to get ads that's out of learning keep them out of learning and let them run okay so that's another element i'm actually now at a point where i'm building more and more to do this even sometimes. Turning off some ads which has been anathema to me in the past never because of performance okay i'm not turning off ads in order to like because an ad is spending and it's performing poorly i'm turning off an ad if it's just not spending very much at all if it has been an account for weeks and it's gotten ten fifteen twenty dollars and spend total and it's not showing any signs of picking up i'll turn it off to get more ads into less ad sets just so that they get less of the ad account in learning of course occasionally it might be a false negative if if it's an ad that I really really want. To to to like force meta to try to respond on there's a couple little things you can talk about i'm going to think i'm going to talk about look at my notes here. Yeah yeah i'll talk more about one little trick for that later that's actually really good it's the last thing i'm going to talk about here so so make sure you hang on for that part of it and by the way if you like this episode you should subscribe right now wherever you're watching listening because this is the highly tactical meta content that i'm trying to produce as often as i possibly can so okay so there's that so.
In any case, I care a lot about the learning phase at this point, really trying to make it so that if I'm whether it's test ads or scale ads or whatever, okay, however you frame it, I'm going to really try to keep ads out of, uh, ads that's out of learning, helps with stability, helps with, um, consistency of delivery. Okay. Number three, and you might have just heard me a little something that you've never heard me say before. In fact, you've seen me suggest is a very bad idea. And that is, I am starting to use creative testing campaigns. And this is like the biggest shift. It's going to sound like the biggest shift from my past strategy, um, that I have introduced in a long time. Um, it's actually not a very big shift, okay? Here's my problem with creative testing campaigns. My problem with creative testing campaigns is that when most people, and I say like 98% of people mean by that phrase is a campaign, uh, that you run on an automated budget and often, uh, at ad set level budgets where you are launching new ads, you're launching those ads without a manual bid, no cost cap, no bid cap, no to your ass target, forcing spend to the new ads, okay? Analyzing the performance of that spend based on purchase behavior and, uh, and then, uh, and sort of accepting the fact that many of your ads are going to perform under target. Okay. And that's the way that a lot of people do this. And then they're scale, they're taking ads from those test campaigns and moving them into scale campaigns from there. Um, now the thing that I hate about that and then I'm absolutely not changing on at this point is the idea of forcing spend to, to, to test ads. That and then, and then secondarily analyzing the performance of those based on purchase behavior. There are countless problems with that. Among other things, the hundreds of thousands of dollars that brands are wasting all the time by doing this. If you're still using gorgeous or zen desk for your customer service, help desk software for your ecommerce brand, what are you doing? It's time to switch to rich panel rich panel. First off promises a 30% savings guarantees it. If you switch over from one of those two pieces of software, if you go to rich panel, which means the only reason you should not switch to rich panel is if it is somehow worse or something like that. And it is not rich panel is very well loved by everybody I have talked to who has switched. I've got another client who's in the process of very likely switching over as well. And so, um, and so over and over people are saying the same thing, which is that switching over has saved money and also been a better experience. In fact, rich panel says that on average, uh, brand see a 30% reduction in customer service tickets because of how good their self service portal is driven by AI to help customers get the answers to the questions that they need without going through some big giant clunky help center, uh, with all these tabs and folders and forms and whatever it is. Uh, on top of that, rich panel is built AI first. It's a more recent entry into the customer service software space, which is a good thing because it came, it was built from the ground up with an AI first mentality. And if you're listening to or watching this episode of a meta ads tactics, you probably know the pain of moderating your comments on your ads, uh, like manually in some way, even with customer service help desk help. I've just never seen anybody do this well. Well, what rich panel offers is actually an AI assistant that go and do the interactions for you and doesn't incredible job. You can do it fully automated. You can do it just with help either way. And of course, there's all kinds of built in ways for, uh, rich panel to help you with your regular, uh, tickets and and making your reply, your response time faster. That's why customers on average rate brands that use rich panel very high in terms of the, um, satisfaction with customer service levels. There's just no indication that is anything other than the idea that rich panel is great customer service software built AI first, built easy to use all with a, um, all while being cheaper, more affordable than the other options in the space. It's just really, really good. You should at least get on a call with them. And on top of all of that rich panel promises to help you get transitioned to rich panel in two weeks time from your old software. They know you don't have time to waste. They know you can't deal with a long slow install process. So go check it out today. At least get on a call rich panel. Com slash a J F rich panel. Com slash a J F. You can look in the show notes for that as well to get a link to that page. Uh, go check it out today. I'm launching a brand soon. I'm starting with rich panel. You should think about it too. Brands are spending so much money, uh, wasting money on test ads that they could just launch with a manual bid. And if they launch them with a manual bid, uh, they, they would, they would have, uh, every chance of success. Meta can see what's happening for those ads. It will amplify the ads. If there's signs of success, it will suppress them if there's not. It will not waste money on ads that are performing under