5 Takeaways From the Meta Performance Marketing Summit

[00:00:00] Richard: Hey folks. Welcome to the E-Commerce Playbook Podcast. I'm your host, Richard Gaffin, director of Digital Product Strategy here at Common Thread Collective. And of course, I'm joined, as always by Taylor Holliday, CEO, here at CTC Taylor. How are you doing? Coming off a bit of a 24 hour turnaround as I understand it.

[00:00:17] Taylor: Yeah, well, I'm old. I can't really hang with these young folk out in the the, the social world. Especially not Aaron Warshawsky, our VP of sales here. He is a, he is bet cher of the machine. And I, I can't hang, so I'm a little bit of a in recovery mode. But glad to be back in the office and excited to share a little bit about what I learned at the Facebook or the, sorry, excuse me, the Meta Performance Marketing Summit.

[00:00:40] Richard: I was gonna say, before we get into that, I was empathizing with you o on, on the sort of topic of getting older, getting into our thirties. Mid to late thirties. 

[00:00:49] Taylor: You might. Yeah, there 

[00:00:50] Richard: you go. And yeah, so I'm starting to feel the thing of like, if I have pasta the next day I'm hungover. So I can only imagine.

Yeah. Like what it feels like to go through that. But yeah. So let's talk about it then. So yesterday you were at the top secret summit, maybe not. So top secret of or rather, is the second annual Facebook performance. Marketing Summit took place yesterday, got all of the, all of the names in our world together, along with of course, representatives from Facebook to talk about really, I guess you would say.

It's like primarily talking about the product itself and things that they're rolling out and the way that they're thinking about the Facebook ad platform itself. But, so what we wanted to do today was to share with everybody here some key takeaways from. That summit things that are going to be applicable to you guys in the near future.

And just, just some interesting observations about what happened over that that day. So why don't we kick it off with the first one. The first key takeaway here, Taylor, which is the idea that Facebook puts on a branding masterclass. So dig into that a little bit. What, what are we talking about there?

[00:01:54] Taylor: So, as a, as a CEO has a responsibility for. Introducing ideology to my internal team, to my customers. And we, we spend a lot of time thinking about this, right? Like C, d, C has coined a lot of terms in the world, right? And so those kinds of messaging and branding things take a lot of effort and repetition and consistency.

And when you go to a meta event, one of the things that I notice is one, like they're just done in a world class fashion. So everything from the setup. The food, the stage, the decks, they all look beautiful as what you would expect from a business of their size. But more than that, They are on point with their message in an incredible way.

So the theme of this event, and we'll get into it more, was really about a s c and machine learning and the improvement or in ai. And they said the same things over and over and over again. Every speaker, same data points. So they do a few things incredibly well. They brand things and name them and say those names over and over and over again.

Re until you are tired of hearing it. And we have this joke that if people don't make fun of you for your phrases, like your phrases aren't embedded yet and mm-hmm. I was making jokes about ASC by the end of it. And then the second thing is like they take a small number of data points that are impactful and clear and they say them over so they don't try and say too much, and sometimes it feels repetitive.

You're like, okay, another slide with the 17% c p a reduction and 32% ROAS increase. And they had four different people say the same slide. But you know what, I know that data point, now I've remembered it. I've taken it away. And then they also eat their own dog food over, like I just received a survey from them asking me to reflect on.

Basically a brand, and I talked to Yoni about this, about what they'll do is like they'll do a brand lift study on their product ideas. So they'll ask me to recall specifically what I talked about, what they thought was most impactful, what I'll use, and they'll actually measure it and adjust the next event on the basis of that.

So I have a survey right now that's like asking me these very specific questions. Which of the following meta solutions do you plan to use in the next 90 days? To what extent do you agree or disagree that meta has these specific solutions? And then list all the speakers, which one did I find most impactful?

And then an open-ended response, and he said that they'll measure two things coming out of these events. Like one is, is there an uptick in adoption of the products? So like literally practical op opera, operationalizing the ideology inside of the organization to their financial benefit. And then two, like is there a recall of the specific ideas that they wanted to inject?

So, clear strategic plan, consistent branding on point, beautiful delivery, consistent repetition, measurement, and survey. Just a pure like masterclass in branding done by the meta team here. 

