What iOS 27 Means for Meta Ads Before Black Friday

Let me saw this way back in 2021 with iOS 4. Oh, I don't speak about that here. Ruffer time. But when Apple does this, Apple is cool in that they like privacy.

But sometimes that can be overly private. And to the point where you lose signal, you're no longer able to see if a purchase conversion happened or tie it back to a particular app, sure. So when you lose that signal,

do you, first of all, might have worse looking

performance on meta? This episode of the Ecommerce Playbook is brought to you by Universal Ads. Here's a question.

“How much of your TV budget actually makes it to the big screen?”

On most TV platforms, it's not all that, not even close. Platform fees, data fees, optimization fees, it adds up fast. With Universal Ads, you can ditch the middleman. It's 100% self-serve TV advertising,

built for brands of all sizes with zero platform fees, and zero hidden markups. You get your brand on premium networks like NBC Universal Fox and scripts, right next to brand-save content, your customers are already watching.

And right now, new advertisers get $2,500 and matched ad credits when they spend $2,500. That's double the TV for your first campaign.

Get on TV without the fee.

Go to www.universalads.com/middleman and use code, uacctc when you sign up. That's www.universalads.com/middleman, code uacctc. All right, hello, hello, a sub-steve, how are you? I'm great, Tony.

Oh, you're doing okay, well, that's good. Glad news, I'm glad to hear that you're doing okay, 'cause we have a potentially interesting tracking curiosity that's bubbling up in our world. And the topic of the day is iOS 27

and the new release, the new Apple operating system released. And what if anything, we need to be mindful of for our advertisers? So that's where we're here for, so I'm excited. Are you ready? I'm excited about this, too.

“I think it's an important thing now that it's crept up”

and we're coming up on Black Friday and Cyber Monday. This is something that really kind of dial in. Check your accounts. Yeah. Make sure things are operating correctly.

For sure, coming into an important point. Time of year. Yeah. So I think like the shape of the conversation that I'm hoping to have today is kind of like what's happening?

When did it happen? Kind of like the timeline and then what's the impact? And what what what what what what what are we folks are? What do we think our our advertisers and our our audience should be paying attention to?

So let's start with kind of like the the big picture. Let's talk about what what what what changed. What's what is happening with iOS 27 specifically? And help me help me understand it in simple terms. Right.

So Apple obviously updates its operating systems. When we saw this way back in 2021 with iOS 14. Oh, we don't speak about that here. Oh, rougher time. But when Apple does this, Apple is cool

and that they like privacy.

“But sometimes that can be overly private.”

And to the point where you lose signal, you're no longer able to see if a purchase conversion happened or tie it back to a particular act. Sure. So when you lose that signal,

you first of all might have worse looking performance on meta in that you no longer have these conversions that actually happened and that is not seeing them. And then you have a worse feedback loop. So meta is not able to see and determine that a purchase

happened. So they lose the ability to determine who's a good customer to potentially show your ad to right or their ability to do so is decreased since they're missing some of that. Right.

So sort of two kind of two impact vectors, if you will, one being kind of let's call it attribution. So it's the platforms reported impact. And then the other being targeting. Yes, great.

Yeah. So signal impacts impacting those two things. I think for me when I think about it,

I'm always trying to wrestle with the idea of,

do you actually have fewer people buying aka? Is there an attribution problem? Or do we have like a targeting issue or are we actually reaching the last people or are reaching people less efficiently?

So it's interesting to separate those two things. So let's talk more specifically. So I guess maybe I have been in the weeds here for or less whatever, trying to be able to articulate this change.

Like what does actually happening canically with iOS 27? And also how does that compare to the unspeakable iOS iOS 14? So let me hit you with this.

And you tell me where my understanding is right or on or in different. So iOS 27 is the particular change relates to technology called WebKit, which is a browser like a browser rendering engine

that's specific to Safari.

So it's really browser based. And there is an ecosystem of companies in the world called Identity Graph Providers. Or essentially companies that sit in the background of the internet.