target. For whatever reason brands have it in mind that just by putting the word creative testing on a campaign, it's an excuse to blow a whole bunch of money on under performing ads. Um, on top of that, the idea of analyzing the performance of those ads based off of purchase behavior on extremely small sample sizes is terrible. People do not understand small sample size bias, they are small sample size volatility, uh, small sample size noise, I should say. And therefore they constantly turn off ads that could be winners and they, and they try to like scale ads that, um, very well could be losers, uh, because they're looking and seeing an ad that spent for three days, got five purchases and they're like, oh, it's a winner in the scale. Okay. I just like that's happening all the time. It's a giant waste of time. It's hugely inefficient. There's countless problems with it. It's putting human decision making in it. I hate it. I hate it. I hate it. It's wasting a bunch of money. One of the biggest cost centers in all of D to C still. So none of my opinion about that has changed. Okay. Um, so I am still launching anytime I launch a creative testing campaign. I am launching it with the same manual bid that I have a scaling campaign with. Okay. So let's say I have product day. I'll have a product day scale campaign. In fact, usually I'll have to the T row S version and a bid cap version. Okay. Same ads in both of those two campaigns. And then all the test campaign, uh, often launched with the bid cap, sometimes T row S, uh, and, and the test campaign, uh, has, has new ads in it and the manual bid in each of those campaigns is the same. Okay. I'm maybe at the same target and all of them. So this has actually been how kinship has bought media for a very long time. Uh, they, they do something similar with their launch new ads and new campaigns with the same manual bid. And I've talked to Taylor. I could say about this a bunch in the past. And like I've always just said, like I just don't, that is not, that doesn't bother me. Either way, you're still telling me to make the decision about suppressing, amplifying the ad based off of its performance. Now like kinship doesn't then move test ads into scale campaigns because every ad is a test and I agree with that principle. You just kind of let the campaign scale it. But because I'm trying to stay out of learning at the same time, if I see an ad in a test campaign with my manual bid that is spending money, um, then I may want to go move that into an ad set that is like a quote unquote scale ad set or a scale campaign. And that way it sits in a, in a campaign that can get out of learning and stay that way and sort of again, potentially reduce volatility over time in terms of, uh, in terms of daily spend, particularly. Okay. Um, so, um, so I'm not picking winners by the way at all, uh, from test campaigns, the way that I am determining went to move an ad from a test campaign to a scale campaign is just when the ad gets 50 purchases. There is also some meta documentation that somebody pointed out to me on X, I might have a Jordan Minard. Thanks Jordan if you did that, um, who's not a manual bid guy, which, uh, is, is, you know, fine. It's just smart, dude. But, um, but, uh, Jordan, um, I think Jordan had pointed out that that meta had specifically recommended that you, that you launch new ads in a new environment to avoid false negatives. Basically, I know my friend Miranda, Akins has done this as well and that had helped her account. Also, they've seen some success, um, at 365 holdings where she works and, as a smart, really smart media bar there. Um, you know, some, some evidence that maybe launching an ad in a fresh campaign, and this is the logic of doing it, gets meta to not, um, suppress the ad too early and get a false negative. Now, I've been meet buying like this for a little while. Again, same, same manual bid in both. And I think Miranda's doing same manual bid in both. I'm not sure. Um, but, uh, but the point of it is just that maybe there is some evidence, maybe there is, uh, maybe there is some reality that, uh, launching an ad against your past winners, um, even at the same manual bid, uh, suppresses some ads too early. Now, I've worked with lots of brands that have scaled their spend fast doing this, right? So I think this is probably somewhat on the margins, but it's something that I've begun playing with more and more of my accounts. And I'll also tell you, I'm not 100% confident that it's actually doing anything. Uh, it's not clear to me that launching ads this way is really helping at all. Uh, in fact, I think it's possible that it's just creating more work. So I, I'm not totally confident in this, but I'm saying it partly just to be honest with you about, uh, a, a convert about a thing that I've been vocal about that is maybe shifting around for me. And also to give you some constraints within which to make this decision, which is like there's some meta documentation pointing this way. I'm willing to do it. What I'm not willing to bend on as the manual bid part of it because I would rather actually have some false negatives than, um, you know, false negatives meaning ads that get suppressed too early when they could have possibly spent. I'd rather have that than the problem of blowing a whole bunch of money on, on other ads because I don't think it's a very big problem. False negatives happen. I don't think they happen very often. And again, I'm going to give you a second away to try to avoid that problem if you're, if there's an ad you're super excited about and you want to make sure it gets a fair shot. Okay. So, um, still running manual bids, not picking winners based off of purchase behavior, higher the picking winners based on spend letting an ad get to 50 purchases before I move it. Sometimes I don't even move it. By the way, I just let it sit in that test campaign. If it's out of learning, because I don't like messing with learning. Um, so yeah, and if I need to, I'll cycle out low spenders. If they're just not, um, if they're just not spending at all, I'll cycle those out. And so I use a test campaign purely as a way to have a fresh environment to launch a new ad in case that helps meta sort of get a fresh read on it. That's really the whole point of it. Um, again, I'm, I'm not totally confident it's helping. But there's some evidence that it seems to have been decent for me. Definitely not the kind of thing that's overhauled my media buying since I started doing this. When we started doing this months ago, it has, I've not seen all my accounts suddenly blow up.