[00:04:36] Richard: Yeah. And a great takeaway from marketers in general about, again, yeah, it sounds like there was a masterclass in the power of repetition.

And I think like one thing that's easy to get into your head is that if you repeat yourself too many times people are gonna get tired of it. And actually, that's sort of the point I think of that's exactly, you know, the wealthiest, 1% of the wealthiest 1% or whatever, like Bernie Sanders is, whatever, like it's a, it's a meme at this point, but that's how you absolutely know what he is talking about every time he's gonna get up there.

So. So we're talking a little bit about how they messaged it and the effectiveness there, but of course then we have to talk about what was the content of that messaging, because you know, let's say for instance, four or five years ago, there was a similar thing where what they were messaging over and over was the pivot to video, and that ended up being false and bad.

Yeah. So what's the takeaway here? What were they, what were they talking about? Especially as it refers to Advantage Plus. And what were your sort of general 

[00:05:26] Taylor: feelings on that? Yeah, so the number one message item is about ASC a Vanda shopping campaigns or ASC plus. And they had sort of two key insights about what this is, is that one, it's a fundamental new AI machine learning algorithm that powers this campaign.

And the point that they made, and I thought it was again, they took a complex idea and boiled it down in a really simple idea, was that the ad auction has to make a decision on the delivery of your ad in one second. And basically what they've done is increase the amount of compute power that they can get through in one second.

That's what AI is doing. So the, the simple way they're saying is, we can process more data in that same amount of time to make a better decision. And that's what's powering a s c. And the result of all that compute is 17% reduction in C cpa, 32% increase in Ross in comparison to b A U business as usual campaigns.

And that sequence of. Here's the product, here's the difference, here's the why it works. And here's the outcome over and over and over again is why. They're just like, and then they want all of us, they gave us an action, a key takeaway, set a goal of getting to 30% a s c adoption across your campaign within the next 90 days.

So data, information, insight, action. And now I guarantee you all the reps are gonna call and go. Where are you at relative to 30%? Where are you at relative to 30%? And it's gonna just be this sequence of ac insight research, goal action, oper operationalizing measurement. 

[00:06:58] Richard: So as an independent observer, then like.

Is this, is this BS or not? I suppose maybe it's not as extreme as that, but what's your take on whether or not our ex or maybe, maybe maybe break down our experience with Advantage Plus as opposed to other campaigns? 

[00:07:13] Taylor: Yeah, so, so I, I find Advantage Plus to not really be that novel other than this alleged AI change.

Really, it's just a structural setup for how you run your Facebook campaigns. It's very similar to what we're already doing. So they force you into broad targeting. They force you into all placements. Right. And then they, you set an exclusion for new versus returning customer percentage, and they, you know, had some debate about what they thought the right percentage of that was and et cetera, but it's basically structurally.

A lot of what we were already recommending in many ways right now, the point number two on the five things I sort of learned at the Performance Marketing Summit is the reason we have been resistant to moving all of our customers to ASC is not cause we don't believe in the ai or we think it's about, it's because they don't offer cost controls.

They don't offer the ability to run cost caps or bid caps inside of ASC Plus. And this is something I've been very vocal about on Twitter and and even said to our reps like, we will move all of our customers there when you give us the ability to do that. And what I can say is, so this is a shout out to Yoi Levy who, for those of you that are on Twitter, he is sort of the, become the public face of meta on Twitter and takes a lot of inbound flack from all of us marketers.

But man, I'm appreciative for the guy Juan. He's an amazing human. He genuinely cares and he's an advocate for the things that we care about. And he showed me, Very directly that inside of the internal meta slack and engineering planning, they don't use Slack, whatever they use in there. That one, he's been sharing my tweets directly with the engineer team saying that, and I can confirm that it is on the product roadmap for cost controls to come to ASC Plus.

And I, I, I, I talked with multiple folks in there and one, like we are known as the people that want this inside of meta and. There. I talked to a woman named Olivia Kendall. She was a brilliant, runs the Auction Insights team, and she totally agreed. She's like, there's no reason this shouldn't be a consideration for the model, for the value of the outcome that you want.

Um mm-hmm. And so the sequence is gonna be, first, there's gonna be value optimization in Monroe as bidding. And then there's gonna be cost controls. That's the sequence that we shared. And so that's the only thing that's kept us from doing it. But we are very eager to take advantage of the new AI and combine it with an appropriate balance and protection of our customers downside to ensure they're more profitable than they were yesterday.

