We Analyzed $1.5 Billion in Meta Spend. Here's What We Found.

[00:00:00] Richard: Hey, folks, welcome to the Ecommerce Playbook podcast.

[00:00:02] I'm your host, Richard Gaffan, director of digital product strategy here at Common Thread Collective, and I'm joined today by The Chopper, Mr. Tony Chop, who is our... And you just told me this, but you have to remind me again. What is your official title, Tony?

[00:00:14] Tony: Vice President of Media Investment

[00:00:16] Richard: Vice President of Media Investment, which for our purposes here today means that he knows everything about paid media. And what we were discussing right before we hit record here was the fact that is we've sort of connected Claude and our AI tools to our own MCP. We've been able to develop analysis at an incredible rate of speed here, or an incredible volume rather.

[00:00:38] And so there's lots of information coming through and we've had a ton of opportunity to explore questions that we haven't really been able to explore in the same way before. And so one thing that we've been kind of, going over recently is analyzing how platforms behave against bidding strategies.

[00:00:54] And what we mean by that is if you set, say, a, I don't know, Monroe S target in Meta, does Meta actually deliver against that outcome and act- how close does it actually get to that target ROAS in, in actual fact? And s- similar for like a cost cap or a CPA target, how close is it getting to those things?

[00:01:12] And so we wanted to talk through a little bit of that research today, but Tony, why don't you give us some background on some of this research that we've been doing and, and some of the conclusions we've come to.

[00:01:22] Tony: Quick, just quick timeout. My internet is real choppy. Can you guys hear me okay?

[00:01:27] Richard: Yeah, you're good right now

[00:01:30] Tony: Cool. Yeah. Nice to see you again, Richard. Hope you guys are still enjoying summer up there in, in Portland. Yeah, it's been-- You know, this year's been a really fascinating, fascinating time. I, I've been at this for a long time, and well, earlier this year, we, we, we, we got set up with Claude.

[00:01:52] So we have a corporate Claude, and we, we connected Claude to the Statlas database via an MCP. And it's kind of always been a d- a dream of mine to be able to really ask some of these, these questions at scale o- on top of this, mm, this incredible database that we have of all of our, all of our e-commerce customers. And to your point, like for a long time, forever at CTC, we've believed, we've had a strong belief in, call it cost-controlled media buying. And that takes a, a lot of different shapes. But in essence, media buying can be volume-based or it can be cost-constrained, cost-constrained. And we have been long been advocates for, for the latter, for forcing the media to achieve some sort of financial outcome in order to earn its delivery.

[00:02:47] So, and contained within that sort of sphere, all of the platforms give us multiple options for each of these pathways, for volume-based bidding and for cost-controlled bidding. And so the questions that we've been exploring and, and trying to scratch at with our, with our broad database is how, how well do they work?

[00:03:16] Right? So if we tell Meta, "Hey, it's a $50 cost cap," or if we tell Google you know, "Give us a $70 target cost per acquisition, target CPA," or if we tell Meta, "Give us a 2.0 min ROAS," or Google a 2.5 target CPA, how effective is that setting that, that we give to the platform? And are there any differences and, and nuances between the different options that are available to us?

[00:03:48] And are those differences or nuances-- How are those differences and/or nuances useful for us on the media investment side and useful for us to, you know, share our perspective internally at CTC, like in, in the canon in how we operate and ultimately with, with our audience as well? So that, that's kind of the setup.

[00:04:08] And there's been, I think, some pretty interesting takeaways so far.

[00:04:13] Richard: Yeah. Well, so let's, let's just then jump right into, into the results, and maybe we talk a little bit about like, the setup of some of these experiments. Like what exactly-- what's like, I don't know, the sample size, what exactly are we testing? Like, talk to that a little bit.

[00:04:30] Tony: Yeah. Yeah. So the, the first one we looked at was was for Meta, and we- the report was generated earlier in the summer, back in July. And we looked at a chunk of investment between beginning of March and the end of June. We looked at 300, approximately 320 Meta accounts and just under $1.5 billion in media spend.

[00:04:58] So that was the sample size of the, the Meta survey. And the Meta survey, mm, came with the, the, the fascinating and on the n- right on the nose title of Do Cost Controls Work? Do they actually have any impact? And the, the big takeaway from the, the research, and we'll, we'd be happy to share the entire report 'cause it's, it's pretty dense.