new creative, just launching and performing much better all the time, etc. Nothing like that at all, but possibly helping. Okay. So there you go. If you want to try it, that's the way I would do it. If you are serious about meta ads and you're serious about building great funnels, offer testing, testing landing pages, adding upsells in cart, adding subscription options, doing all the things that make performance marketers do their job effectively with landing pages. You need to consider going to for MOT. Go to for MOTcommerce.com/af to go check it out. But for MOT is the most robust landing page software I have ever seen and it's upgrading all the time. They just raised another giant round of funding to keep putting money into making this incredible piece of software that allows you to iterate on landing pages and on funnels faster and easier than ever before. The main core thing that for MOT is driving at all the time is trying to make it so that performance marketers can make very fast and easy changes to their post-click funnel experience for customers without getting in the way of the actual website without going and switching things around. They're going to screw up your PDP and get somebody your director of ecommerce or CEO or whatever it is. Mad because you keep wanting to run all these tests. Now you do everything in the self-contained funnel environment and you can match message on your landing page to message in your ad faster and easier than with any tool I've ever seen. You can also test all of it in Firmots software really quickly and easily. And one of my very favorite features is the ability to actually take one Firmots link. They call it a forever link. Put it in your ad, run your test on the back of that link and then automatically make the winning version of your test take over the link. So you don't have to go and every time you have a new test that wins, go change all your ad links and put everything back into learning and whatever it is. So Firmots is just incredibly robust landing page software. If you're serious about landing pages, if you're serious about offer testing, you should be considering trying Firmots at least get on a call with them, check it out for yourself. It's really awesome software. Again, we use it for multiple clients. One of my clients added it and saw like a 50% spend increase really, really fast. You should check it out too for my commerce.com/af from my commerce.com/af is the place to do it. Link is in the show notes. Go check it out. Number four, I am making much smaller bid and budget adjustments than I used to. So all of this in some ways goes back to the issue of the learning phase and really trying out to mess with ads that are actually spending. That's the whole point of the clip testing campaign thing. Like, or it's one of the big points of it is to launch an environment that's not going to mess with the learning of a dish of already existing ads. Okay, but also some 20% budget and bid adjustments steadily over time. I used to be pretty willing to go make gigantic budget and bid adjustments. But now, because that will often kick ads that's back into learning, I'm much slower to that. And in general, again, meta has pointed this out in some places. They had some documentation saying, maybe it's a good idea to make smaller adjustments over time with your bid and your budget again, as a way of sort of decreasing the volatility of your delivery and helping meta's machine learning perform at its best. For most of my accounts now, I'm doing this pretty consistently. If I'm in a, if I want to be more aggressive on an ad, like, I'll launch it at a lower bid and then just take that bid up 10% per day until it gets to the point that I want to get to or something like that. If I'm trying to try and launch and new ads more aggressively for whatever reason and there's times when you need to do that. Okay. So I'm not just going in like getting a bunch of learning on an ad set and then ripping the budget way up and ripping the bid way up. At some point, I have no problem with the idea. In fact, still I'm trying to get to a point all the time with my manual bids, where my ads are not hitting budget. But if I have an ad that's hit, if my ads are hitting their budget, okay, all the way through, then I'm ticking up by no more than 20% in a day as opposed to going in like doubling the budget or something like that, which I might have used to do. Again, all with the goal of staying and learning. So making smaller adjustments until I get the budget and bids to where it seemed like they're going to sit steadily, trying to stay out of learning, trying to keep those ads that's active all the time while at the same time moving the account forward. Okay. Number five, this is a bigger one. And that is that that I am making almost no true iterative ad creative changes at all. And what I mean by that is, you know, I put out an episode a long time ago. This is something I think I was just wrong about. Okay. I'm happy to say that when this is the case. I put out an episode called how I turn one ad into three hunt or one idea into 300 ads. Okay. And the logic was when you launch ads, test multiple hooks for the same ad, that way you give yourself the best chance for your ad to hit. If you're launching with multiple hooks, you've got more opportunities for your ad to hit. And and that way you don't sort of just like ruin your chances of having an ad hit because you wrote a crappy hook for what is otherwise a good concept. Okay. The problem is meta very clearly at this point, what