[00:09:22] Richard: And so maybe breakdown for us a little bit then. The advantage of, or, or why, rather, why we believe cost controls need to be added to this product. That's right. And maybe break down a little bit too, it sounds like maybe it's just been a matter of time, but why cost controls weren't included in the product at 

[00:09:35] Taylor: first.

Okay. So let's, so two separate questions. So why are cost controls so important? I think this is, this is. Two. Well, two things. One is they have an incredible feature of downside protection, which is that in the event that the model says that you cannot generate the finance, the outcome that you want, the efficiency target that you need, it will not spend and not losing money is the key to making money.

Okay. The way that you get better return on your investment is to not lose money, to not run bad spend. And so it is really, really critical to hedge your downside in advertising. And the machine can do that for you. It will just not deliver if you're not gonna get the result that you want. The other thing is, is that I think that the way that ad products are built have a fundamental flaw, which is the idea that the starting point for the construct of delivery is your daily budget.

I don't want my advertising to pivot around my spend. I want it to pivot around the available performance. So what cost caps do is they flip the script. They say, okay, this is my target or, or my expectation, my maximum or minimum or whatever, whether you're using bid caps or cost caps. The outcome that I want, what I want the variable to be is volume.

In the current ad product, the, the constant is your budget and the variable is performance cost, caps flip, that they make the constant performance and the variable volume. So I'll give you an example for Bamboo Earth on Saturdays and Sundays for whatever reason, whether that's the competition in the ad auction, whether it's, you know, our customer using the product more, we can spend about two and a half times as much money on Saturdays as Sunday as as we can on Monday, Tuesdays, and Wednesdays.

Without cost controls, I wouldn't be able to take advantage of those differing daily dynamics. And it is so important. It is, I can't even understate how important it is to be able to have performance be the constant and volume be the variable that I work off of. Versus every day I have to spend a thousand dollars regardless of what, how good the performance is.

So that's, that's why cost controls. Second question of. You asked, why didn't they do that? Well, now we're out. Now I'm gonna put on a tinfoil hat. And I'm gonna say that I think cuz it's not in their interest at all because they are rationally self-interested and that cost controls limit spend. That, that, like, that's what they do is they restrict unprofitable spend.

And the reality is a lot of Facebook spend is unprofitable. That's it. So, now that could also just be, you know, like they, they serve a very broad market. They serve app installs and, you know, general media buying objectives that are not all e-commerce, trying to drive profitable growth. So I recognize that I am a small fish inside of a very big pond with lots of objectives.

And I don't drive the meta product roadmap, although we're trying to be influential on it. So, so there's a lot of things, a lot of interest to serve in there too, and I recognize that. Right. 

[00:12:32] Richard: That makes sense. Okay, so let's move on to takeaway number four then, which is around measurement 360 or some of the discussion of a particular, one thing that you mentioned is mm m, which is media mix measurement.

Is that correct? 

[00:12:44] Taylor: Yeah. Media mix modeling or Yeah, media mix measurement. Yeah. I think probably you can see both, but 

[00:12:48] Richard: yeah, so talk through some of the discussion that you had there around how Facebook's approaching that and so forth. 

[00:12:56] Taylor: Yeah, so they are, Facebook's rolling out this sort of methodology, and you've probably seen Dex, this isn't necessarily new for something that they call Measurement 360, which is basically a framework for how they're encouraging brands to think about measuring media performance.

And the idea is that there's no. One way to do measurement. There's no single data point. And again, we know this, we recognize this as well. And so they're saying, okay, they really focus on three different tools. They talked about attribution, they talked about incrementality and they talked about mm m and so they talked about how to, the best brands are sort of using combinations of all of these things.

And they have this, like, they have this graph, this table that sort of takes the three different. Measurements and it looks at them as like speed, reliability and you know, I think how replicable they're, and they, they sort of show that like, mm m is like very accurate or, or is not very accurate, but it's like replicable or, or it's, no, this is what the three categories were.

I'm sorry. Cross-channel application accuracy and speed. And so attribution is like, great for speed, terrible for reliability and useless in cross-channel application, right? And so they would say that like, this helps you make daily decisions really fast and directional accuracy. And, and then they're like, okay experimental holdout studies.