And when you log into a website or fill out a form, your email address, as one example, is getting captured in this database. And these identity graph providers use that to sort of connect dots in the advertising ecosystem

between a person and an added interaction. So as I understand it, the iOS 27 focus was focusing on WebKit and focusing on detaching these identity graph providers ability to capture that information. Help me say if that's right, wrong and different,

or if you describe it in a different way.

- That's basically right, yes, absolutely.

- Wow.

“- You read up on it quite a bit, I think,”

since we've had this issue, basically right. I think that's the best compliment you can get from it. (laughing) - Yes, I didn't dig into as many details as you honestly. - So I'm respect to this.

- Okay, cool. So I think I have the rough shape a bit. I want to try to relate it to iOS 14, 'cause they are both privacy related but sort of mechanically different.

iOS 14 was a requirement for app users in the iOS environment to explicitly give permission for various tracking. So the scope of the iOS 14 change was in my simple brain like just much bigger, much larger scope.

I have to say that this is not impactful

or not potentially something that we should be concerned about

or mindful of, but just like as far as like Tony's ability to sleep at night, the iOS 14 thing, price still gives me nightmares. I remember it happening like I was in September, like leading up to this type of year,

however many years ago, we got the scope of this doesn't feel as big yet still. We are kind of like scratching at.

“What are we seeing any patterns in the media platforms?”

It's like conversion rate, like are there maybe there's a little smoke here, is there any fire? So I'm curious if you can sort of give me your POV on what if anything are you seeing in the data at this point in time and what's jumping out for you?

- Well, let's start looking at the dates too. - Yeah, that's great, that's kind of it. - This is an important part of the story, sorry. I jumped ahead, I got to data my apologies.

- So the iOS 27, the dates, how is this actually

rolled out into the universe? - So they kind of introduced it on September 14th. - Yeah, so they had about a 3% adoption at that point, but it's kind of been pressed a little bit more. In fact, I think you've said on your device,

you've recently your own personal device, you've recently received a, "Hey, upgrade to iOS 427." - That's right, a push notification, a push notification on your own device recently. And it really kind of started ramping up,

I think October 8th is when they officially like switched over and started pushing from background to push notification, exactly. It becomes like the recommended upgrade. - Yeah, at that point in time,

instead of just also available. - Yeah, so let me play that back to you just the shape of it. So September 14th, like a little over a month ago, it became available, iOS 27 was available, which you'd have to go into settings

and actively pursue the download. And at that point in time, low adoption, 3%, something, whatever, since mid-October, I think you said October 8th, it started to become a push notification

for more and more users. My own anecdotal experience, I got the push notification. I did the update on the 19th. The point being, like, I think maybe the way to sort of

visualize it is the shape of the curve, the adoption curve is probably increasing. - Yeah, exactly, it's not like a step function where all of the sudden everyone was forced to be switched over, it's, I mean, Apple has a massive user base.

So it's going to be a climb in adoption of this iOS 27th. - Got it, got it. - So, and at the end of September, I believe the data we were looking at is around 26% of users.

- Okay. - Had updated. - Understand that, obviously, it's 21 days now,

“22 days now, since we're recording this on the 22nd, I believe?”

- Yep. - If you were a D to C brand and you're only running at on social, you're fighting for the same crowd feed as everyone else. So why not show up on the biggest screen in the house

with ads that can't be skipped? Before you say, TV's too expensive, here we are. A lot of TV platforms have a middlemen taking a cut of your budget before you're at ever-airs, but with universal ads, you can ditch the middleman

and put every dollar of your budget on the big screen. Don't have a TV idea, they can help you get one made with self-serve tools in creative partners. So get on TV without the fee, head to the link in our show notes, and use code UACTC to get $2,500 in matched ad credits

That's code UACTC.

- So that started adopting that.

We didn't necessarily see on September 14th, like a drop in your version, right? - Yeah. (laughing) - Not immediately, it wasn't as we're talking about,

like a step function adoption and a step down in conversion. - Yeah. - What we did see was around October 5th. - Interesting, the decrease. That's when we started seeing a decline in conversion rate.