is the case? And this is again, meta is again, very publicly said, create real variation in your ad accounts. This has been reinforced me with reps at every level. And it's been something that I've definitely noticed has made a difference in ad accounts. Okay. What they have said basically about this is you need a wider amount of variation. I've seen other people point out the same thing. Cody Ploughker, I know has been hot on this point as well. And the idea is that if ads are too similar to one another, meta will just read them as the same ad into the auction. Cody, Cody Wittick, Taylor Agstave said the same thing to me in the past as well. Again, the Kinship guys. If the ads are too similar, meta will will make no distinction between them as they enter into the auction. And in fact, if you just like have three or four ads that are all the same, they will, meta will just sort of randomly pick one and then it will enter into the auction. Like that. And so it's just sort of just adding randomness to the decision. It's not actually probabilistic decision. And I've definitely seen that where I'll launch four versions of an ad and one of them will get spend the other three won't and the other three will never really get any spend. And that appears to be because meta is sort of blocking those ads together, treating them as one ad and only picking one over time. Sometimes you'll see multiple of them spend little bits here and there. But that's not really mostly what it will do. In fact, instead, you should be thinking at the baseline strategy level, how do I create more variation in my ads? Now, this creates a problem because at the same time as you're thinking that, okay? If that's really the goal, creating more variation and more volume at the same time, you have an operational and production level problem, which is that it's much easier for an editor or an ad writer somebody to iterate off of version one of an ad than it is to create a whole new ad. And I still want to maintain some efficiency in my workflow in my creative supply chain. Okay? And so how do you sort of thread the needle here where you're where you're multiplying the output of your work without every single ad you make being a truly fresh start from the ground up? And at the same time, how do you learn from past ads that are winning and create variations with angles that you know work? Like, how do you sort of do all that stuff together? And here's where I've landed on this. We are still creating quote unquote iterative variations, but we are iterating more widely. So let me give you a sense of what I mean. In the past, I would have said one ad concept, four different hooks, something like that. Okay? That's the way we would write it. And I think there's still something right about that. I think it's still not a bad idea to write four different hooks. But instead of taking that same, let's call it a one minute ad with four different hooks. Instead of taking that same thing and saying it's literally just four versions of one minute out of the four hooks, what now we're doing is saying, okay, I'm going to take that one concept and turn it into a two minute, a one minute, a 30 second and a 15 second. Each one's going to get a different hook with different intro, B roll. Okay, a different visuals to start the ad and maybe some rearrange visuals throughout the ad, but it's going to be very similar. And what that allows you to do is is actually still take the same principle, which is that perhaps different versions of the same concept would work better or worse than others. And it's not a true split test. It's not a true hook split test or something like that, which is fine. Meta is a really bad split testing machine. Unless you run a actual AB test with four spends, Meta isn't an AB testing machine. It's a probabilistic machine. It's a probabilistic forecasting machine. It's Bayesian, not randomized controlled trials. That distinction makes any sense to you. Okay, so don't treat meta like a split tester. Instead, what I'm trying to do is create multiple options while some multiplying my work and have them be different enough that Meta enters each one separately into the auction and gets separate ad reads, separate data reads on all of them. And I have found this to work pretty well in terms of getting spend to more variations of the same ad because I'm now taking that sort of iterative idea and just making the iterations broader. Okay. Again, meaningfully different hooks, meaningfully different intro visuals, hook visuals, meaning and different lengths. Also, there's more distinction in each of those ads. And yet, it still allows me to multiply my work some in the sense that it allows me to go and say, when an editor is working through B-roll for a product, it's really easy for them to go and shift some things around to make the, especially if they start with the two minute version. They've already got all their B-roll selected. Now they can rearrange and play with it down for a one minute version, the 30 second version, the 15 second version, really fast. And it allows you to have some efficiency in that creative supply chain process while at the same time having real variation in it. And still iterating off of one angle or message that you know works because at the end of the day what I believe works and advertising is messaging. And so I do want to take multiple cracks at the message that I believe works, especially if I'm expanding on a message that I already know works. If the message is, this product is better than the product.