Really highly accurate, not very cross-channel applicable and kind of slow. And then they're like, mm, m like really cross-channel, applicable. That's what it's for. Not that reliable and really slow. And so they're like, each of these sort of serve a different purpose in answering a similar question. And so they, they have this like triangle that they talk about and they had the, the head of performance from.

Botox and Juvederm, who is an incredibly smart dude, talking about how they, how they utilize this sort of framework of always on conversion lift studies with mm m and attribution modeling. And so they're sort of talking about how all these different pieces help paint a picture that helps you make decisions and ensure that you're making the most impactful outcomes.

And it was really interesting. I had a conversation that that woman, I mentioned Olivia Kendall, and she talked about how. Because I asked them like, so Facebook has this API o open source mm m called Robin. But they don't ever talk about it. And so I was asking her, I was like, why don't you guys talk about Robin from the stage?

And she's like, well, one of the really interesting things about mm m culturally is that she said in Europe and Asia, the adoption of mm m is like incredibly high. That the pacing at which people make decisions in those worlds is much slower. And so for the data to lag, to require a holdout study that then compares multiple channels, that takes the course over a month, that you only get one data feedback loop every month is like totally acceptable for decision making in the culturally in that world.

But in the United States, the adoption is almost none. And it's because we want answers every day about what to do, and we have quarterly earnings and monthly business goals. And so something that would take us a month to provide direction for action is almost useless and nobody values it. And so she said it's really hard to, as a data scientist and as someone responsible for.

Good thinking and data work that takes time, takes really structured testing, takes a lot of dollars to get to good data to introduce these tools in the culture that wants now a decision making framework to use it. And I thought that was a really, really fascinating insight. 

[00:16:29] Richard: Yeah. Let's, let's actually quickly, cuz I don't think we've done this yet, breakdown.

So MMM we said it's media mix modeling, but what exactly marketing mix modeling 

[00:16:37] Taylor: is, yeah. Yeah. What exactly is that? What, what. What an mm m is gonna do is it's gonna attempt to answer the question about the incremental value of every channel that you're advertising in. So when I say channel, I mean meta versus Google versus, you know, linear TV versus podcast versus everything else.

And the way that it does that is by taking periods of holdout reference for different customer bases. So they'll take a group of customers, And they'll make it so that for some period of time they never see any of your search ads. They, they hide all of your search ads from them and they compare the conversion and performance of that group of Cubs.

Customers absent seeing that ad product. And then they'll compare it against customers who did see that ad product and that Delta is the incremental value of that chat channel. So you'll run this for every channel over the course of a month to get an incremental performance factor of each channel that's supposed to allow you to make a comparison across channels to see which one is really driving the most incremental impact for you as an advertiser.

So that's what an MM M is doing, and it's great at giving you a snapshot of what happened in the past over a period of time and insight in comparing where you may be overspending or underspending dollars. The challenge I've seen with MM m is people want to turn that into a, they'll take those multipliers and they'll apply them every day to every channel and use it to make daily decisions.

And that is not what an MM is for. And it should not be used that way. And anybody who's using an mm m for daily measurement, I like, I think you are making a. Fatal, like a fatal flaw because every time I've seen it, what happens is after you actually run the mm m on the period of time that you were using, the actual weights and measures come back different than the ones you were using to make the decisions.

Because that's just the reality is that these, the output is not constant every month. It is not that like every channel performs exactly the same in every period of time. And so, What it's great for, and the way that I would recommend using an MM m, but this is again, why it's slow and hard, is that let's say in the month of March, your weights came back and it was like, oh, we're underserving Google versus Facebook.

What you have at that point is not an action. You have a hypothesis that now needs to be confirmed over multiple periods of time. So what you should be saying is how interesting we are potentially underserving Google. Let's see if the next month replicates that same idea. If it comes back again and it's like, oh, okay.

Consistent Again, Google was underserved. Now you could make a hypothesis, which is. If we increase the investment in this channel, that result will hold, the incremental weight will hold. Now you make an additional investment, and then a month later you assess whether that was true or not, that the action that you created in response to the historical data created the future event that you want.