[00:05:27] But the big takeaway, the punchline if you will, is MinRoas is highly effective at achieving its outcome. Highly effective. Within several percentage points of achieving its outcome at, at the portfolio scale. And I think this is gonna be a really important A really important part of this conversation is that the things that we are-- the things that we're talking about that are true at the portfolio scale there is obviously differences within any individual account. But we have to hold these ideas, these ideas in tension that it's, it can be true for the portfolio that min ROAS is very likely to achieve the setting that it, it is that it is set to, while at the same time there's individual account variance.

[00:06:17] We can hold these two ideas at the same time. So min ROAS, highly effective at achieving its target. Cost per result, very much less so effective. On, on average across our data set, the cost per result goal only achieved its target less than half of the time. So in one sense, we have a bidding strategy like if you think about it, we're, we're media, A-as a media investor, I wanna go and give the money to the platform and have a certain level of confidence that the outcome that I need I'm gonna get, right? And what we've learned from this initial pass with Meta is that if we're gonna make a bet somewhere, min ROAS is the bet to make. And this ultimately has informed CTC's Meta canon in how we think about applying our account structure and our bidding principles because while there's no guarantees on an individual case-by-case basis, we can be quite sure and quite confident that this is the right initial bet to make

[00:07:26] Richard: Yeah. So the idea then being obviously we've been advocates for cost controls for a long time, but that was sort of under the-- because we couldn't test the assumption that Meta's cost control product, let's say, actually worked properly. And so there's like a little bit of an element here of cost per result does not deliver the way that it ought to deliver, and min ROAS does.

[00:07:51] And so at that simple level, that's the preferred method, or is there some more nuance to it, or?

[00:07:59] Tony: Yeah. I, I think that's a, I think that's a, a fair sort of way to interpret it and just kind of let it sink into mind. I think just kind of zooming out a s- a second the ROAS bidding product is the newer bidding product. It's the newer optimization option in, in Meta and in Google. The cost cap or target CPA were earlier forms of cost-controlled bidding.

[00:08:27] So you can think about the R- ROAS or s- essentially value bidding as like the, the newer technology. Also, if you think about the, the mechanics of how the, the-- ultimately the, the bidding engine works in, in cost-controlled bidding whether TCPA or, or cost cap there's essentially a ceiling that the system is trying to operate under.

[00:08:52] It's trying to ensure that all of the conversions sit below this threshold. In a ROAS outcome, the, the cost per conversion can vary based on the AOV to create the marginal outcome. So there's-- One way to think about it is there's more flexibility available to the system to go find high-value buyers that achieve like a higher basket, a higher cart size, and ultimately create more value.

[00:09:21] But the cost, the sort of cost per purchase might actually be higher on that, on that particular order, but the ROAS will achieve the outcome. So the, the way that I think about it is that the, the ROAS bidding options give the algorithms more flexibility to go find different types of buyers, where the cost-constrained bidding options are gonna be more more constrained effectively.

[00:09:43] Richard: Yeah. Interesting. Okay, let's, let's switch over to Google then and talk a little bit about our results there, 'cause they're similar. And then yeah, talk a little bit about that and, and the upshot of it

[00:09:54] Tony: Well, yeah, I think this is where, this is where things got really interesting, right? So, the Google study is the more recent one that we just, that we just put together. So, we're looking at a big chunk of data. So, 24 months worth of data back to June of 2024 through May of this year.

[00:10:08] 134 stores in the sample size, and just over $100 million in qualified spend. So- Very interesting that the result of the analysis on the Google set, which is does Google adhere to the targets that we give it in the platform, are very similar to Meta. In the sense that tROAS, target ROAS bidding, is actually achieving above its target 1.3x the time.

[00:10:36] So m- more often than not, tROAS is actually running above the target. TCPA is the exact opposite. It's running below its target at a factor of 0.8, okay? So I think while we have-- Again, this is, this is back to, like, there's gonna be-- You're always gonna have variance at the individual level. But if you, if you're gonna, if you're gonna compose a canon for how to do media investing and you're gonna-- And for your brand, if you're thinking about a place to start, it is, it is 100% advisable to use ROAS bidding options either on Meta or on Google, and you are, you are absolutely more likely to have the, the outcome that you need as a media buyer to say, "I need, I need a two-to-one return on this media investment or I lose money."