the competing products in the marketplace. Man, you can make a lot of different variations of an ad that has that basic framework. And maybe this product is better than the other products in the marketplace for X reason. Okay, you can again, go really, really far on making a lot of variations of that basic message and still have widely different ads between creators and still images and post-in-ode ads if you want to, right? Explainer ads and negative hook ads and myth versus truth ads and listicle ads and all these different things. Now, those are all really different from one another. So in that respect, one idea 300 ads is still true, okay? And that you are taking one core message and trying to multiply it across what you're doing. But in that variation process, and we're certainly not taking one still image and changing it with five different headlines anymore or anything like that, okay? Now, again, you can still take that same principle at the still image level, but instead of making one image, five headlines do five different images, each of the different headline, all the same message though. Now you've actually got something where, again, you're multiplying the output of your work in significant ways and taking that one message and trying to take a lot of cracks and creating real variation in the way that you accomplished, execute the delivery of that message, but at the same time, you don't run this issue where meta reads the ads as the same. So that's a big change for us. We're really trying to build that into a process more and more all the time. My friends behind the scene studio are still my creative partner for that. They've helped rebuild their process to reflect this idea. I think it's a really good way to do it. And I think you should be pursuing it that way as well. All right, number six, I am not technical, but one of the things that came out to me in the meta performance marketing summit was just, I continued to emphasis on data quality. And I've seen it much in this. I used to work with one company that was a pixel installation company. We had some trouble with that company for a while for a while and that made us that we ended up going away from them later on. This was mostly, I think their tools actually mostly really, really good. I don't have any issues with that. But there were a few issues with our interactions with them for our agency. And then some partners that they didn't integrate with as well. So now I've just started using LFR for more stuff. This is not a paid notice. Like they've just been great to work with on a bunch of different things. And what I have now consider LFR, and I hear great things about Blotout, by the way, I don't have a dog in that fight at all. I just knew LFR had interacted with Brad a long time ago, trusted them. And now we're just going to a point where we're just kind of making it SOP that we're using LFR as our Pixel installation, Cappy installation partner on across our clients. I really think of it as sort of insurance policy for your out-account. You've now got somebody who is a technical partner of something breaks that you can talk to who just like does best in class pixel installation and integration and you can do that, of course, across everything that you're doing with GA4, Google Ads, all those things. It's not overly expensive at this point. In fact, the prices have all kind of converged on each other as far as I can tell across the different options. And I've just found that's sort of a great thing to have on hand as a partner that allows you to sort of take your mind out of the problem of, is my data as good as possible? Is the data that I'm sending to meta as good as possible? Even from meta-side, I've heard extremely mixed things, in a really ways I can't even really talk about, about sort of what the best setup is here. So it's just not super obvious what that is. And so again, it seemed to me that working with a company that is trusted, that's in the space that people know and that allows you to sort of get that installation going pretty easy, really fast setup and is reliable. Like that's just the way that we've approached this. Like I said, it's sort of an insurance policy on the out-account. Particularly by the way, if you're in the health and wellness space where you've gotten tagged as a health and wellness brand and there's challenges with lower funnel optimization, et cetera, those technical partners can be especially helpful there because they can help you with the work around solutions to those kinds of problems that certainly help one of my supplement brands for that. And so yeah, so shout out to them again, that's not paid. I make no money on this. Tell them I said to you if you get this because they'll be glad to know that because I effort do want them to be a sponsor. But for now, just happy to represent them. They've been great to work with so far. So that's something that has been important to us. For a while, I was using the native Shopify integration after that, was using another partner right now, we're using LFR. Number seven, and this is kind of the most interesting thing that I learned a long time that there isn't very much public about though, there is public data about this, which is that if you have an ad that you want to relaunch totally fresh and you want to make sure that meta does not attach any past information to that ad, okay? There is a little trick you can do and that I start to do sometimes and that is to duplicate the ad but change the thumbnail because what I have learned, and this