And it's a very slow feedback loop that can help you allocate dollars between channels on a monthly cadence, but it can't help you make day-to-day decision making. 

[00:19:28] Richard: That's interesting. Yeah, because it sounded like that the implication of that woman who you were talking to was that, or maybe this is just sort of the way I read it, that Europe and Asia are kind of doing it the right way cuz they're, they're willing to be patient with their data and us Americans are so go, go, go.

Or whatever. But the way you're describing it, I mean maybe again, I'm also American, but you know, Thinking through data and taking action on it in like three month cadences or whatever seems like. So to what extent do you think it's, it would be helpful for us on this side 

[00:19:58] Taylor: of the role? I think embrace mm m.

So if you think about this triangle that they're talking about between measurement, incremental studies and attribution, I think that MMN should be running in the background. And every month you should review it to see if you have the right channel allocation or mix right. And you should make decisions and think about that allocation.

It should never be used inter month to make decisions and it should never be used inter channel to make decisions. And I think that the problem is what most people need actually operationally, is a tool to make decisions like that. They're actually making daily decisions across their budget and channels.

And so either that tool gets substituted and used in a way that it's not useful to be used or attempts to be done that way, or it's dismissed because it doesn't solve that problem. The idea of the measurement 360 is that you need multiple views of this. You need a lagging, longer term performance. Think of it as like always on brand lift studies, like you have a always on brand lift study that's showing you, am I getting to more unweighted unaided awareness over time?

That's a thing that you should be always tracking and reporting on regularly, but not using to make day-to-day decisions. And so inside of a really well-structured organization, you have different tools that serve different roles of answering different questions, but if you start using them ina inappropriately, Then it becomes problematic.

And it's also one expensive to get to that level of data complexity in stack. It requires a lot of sophistication in understanding how to use them and not, and to be able to educate internally and operationalize the ideas. So it's really challenging to pull this off I think in a way that's super effective, but done well, they can be super helpful.

Cool. 

[00:21:33] Richard: All right, so let's move on to our, our fifth and final key takeaway. Which is more sort of a, a general or maybe personal one for you, I think, which is the idea that at, at the, the marketing summit ended up being a meetup for a lot of people who have been through the CTC e ecosystem that, you know, we've met in the broader e-commerce ecosystem.

So just talk through that experience a little bit as well. 

[00:21:54] Taylor: Yeah. One, it's just good to get out from behind the screen and to remember that you're a part of a bigger community, community of people doing cool stuff that you can learn from and to share with, and so, And then for us, you know, we're 12 years into this and we have had a lot of awesome folks come through CTC that are now off doing cool things and so.

You know, we were there and saw, you know, got to hang out with Nick Shackleford who knows, a longtime C d c employee and friend and just catching up with what he's doing. And then, you know, we have Jimmy, almost, oh, who's leading paid over at Ridge right into him and the Ridge team, and got to sort of banter with Sean and Connor and Jimmy about what they're doing and just catch up.

In, in his world. And then, you know, Puja, who's a former media buyer at CTC is now over at ri I ran into Sam who she's, she's doing work for an agency out in Minnesota and just like, there's a bunch of folks that are out doing really cool things that are a part of it. And then to just get a bunch of people come up that, you know, are lurkers on Twitter and go, Hey, I don't really engage much, but I'm so grateful for your content.

And, you know, I gotta hang out with an awesome guy by the name of Hari who is the head of. Paid over or head of growth over at at house of Gaga cosmetics brand. And he, you know, I just had no idea that this person existed in the world or was in my community. And then got to meet him in person and share stories about things we learned.

And he gave me one of the most incredible insights that is like, blew my mind about L T V in a way that like I'm gonna be talking about more soon. So it's a little teaser, but just like, man, that was really cool. And there's just so many folks got to hang with Jason Panzer from Hex cloud and just.

Spend some time amongst the community and get out of just the Twitter timeline and actually see people and hear their stories. And I think that part of it is where, why I would encourage people to go out to these events, get amongst the community and, and, and build the connection. Cuz there is a special thing and it's, for the most part, it's, it's really PO and even the meta team, man, Yoi and Jordan Sterling and you know, there were so many people that I saw and met that like.

