The Most Consequential Meta Update of All Time

[00:00:00] Richard Gaffin: Hey folks. Welcome to the Ecommerce Playbook Podcast. I'm your host, Richard Gaffin, Director of Digital Product Strategy here at Common Thread Collective. And I'm joined this fine morning. It's a Monday by Taylor Holiday, who is of course, the CEO here at CTC. Taylor, what's going on today?

[00:00:15] Taylor Holiday: You know, just going to try and live up to this headline. So I don't get accused of clickbait of calling this the most consequential meta update of all time. So let's see if we can make good on 

[00:00:25] Richard Gaffin: do it. Do you get accused of clickbait often, Taylor?

[00:00:28] Taylor Holiday: Yeah. Regularly. Regularly. 

[00:00:30] Richard Gaffin: That's just cause he got something to say. That's what I say.

All right. So. Today, here, and speaking of which, we're gonna, we're gonna jump into, this is essentially another segment of Defend That Tweet, where Taylor has tweeted this is last Wednesday, I guess, so it's a little bit older, but, Today marks the most consequential update in a meta ad product of all time. Meta is releasing one. Optimization for third party measurements, starting with GA and North beam with triple well shortly behind to optimization for incremental conversions. One of these has the potential to make massive business impact for you. I'll let you debate what you think it is, or another way you could phrase that as choose wisely.

So, I think like maybe a good place to jump in is just some overview of maybe the messaging or the language that we got from meta about what this is. So, do you want to maybe jump into that real quick? Like in

[00:01:18] Taylor Holiday: Well, let me, let me, set a historical foundation for why it's so consequential. So if, if we go back to the. Attribution wars of the post iOS era, where I was fighting all the attribution platforms. It was fundamentally rooted in this problem of disassociating optimization and measurement. So people, if you're not familiar with the meta ads product, there's a really important selection that you make in the process of building an ad campaign.

And that is, what would you like meta to optimize for? Okay. What, and what that is saying is which purchases. That occur in different views. Do you want that at the count as a conversion and then try and optimize for more things like that. So if we think about what the choices are right now, you can choose the default is seven day, click one day view.

That means that every purchase that occurs within seven days of a user clicking on an ad or 24 hours of them seeing an ad will be counted. Buy Meta's algorithm towards a purchase that meets the objective of the campaign. And they will optimize to accomplish more things like that. You can narrow that to just seven day.

Click. You can narrow it to one day. Click one day view. You can narrow it even further to just one day. Click. And they recently at a few months ago added engaged views, which is another optimization setting. And the most consequential thing you can change inside of a campaign is not the attribution setting attribution actually has no effect On optimization, but it's related to optimization.

And so my problem with all the third party tooling was that it had no effect on optimization that it was. And so what would happen is as we, as advertisers, one come in. We'd try and change things in the ad account, but you couldn't move a lever in the ad account to affect the attribution view. This is the problem when people would use last click Google analytics, it's the problem when they would use triple whale, it's the problem when they would use North beam, any third party measurement solution that was disassociated from optimization was a problem.

This is a really important foundational thing to understand about why this is so consequential in terms of what they're doing. Metta has never allowed you to optimize for a business outcome that wasn't. One of those five settings as reported by their internal tools. People's challenge with that is that the reported results inside of meta are not often directly related to your business outcomes.

Okay. So that's why everybody would be frustrated with it. Okay. So that's an important piece of historical context. The second important piece is to listen to something that Mark Zuckerberg said in an earnings call this past week about where meta is headed and how AI plays into this story as well. Mark said that eventually the ad product, you are going to simply give meta your business objective and they will handle the rest.

And if you think about every piece of tactical execution, That exists from media, buying account set up to creative, to targeting, to all these things that the user usually controls. I believe that AI is going to take, is going to do all of them more effectively. But the one thing that the human is going to have to always decide is what is the goal for my business?