is again something that meta has said publicly but that they haven't been very public about, what I've learned is that, or they haven't pushed very hard on is that the ad ID and meta system is stored at the level of the thumbnail, which is like shocking. And so if you change the thumbnail, you're essentially making meta re-id the ad and treat it as a fresh brand new ad. So what I was saying earlier, if you have an ad that is a piece of new creative and you're like, man, just not getting any spend and I really want to make sure that meta gives us a real shot at winning. Here is the solution, duplicate it, change the thumbnail to a manual thumbnail, doesn't really matter which one. Just pick the manual thumbnail that you want and see if and see what happens because it's like a good trick. I think if you really have a piece of creative that you're just excited about, you want to make sure, if you put real time in effort into it, you want to make sure it gets a fair shot in the auction. You're very concerned about a false negative situation. Again, this is not something I'm very concerned about and meta ads, I think every ad basically gets a fair shot, but there's volatility in the early stages of signal or there is noise in the early stages of in relation to signal of any sort of small data set. And so yeah, if maybe meta, it's a probabilistic machine. And so that means it could be right about the probability but wrong in the individual application, if that makes sense. And so if you really want to relaunch it, change, duplicate it, change the thumbnail, see what happens. That's something that I've done a few times. I've taken some old ads also at times it seems to have died out and I've relaunched them at times. And I'm just like, I don't know why this ad died. Something is weird about this. So I want to sort of wipe the white meta sense of the idea and launch it truly fresh. I'll duplicate it and change the thumbnail. So a bunch of things like that. It's a great little trick. Can give you some peace of mind as you're running ads to make sure that again, you're not doing something wrong in terms of or that an ad isn't sort of like inappropriately suppressed. Again, I think this is a very small problem. The grants came in things from meta ads, but there are times when it could be a useful little tool. So there is, I'm going to run back through all seven of them real fast just so you have them here. Number one, 15 to 20 ads per ad set. Number two, low learning phase matters to me a lot more than it used to. It should matter to you as well. Number three, creative testing campaigns, maybe helpful. Maybe helpful. Only if you're using a manual bid still just like usual and really again, partly away to stay away from interrupting ads that are in the learning phase and also just getting ads launched in a fresh environment. Maybe helpful, maybe. Number four, much smaller bid and budget adjustments seven, 20% per day. Number five, iterative variations, none at all except for, or at least not the way that I used to do it. Number six, using LVAR in a third party pixel installation. It's been helpful. Number seven, change the thumbnail if you want to re-idean ad and relaunch it. Stay tuned, let's talk about a couple little things. We'll be done. Thanks for watching your listening. I have some really, really good episodes coming up that you are not going to want to miss. I believe by the time this is launched, I have not launched yet my Bill Dahl-Sandro episode, which really is good. Bill and I talked about sale of natural dog co. So if you want to hear about a really good win from somebody who's built a, you know, Lumedade figure e-commerce brand, I don't even know what the final revenue number was, but a figure e-commerce brand that has done well, gone to exit, had a good moment, listened to that, Bill's great on this kind of subject. Roman Khan coming on, talking about supply chain actually, which is going to be a really fascinating episode because he says the Roman's building one of the most impressive aggregators, like just e-commerce businesses I've been around, incredibly smart to a great conversation there. About Jordan West coming on creator relationship soon. So, and of course some individual episodes, including I'm going to walk through the brand that I'm starting, the thesis that is behind the brand that I'm starting to take you through all the economics that I'm thinking about from launch and brand. So if you're interested in that process of why I'm starting a brand and how I'm hoping it will work, I'm going to walk through that whole thing. So subscribe wherever you're watching listening. I think you will like a lot of what I have coming. Thanks again to Rich Panel and to for both great partners in this episode. You can go check them out in the show notes there. Partners, I'm really glad to represent. I use both and I'm very happy to endorse them to you. And of course, agfgrowth.com is the place to reach out to me. I'm just barely maybe beginning to think about taking lots more clients. If you're interested in that, go onto agfgrowth.com, submit your information there on the intake forms and we can get some conversation going. It's probably not going to happen right away, but if you want to get a little get out in front of the conversation, we can talk about it. Of course, subscribe to my newsletter there as well to subscribe to my email list when you go to agfgrowth.com, follow me on ex@androidjfarice. Email me
[email protected]. Thanks so much. Talk to you soon. (upbeat music)
PodcastTranscript AI