[00:11:28] And I think one of the things that's, that's really, I think im-

[00:11:33] Here's the thing, Richard, that I think is hard about media buying, okay? There's lots of things that are hard about media, media investing, but here's, here's the thing that I think it really, it really cuts on. You can either produce a lot of value or you can destroy value.

[00:11:56] Richard: Mm-hmm. Yeah

[00:11:59] Tony: Right? And if you think about that, like one, one of the things that we, we do at CTC is we get really sharp about how, how businesses make money, right? We understand the P&L, we understand the cost profile, we understand we understand, we understand the customer, and we, we use all of that business acumen about how e-commerce businesses make money and our specific client, how they make money. And we, we get to an, an acquisition target. And then you and I have talked about incrementality, right? We sort of take that acquisition target and we f- we fold into incrementality and our understanding of the incremental contribution of, of the channel. And then that leads us to a target for the channel, right?

[00:12:46] So we go AMER for a business is 2.0. The incrementality of Meta seven-day click is 1.2, therefore our on-platform target for Meta is 1.8, okay? Dot, dot, dot, dot, dot. Now, but think about that. We've just walked this-- We've like threaded this needle. Like this is the number where if we, we-- This is the number we believe gives us the, the maximum amount of opportunity to invest, AKA we have the most aggressive bid.

[00:13:21] You know, I think Taylor's mentioned the, you know, Zuckerberg quote about the the person that's willing to spend the most to acquire the customer is the person who will acq-acquire the customer. So we're always pushing, always pushing the math to say, "How low can we set the target and still make money?"

[00:13:38] Because that's gonna make us the most aggressive in the auction, the most competitive in the auction. It's gonna make us the most likely to get in front of the, in front of that user, right?

[00:13:47] Richard: Mm-hmm.

[00:13:47] Tony: How low can we go? Now, the result of pushing that target as low as mathematically possible to-- with a threshold of, you know, break even or make a little bit of c- profit per order, whatever it is that we come to, is that on the other end of that, on the other side of that is lose money on every order, right?

[00:14:06] So th- this is what I mean when I say like the, the confidence that we have when we go and say, "Meta, give me a 2.0," or "Google, give me a 1.8 TROAS," our confidence in the, in the system's ability to deliver that outcome is, is paramount for us to be able to do, do our job. And that's why I think this, you know, for forever I've had, you know, anecdotal experiences with using this bidding strategy or that bidding strategy and M- again, individual circumstances will continue to vary, but it gives me a great deal of confidence in the CTC canon, the principle of this is where we start with media, with media buying for our projects

[00:14:54] Richard: Yeah. But I think you, you make an important point. To sort of summarize a little bit is that, like, in order to-- because it's a risk, paid media is a risk, there's an incredible amount of precision that needs to go into finding what that number is. And then, of course, like you're saying, that has to be balanced with that you have to make that number, I mean, we're talking about T-ROS specifically, as low as possible in order to spend as much as you possibly can before you start to lose money.

[00:15:19] And so it's all about riding this line between those two things. And in order to get to what that line is and not fall over the edge, you have to have a ton of tools in place. You're talking about incrementality, of course, like having confidence in the product itself to actually do what it says it's gonna do.

[00:15:34] All of those things have to be in place in order for you to ride that line the way that we like to ride it. Which is why we do, you know, produce these volumes and volumes of, of reportage and research and whatever, just to make sure that the data is-- that the feedback we're getting from Meta is accurate, I guess.

[00:15:51] Yeah.

[00:15:52] Tony: Yeah. Yeah, I just, I wanna double, double-click on like just reemphasize, reemphasizing like the reason why the math to arrive at how low can we push the target is because all of the media platforms are auction. It's we're participating in an auction against other advertisers. So au- the m- the m- better-- like from, from top to bottom, the, the more r- the more...

[00:16:28] What's the word I wanna use? The better the DNA of the business. So like subscription brands are like the sort of the,

[00:16:38] Richard: Yep

[00:16:38] Tony: the poster child for this. Like s- s- so much LTV they often can and do and are willing to acquire customers at breakeven or potentially even a loss. So this is really good business DNA that allows for, you know, a really aggressive sort of front-end advertising strategy.

[00:16:55] That's business, business makeup incrementality understanding, the contribution of the channel. All, all of these things lead to our-- a posture of us having a high level of confidence in a, in entering that auction as competitively as possible.