Because that is very fundamentally a subject. There is no objective answer to that. It's not a mathematical formula. It's always about what you as the user want. So understanding that context of where this is headed, where has it been? Is the right foundation for discussing now this next set of specific updates and what they just enabled.

So that's my context.

[00:04:58] Richard Gaffin: gotcha. Okay. So, well then let's go into this sort of binary choice that you've set up here, which is. Okay, so now, obviously, the two things that are being released are one now you can optimize for third party measurement, which seems

like it sort of erases the problem that you were talking about.

Although I suspect it doesn't totally. And then the 2nd piece, of course, is now optimization for incremental conversions, which essentially, and here, here's the language that that Medi uses. Incremental conversions are conversions that would not have occurred without the ad being shown in contrast to conversions when optimizing for volume, which focus on driving as many total attributed conversions as possible.

So maybe it would be helpful to dig in a little bit more to actually the definition there on the second one

[00:05:39] Taylor Holiday: Well, do you want to go to incrementality or do we want to start 

with the third party tools? Cause there are two different updates and I want to make sure we, let's, let's start with the third party analytics tools. So can you read maybe what they've said about what's possible there? And then I'll dive into it.

[00:05:53] Richard Gaffin: let me quick. So this is what Metta has to say, essentially, about including third party optimization tools. We've already received meaningful aggregated insights from test advertisers who have integrated their analytics tools. Using those insights and combining it with additional internal learnings, we are beginning to roll out changes to our ad system that take more of the cross publisher journey into account to help drive more value for businesses. Based on this change, we expect advertisers to see improvements in meta attributed conversions within their third party analytics tools, although CPMs may increase as a result. When we tested this in a recent experiment, advertisers saw a 30 percent average increase in their meta attributed conversions as measured by their third party tools.

[00:06:33] Taylor Holiday: Okay. So what Metta is saying is, wait, if you're going to measure us based on Google analytics attribution or North beam attribution, or coming soon, triple whale attribution, fine. We're now going to optimize for those outcomes. And what they're saying is in the process of connecting attribution optimization, we see an increase in purchases.

Here's the important part as reported by that tool.

So you want me to make last click GA go up? No problem. We will now make that number go up by optimizing for specifically those purchases, because before what was happening is you were measuring last click GA, but Metta was seeing and reporting and optimizing for a bunch of other purchases based on the optimization setting you have.

This is why the whole thing was a scam. And this is like, this is actually like so validating to fighting this battle was that. You cannot disassociate these things. You can not disassociate optimization and measurement. And that's, and that's what this is all proving. So here's the really important thing is what meta is capitulating to that.

I think they're ultimately going to regret doing this is basically saying, fine, you want to use that to measure our results. Go ahead. We'll just make your magical number inside the third party platform. Go up. Cool. Now, the question is. Does that measurement actually produce a business outcome? And this is where the second thing that they introduced actually, I think is the much more important thing.

And what I was being a little snarky about in my tweet is one of these matters and one of it doesn't, because I think plugging in Google analytics and optimizing for GA is a terrible idea. There, I said it, 

it's a terrible idea. It is not what you actually want. What you actually want are incremental conversions to your business.

In other words, purchases that would not have occurred otherwise. And so Metta has offered for a long time, conversion lift studies, where they do this, they run a geo holdout to help you understand what percentage of the reported conversions are incremental. And it's so hilarious to say like, well, if you'd rather have us just report all the purchases.

We could do that, or we could just show you the ones that we actually created, which is what incrementality really means. It means purchases that would not have occurred 

otherwise that were causal to the advertising.

That is the gold standard of what we all 

[00:08:53] Richard Gaffin: So talk to me about, I think there's two things to dig into. One is Maybe talking a little bit more about why optimizing for an outcome reported in GA is generally speaking, a bad idea. And then the other would be, why are these two things mutually exclusive? Maybe

[00:09:07] Taylor Holiday: So, yeah, and that's a fair question. They don't necessarily have to be. So let's talk about the default general reporting system with Google in particular is Google used to default report on last click. If you were to optimize your media to increase last click conversions, what do you think it will begin to optimize for?