They're, they're really advocates and they really care and they're really trying to do good things for us. I know as media buyers sometimes it's easy to be, to feel like, man, I just doesn't feel like Facebook's always that helpful, or there's a bunch of issues, but there's a bunch of really good humans in this world that are really out there trying to be as helpful as they can.

And so, I don't know. I love feeling affirmed and positive and appreciative for the community. 

[00:24:09] Richard: Yeah. That's awesome. One other advantage perhaps of, of meeting up in person like that, can you confirm for me that Jimmy Olmos, formerly of ctc, currently of Ridge, is in fact six foot eight? 

[00:24:20] Taylor: Is that true? I, I mean, he is a giant, I'll tell you that.

And it's funny because Jimmy and I have actually never met in person before that moment. Mm-hmm. And so it's this weird thing where you're like, oh my gosh, it's so nice to meet you. And they're like, wait, didn't you guys work together for a couple years? And it's like, yeah, but we literally never met each other.

So that's always fun. But yeah, Jimmy, he's not, I'd say probably six four, maybe. He's a big dude. Big hands too, like real good, good grip strength. Wouldn't wanna get into an arm wrestling match with him. Yeah, 

[00:24:47] Richard: he's an athlete. We used to be on a team together and and actually, yeah, we would, you know, be slacking each other at 1:00 AM like that kind of thing.

But he was on the, the sort of video box, he always sat really low, so I thought he was really small. And then one day he told me how tall he was and I was like, I'd have to see that in person. And then he left and I never got the chance, so I'm glad. Yeah, he's glad that you're able to confirm that for us.

He's a big dude. He's a big dude. Yeah. Okay. One, one last takeaway then before we go out there, which would be like, what's. What's the real maybe practical or concrete takeaway from this in terms of decision making that needs to be made in the next month or so? So we've got our numbers from what are the, what are the Advantage Plus numbers again? 17%? 

[00:25:21] Taylor: 32%, 17% decrease in C P a. 32% increase in ROAS over baus. Business as usual campaigns. 

[00:25:28] Richard: So are we, is our advice then is the house view. We move away for ba baus on mass, dump all our money into Advantage Plus. What, what's your, what's your key advice? No, I 

[00:25:38] Taylor: mean, we're, we're still, we're still hold, right?

For cost controls. That's a lot of our belief. But that said, not everyone shares that belief. And I think amongst the campaign concepts like. You wanna align yourself with the interest of the organization. Like Tony, our VP of Media talks about this all the time, is that you can't fight the internal direction of the organization.

Mm-hmm. Like if PAX is gonna be the thing for Google and advanced shop shopping plus is gonna be the thing for Facebook, at some point you have to adopt the reality of where their organization is going. Cuz they're gonna sunset all the other stuff, like they're going to move in the direction that they want to.

So figuring out how to make it as useful for you as you can is really critical. So I think that it is worth. Checking that out and then like, I would go watch the replay. I would download the materials. I would understand those things. There, there's a lot of smart folks. Another like crazy one was the woman, the, a woman that leads the growth at Fab say that Fab, she said that Fab produces 50,000 ads a year and Oh my God.

So they, they, they had this other data point that was like, After a frequency of four, the conversion rate on the expected conversion rate on ads drafted by like 40%. And so creative diversification and various formats is a big also part of the, the AI opportunities and. So I think there's a lot of like creative tooling that's gonna come to help empower that and enable that.

And there she was incredibly smart. She like the, what the program that they had going on was really cool. And, and like another example, I ran into Alejandro from Martin Bow. He was there. Mm-hmm. Oh yeah. Who's just the best human on earth? I just love him so much. Long, longtime customer of CTC’s.

And he's outsourced all of his labor now to to Latin America, which is really cool. And he has this really interesting structure internally where he's got pods, like sort of old school CTC pods, they of it as a videographer, creative strategist, and a media buyer for every product line now. So he has a pod that works on men's genes.

He has a pod that works on women's jeans. He has a pod that works on men's basics and like, and they're all responsible for creative for just that product line. And they get their own campaigns and it's just a very robust. Expression of the amount of effort going into all these things. And I think there's a lot of that occurring in a way that's, that's pretty cool.

So that was like four more takeaways for you. 

[00:27:48] Richard: That's right. Yeah. All useful though. All right folks. Yeah. Well, thanks again for joining us on the E-Commerce Playbook podcast and yeah, we'll see you all next week.