[00:17:13] Richard: Mm-hmm. Yeah. No, no, that makes sense. And, and this is something that we've d- I've been discussing recently with Joy, and we've been discussing for a long time too, of like there's... It- maybe it's like a, it's not necessarily that a business' DNA needs to be good versus bad as much as like there's certain types of business DNA that are, to continue the analogy, symbiotic with a platform like Meta or e-commerce in general.

[00:17:37] Like, there's certain types of brands and industries where s- where paid media in this environment works well. And so you mentioned the subscript-- like a high LTV brand because, because you win if you can, if you can afford to lose money on first purchase, that's a great way to win, and, and it gives you the opportunity to spend more money to bid higher in the auction to win more auctions and so forth.

[00:17:59] But is that a fair way to think about it? Like, there's certain types of businesses, there's certain types of DNA that work with paid media well, and it's our job and a lot, a lot of this research is identifying how your business can fit in with this environment, I guess.

[00:18:17] Tony: Yeah. Yeah, 100%. Yeah. It's all, it's all part of like one big cohesive whole, right? So, you know, understanding your business DNA Understanding ha-having a re-really good understanding of a measurement stack. You know, and for, for us at CTC, it, it's MMM and incrementality and con-connecting like-- My, my title as media in-investor is sort of like indicative of like what, what we actually do is we deploy capital to create a return And the, the tools of the trade are incrementality and MMM and on the business financial planning side, like the spend and AMR model and returning customer model that helps us understand the, the spending power of a business.

[00:19:05] And then ultimately when we get into the platforms, you know, the bidding settings, like what, what bid do we u- what tool do we use? Like, do we use you know, so use the metaphor of tools like we, we could kinda think about like the older versions, the TCPAs, the cost caps is like maybe a little bit more older tools, a little bit more rough around the edges.

[00:19:26] And I think what we're seeing from the data is that the newer tools, the, the MinROAS, the TROAS are, are sharper. They're more-- It's more of a scalpel. It's more predictable. And that's really to, to me as a, as a media, as a media investor that's walking the line on, you know, how do we de- deploy this.

[00:19:44] You know, we-- Richard, we sit, we sit across the $800 million media investment across our portfolio of clients. Like, it's almost a billion dollars, bro. Like, how do we, how do we deploy that y- confidently? And I think that's, that's where we're that's where we're pleased to be able to do this, this type of portfolio-wide research.

[00:20:02] But I, wanna hit, I wanna hit one other topic on this because I think there's some interesting, there's some interesting sort of things that are happening around the sort of little flourishes around the, the bidding systems. And so I wanna, I wanna call out something on the Google side. So, for anybody that's logged into a Google Ads account, they've recently over the last several months, you've seen a, a notification in the account that Google's changing the their, their bidding mechanism for-- They're saying, they're saying it's gonna be more consistent, predictable performance against TCPA and TROAS bidding including when budgets change.

[00:20:41] So I've had l- some, some laughs about this with our, our Google agency partner 'cause I was sort of like, "What do you mean more consistent and predictable? Like, isn't the, isn't this the promise of the bidding, like the TCPA thing like a-already? Like, isn't that what you're saying?" So, that was sort of like i- isn't it like isn't this how it's been supposed to work?

[00:21:03] But anyway, I think it's, I think it's the, the language that we're hearing back from our, our partners is that it's, it's really a function of campaigns in Google that are limited by budget. Google Google gives us a, a really cool indicator in the platform that is we don't, we don't see it anywhere else, where it'll throw a flag that says limited by budget, AKA if you have your TROAS set to two and you have your budget set to $1,000 a day or whatever else after the system processes for some length of time, it will give you a flag to say, "Hey, you, you could actually turn up the budget here and maintain the same result."

[00:21:39] Now what, what's happened and what can happen and what we've seen happen is you go to that same campaign, you, you know, triple the budget or whatever else, and then the the, the return, it's really unpredictable, okay? And that-- So back to this whole like the spirit of this whole conversation, like what, what, what's really valuable to me as a media investor is predictability.

[00:22:00] So what, what we're hearing from Google is that we can expect more predictability specifically around campaigns that are limited by budget and specifically around increasing that budget. Which to me is like hallelujah. That's, that's amazing. So that, that's like one little embellishment around this, this whole, this whole universe around bidding and predictability and expectation.