Richard.

[00:09:31] Richard Gaffin: last click conversions,

[00:09:32] Taylor Holiday: Yep. Which is more likely to push you towards what part of the customer journey.

[00:09:37] Richard Gaffin: Bottom of the funnel.

[00:09:38] Taylor Holiday: That's right. So what happens is if you obligate yourself to only count conversions that happen on the user's last, very last click, you are going to begin to move towards the very end of the customer journey. So it is going to, by definition, move your dollars into a stage of that journey that I don't think most businesses actually want.

And so the end result is going to be, now you, this is the problem with attribution is that Google offers lots of different attribution views. So if you're optimizing for Google's MTA view. Which I have found, if you go look at the new, in GA4, their default attribution view is now their sort of Google modeled total impact view or whatever they call it.

That is an MTA solution that hyperweights towards last click, in my experience, when I compare it to different attribution models. So it's still going to give a preference to that stage of the journey, which I think is ultimately any proxy measure. For what you actually care about, which is new customer revenue or overall business contribution margin is problematic and a triangulation of what you actually want.

That's not necessary. What you should just measure is the ads result on your business, new customer revenue, contribution margin. You don't need the proxy. You don't need to triangulate out to a third measure solution. That's every time I see a business get, so that's, that's my concern is you then have to go establish a relationship between that third party measure and the actual business objective you care about because Google revenue, Google reported revenue is not what you're after.

You're after money in your bank account.

[00:11:13] Richard Gaffin: that makes sense. So it kind of sounds like what you're saying is not only the, so the previous issue is that you couldn't optimize for these third party attribution tools results. And now kind of what we're saying is that those results are actually never really, were never really the results that you wanted anyway. So

[00:11:29] Taylor Holiday: That's right. And I'm sure you can hack them into a relationship, but I just think it's unnecessary when the second opportunity of what meta released is. We are going to do a geo holdout and isolate to make sure that our conversions would not have occurred without us. Those are incrementally positive, contributing conversions, which is what I think we should all be after.

So I would just say that of the options, I'm not saying that if you are, as an organization, ruthlessly committed to measuring your ads in GA or North beam, fine, use the optimization towards that. That makes sense to me. If that, if you're confident in those results, cool. But if you get to build from scratch and you're choosing, I would choose the second option, which is incremental conversions.

Now, the challenge with this, it's important to know that Metta has stated that their ability. To determine incremental conversions is predicated on one you've run conversion list studies in the past, or ideally you're running them all the time that data will then inform the optimization setting. And if you've never run one, they will use the aggregate data set on your behalf.

So it is most likely to be most valuable if you are running incremental optimization in compliment to conversion list studies so that they can have data on which customers are most incremental to your business individually. This is the same as measured or house or anybody else is that if you use a default set of incremental weights, they will use those.

If unless you have individual studies of your own brand, in which case that data will inform the optimization. So it is, it is a process still. Of doing testing in parallel to optimization.

[00:13:11] Richard Gaffin: So on the so with the incremental conversion optimization. Is part of what that offers then, cause we talked about, like, so if you use GA, you're optimized for GA or whatever, you're basically privileging the bottom of the funnel is part of what the incremental conversion optimization part of what it does is actually allows you to think about the top of the funnel a little more clearly, or actually drive acquisition in a clearer way than you were able to before or something along those lines.

[00:13:36] Taylor Holiday: Well, and I, I don't know that it's necessarily specifically only related to acquisition. It could also drive incremental returning customer revenue. It's just saying that those purchases would not have occurred without this ad product existing. And so it's really important. And therefore, I think, more likely to not conflate A bunch of purchases that already had a bunch of work done to get them to that very last stage.

So if we go back to the, I did this video where I sort of break down the sort of first principles of of incrementality where it's like the customer journey stage and then the optimization setting and the further, the further down the customer journey, you move from like.

Net new visitors with extreme holdouts all the way down to people with stuff in their cart, like you're going to get less incremental impact if you advertise only to people who already have things in their cart, they're already highly likely to purchase anyways. Right? So yeah, incrementality will tend towards.

The obligation of new customers or causal impact really, which will tend to move towards that. Now, the missing variable is the expectation of efficiency on that incrementality, right? So, and how much volume you're able to produce against that expectation.

[00:14:47] Richard Gaffin: Okay. So it sounds like kind of like a way to, a way to summarize this maybe it's like, what's happening is that. So 0. 1, which is the optimization for third party measurement. The reason that those third party tools exist to some extent is just when iOS four happened. And the narrative was essentially that Facebook's attribution it's in platform attribution is no longer clear enough for us.

We have to have these third party tools in order to understand what's happening in rolling out this optimization. Or rather, yeah, the ability to optimization for the third party measurements, they're saying, yeah, okay, you can now use those tools, but additionally, by rolling out the incremental conversion optimization, they're saying, and now we actually have a much better attribution, a way to understand attribution here than any of these third party tools.

So really, they're saying, like, actually, still, the best way to know what's happening is via our in platform tools. And this is a way that we've vastly improved that, but

you can use your little third party tools if you'd like.

[00:15:43] Taylor Holiday: well, what's funny is it's basically relegated third party tools to just a data point because if, if I'm optimizing for GA or for North beam, why do I need to go look in North beam? Because what you're saying to me is that my reported results in meta will now look like the, they're, they're going to create alignment between them.

Right? So really all its value is at that point is just passing you back an MTA data number. Really? Like that's, at the end of the day, like. A lot of these tools will just get boiled down to a model with a UI on top of it. And the entire value is on the output of the model which is just a number, right?

So it's like that, that in a lot of ways, it's funny that it does, it does diminish that, but you're right. It's also saying. You can do that, but we still think the best way to do it would be to use this incrementality optimization alongside for conversion list studies a year that informs it. Otherwise, we're going to use our repository of all lifts tests and specific advertisers, historical lift tests to optimize the report.

So yes, the point though, is that it's moving towards trust your meta data inside the platform more for generating the business impacts you care about.

[00:16:48] Richard Gaffin: I think what went interesting. So one of the commenters on your. Tweet here, Rishabh. So we'll shout him out here. But he said it reads like one opt or option one is the optimization of a sense of control and opt or option two is the optimization of money. Which he's framed

as not a trivial choice.

But yeah, I think that's really well said is that like, that's always kind of what the third party attribution tool has been doing is giving you a feeling that you know more. Or that you have more information and more information is therefore better. Whereas what the second option is giving you is actually the ability to put money in your pocket.

[00:17:17] Taylor Holiday: That's right. That's right. And it's nice when when someone else says it so that I don't just get accused 

of always being that voice. But yeah, I think that's exactly right. Is that What, above all else, if we just work from a first principle, what you care to measure is your bank account. You care to measure Shopify revenue.

You care to measure contribution margin to your business. That at the end of the day, that's the governing truth. Everything else above that is like an attempt at directing you towards that measurement, right? So my experience is that every organization uses that as a measure. Nobody in the world doesn't look at their revenue.

Some don't look at their profit, but most look at have some sales dashboard. Everybody has that. And then whether you like it or not, every media buyer is looking in the ad account. It's the introduction of the third data point. When it becomes a triangle, then it becomes a complex mess. It doesn't actually get clearer.

It gets worse. So the more that I can tie that a relationship directly between the platform and my business, the better. And this is, I think, going to be a helpful way to do that.