[00:22:27] There's another thing that I've, that I've seen recently that I'm, that I'm, I'm pretty excited about, and I think it's, it's like a different, it's a different take on the same idea. And it's happening on, on TikTok, and we first spot it on, spotted it on GMV Max ads, so ads in the shops ecosystem. And it was a-- It's pretty cool.

[00:22:49] It's like you give the campaign a daily budget and then TikTok, the GMV Max campaign has an option to a budget scaling option. So, s- said simply, if at target scale budget, right? And then you can set-- There's like different settings. You can set the interval, like look at it, like check like five times a day and scale up by 10 or 20%.

[00:23:11] So it's all kind of configurable. But the premise is pretty simple. It's if the system for whatever reason finds a pocket of demand, go and scale into it if you're at the target, right? And I think the, the what, what this solves for is the, the fear and the apprehension and sometimes the, the, the challenge in this whole world is like on the Meta and Google side, what the platforms want is budget liquidity.

[00:23:38] They want a big daily budget, right? Okay? What I want as a media buyer is predictability, confidence that my media investment is gonna return at this level because I'm threading a needle here, Richard. Like, one way creates a bunch of order volume and contribution margin, the other way torches it, right? So Meta, Google, they want big, like, daily budgets for budget liquidity.

[00:24:04] That's how the system works. I, I talked to you before about like, MinROS and TROS bidding having the opportunity to go for more like higher value buyers at a higher CAC, blah, blah, blah, blah, blah. Budget liquidity is the thing. TikTok is approaching it a little bit differently where they're saying like, "Hey, you don't have to give us unlimited budget liquidity.

[00:24:21] You tell us the target, you tell us how much we can roll this budget up if we're-- if we find a pocket of demand, and then the next day the budget resets back down to that sort of normal baseline." So, I, I really, I really like that mechanism, and again, we saw it in GMV Max ads a couple months ago, and we're starting to see it percolate into the, the web ad side as well.

[00:24:42] But I, I think it's a, it's another sort of cut of this whole idea around what do the platforms need in order to be successful? What do I need to be successful as a media investor? And you know, how, how do we, how do we put it all together?

[00:24:56] Richard: No. Cool. So let's let's quickly... I, I, I mean, we've gotten a, a lot of takeaways out of, out of this, but just, just to make sure we summarize here, use tier ROS, min ROS, don't use cost caps. That's, that's the canon stance right now, yeah?

[00:25:11] Tony: Yeah, I don't like-- I, I hate these-- I hate, I don't, I struggle with like black and white things. Like the fou- the foundation, the found- if you're gonna build a media, media platform, your foundation should be ROAS, ROAS-based bidding, value-based bidding. Okay? We have, in our canon, we actually have some language around bid surface expansion through bid cap specifically.

[00:25:31] Okay? So I, I just want you to think about the takeaway needs to be use ROAS bidding as your core. It's not like, hey, don't ever do anything else, okay? I don't wanna be that about it. I wanna be-- I wanna say use ROAS bidding at your core because it is, it is across the portfolio the most likely to help you thread that needle

[00:25:52] Richard: That's right. All right, cool. Well, you, you heard the man. So I think, I think we'll, we'll cap it there. But the idea is the, the headline is that min ROS, T ROS work the way they say they will work or even better in some circumstances. There are maybe circumstances in which the other bid types make sense, but in this particular case, this is what we've, we've discovered.

[00:26:12] And as we, again, like this an incredible volume of research coming out of, of the CTC think tank, so we'll have plenty more to share with you and, and go over with Tony. So one more thing I wanna say that is if you wanna join us and join Tony Topp himself for a in-person workshop on building out your paid media account and thinking through some of these things that in-person workshop is next Wednesday, September the 2nd. Tony will be there to break down Meta ad accounts, kinda run through his stance on paid media.

[00:26:44] So please join us for that. You can check out a link either on our website or in the show notes for tickets to that event. But anyway, I just wanted to say that and I think, Tony, anything else you wanna hear here?

[00:26:55] Tony: No, I think that's it for this one. We'll see you, we'll see you next time, Richard, with way more a bu- a bunch analysis and reports

[00:27:03] Richard: With hundreds of thousands of words of analysis in between now and next time. So we shall see. But all right, folks, thank you again for joining us, and we'll see you all next time