[00:18:17] Richard Gaffin: so maybe I'll say like, or we can kind of leave it with what Facebook or Facebook's language around kind of what you're talking about. About their eventual goal being to move meta towards being a platform where you click the button that says increase contribution margin, and then that just kind of happens. So, as part of

their ongoing work to build AI enabled automation solutions that allow any business to grow, we're evolving our ad system so that the results we drive are more customized to each business's objectives and the way they measure value in whichever external analytics tool they use.

So some

businesses want to maximize total conversions.

Others want to increase the value of every conversion. Understanding exactly what a business values allows us to bring the level of customization to each of their ad campaigns, which is integral to making sure our AI models continue to maximize. So I think previously, a lot of the times it was just like the purpose of running Facebook ads is to get a return on your Facebook ads, but we've evolved to sort of a much more complex way of thinking about that.

And this

[00:19:11] Taylor Holiday: What? Yeah, 

[00:19:12] Richard Gaffin: that.

[00:19:13] Taylor Holiday: that's right. And it's much more honest in the sense that it is true that every business has a different desired outcome relative to their state,

right? Where some businesses are in such a desperate position that what they need is only dollars that make the money today. And they need to be able to optimize for that.

Other businesses have an opportunity to Take a more latent payback period. They want to drive ongoing brand awareness and demand for the future. They're making long term investment decisions. It is different for every business. And this is the same thing that we, when we come in with every partner, it's a question of what do you want to optimize for?

And so you think about our spend in Amy, our model, it's basically this, what do you want to optimize for new customer revenue, new customer contribution, margin, lifetime contribution, margin, lifetime revenue. We, we have an ability to set the ad budget relative to the business objective. This is just a downstream optimization setting the ad account that matches that same principle.

What do you as a business want me to accomplish with this ad product? And that's what every business should be asking. Or that's what, this is why I'm so bullish on the way that Matt is going about this is they're saying you give us the business objective that you want, and we will make our ad product accomplish that thing versus going.

Look over here. We're just going to always report on the same standard view. And you can't change the optimization settings. And we're just going to tell you why it's valuable, which I see some of the other platforms kind of stuck in is that they're not willing to change the optimization to match the business outcome.

And that's really the key here for everybody.

[00:20:36] Richard Gaffin: Cool. All right. Any, anything else you want to hit on this?

[00:20:40] Taylor Holiday: Look, all of these things are subject to efficacy and testing is that incrementality works if it produces incremental results, but it's a higher standard of, it's a higher bar. And so in some ways, I'm curious to see if brands struggle to produce the IRS that they want. And if it leads to frustration or this sense that it's not working, From meta as we increase the standard of measurement.

And so I think it's going to be really interesting to see what happens is that I, I just don't know that for every business there are incremental profitable conversions, given the present construct of their margin profile and creative mix and et cetera. And so I think there may be. An initial pushback to, I tried it and it didn't, it quote unquote, didn't 

work. And so I, I just, it's going to be really interesting to see, but I wouldn't be surprised if the initial wave of responses, what the heck I, it's saying I got a crappy incremental Ross and it's like, yeah, maybe it's revealing to you what was hidden before.

[00:21:37] Richard Gaffin: Well, so it's the clarity of this new tool is going to reveal that meta. Maybe it wasn't working as well for some brands as they thought it was, which is going to

have an interesting impact. Yeah.

[00:21:47] Taylor Holiday: that's right. And that may that may evolve the settings and structure of the ad campaigns. I think it's going to put pressure on things like meta's constant push towards reach and traffic and video view campaigns that I have yet to see produce a good incremental Ross. So. I think it's going to put a pressure on all these ad products to really make business impact, which is what we as an industry want.

We want pressure for performance. We don't want obfuscated results. We want real business impact. And so I think it's a, it's a good standard and it's the highest standard and it's a, it's a positive for the industry.

[00:22:17] Richard Gaffin: All right. Well, speaking of real business impact, if you want us to make one for you, all you got to do is go to our site, come to our co. com, click the hire us button drop us a note. We would love to chat, but I think that'll do it for this week, folks. Thanks for listening in and we will see you all next week.