- Okay. - On meta. - Okay cool. - Not cool, but on the right. - And that particularly around the eighth

that became more apparent because it was bold, significantly below last year,

“and I think down like 15% compared to last year”

from the eighth to like the eighteenths.

So in the last set of data, I looked at those 11 days,

down about 15% year over year, whereas it was down before, like a little bit, 6% I believe, but down 15% really caught our attention. - Yeah.

- And that last a lot of days. - Yeah. Okay, so let me recap here, so plenty of bags. So we have iOS 27, WebKit, Identity, Browser, different pathways, and iOS 14 for sure,

different mechanics, both privacy related. We have the shape of the adoption, which started as a background, so if you could go do it, move to push notification and increase the curve.

And then we have, we're beginning to see the shape of the data. And I'm sorry, the numbers were the comparisons were year on year, CVR, 15% on meta, which is like, to Steve's point,

that starts to get our attention pretty rapidly. So I'm curious if you and I earlier this week started to explore, well, what do we see if anything on the other platforms? Was there anything that popped out from that?

Matt initial analysis? - I really looked more ticked off in that initial analysis. - Yeah. And it was down slightly. - Because that's another platform we were seeing that could potentially, would maybe be impacted equally.

- Yeah, it was rough. - Didn't really see that much impact on Google. And Google, everyone defaults typically to click date for the attribution of the purchase.

So it always kind of lags on conversion.

So I can't tell as well on Google if the most recent, like, great, the last five days have been down, but that's just because you're not getting conversions counting that'll happen in the future. - Yeah. So recent data always looks

a little bit rough around Google. So I would not Google data doesn't have me concerned right now. Got it. And TikTok is a little bit more sparse and that we need good year over year comparisons

for that. And we don't have as broad of our data set putting a lot of money into TikTok on a consistent basis. But yeah, there's a solid, there's dozens on there, but it's not as pervasive.

“I think I don't know if there's any of our 200 brands”

that are not on that. 200 plus brands. - Yeah, for sure. - Yeah, represent some majority of our portfolio. - So let me see if I can say it back to you.

- So the early stage signal that we have is strongest on meta. - And the pictures last clear on Google and TikTok due to the scope of how much signal we have to relate. - And I think for me,

like, I'm still kind of like scratching my head around the, like the specific mechanical pathway, like the failure pathway that could present itself in lead to. So if I one hand were saying,

okay, we have these changes in the ecosystem, right? And then on the other hand, we're saying like, okay, maybe we see some CVR stuff and meta potentially. I am racking my brain around like,

trying a specific example of how that could be the case. Okay, so I wanna, and I wanna give like, I wanna give you my understanding of what I believe is still currently stable in the environment and then we'll go to the failure pathway.

“So let's say, we got Jane, Jane, find something, okay?”

Shopify records the sale and the, in Jane's meta account, there is a pixel and a conversion pixel, it's also installed on Jane's website and then there's a copy integration. Shopify records the sale,

the browser sends the conversion signal directly to meta, the Shopify copy integration sends that same conversion signal, the data matching is happening, the server server, on the identity pieces that are capturing Shopify and transmitted directly

to meta, including email and other PII that's hashed between those two. Essentially that, let's call like BAU setup or maybe like, maybe I might describe that it's kind of like the best and class setup, right?

We get it, like there's no, there's no issue there, right? There should not be, right? Yes, right, right, right, right, right. Good, thank you, I feel better now, yep. So let's sort of explore

some of the potential failure paths, right? Because, and I'm just gonna talk this out loud and you got check me and bring me back down to hers whenever I'm like, you're sort of like, don't know, don't you can't say that, that's not supported

Okay, so, all right, so we had like,

kind of a really good setup where we have Shopify, we got the meta sales channel installed, we're firing the pixel with the sales channel, we're, we're bringing the copy into the sales channel, really tight.

Okay. Let's pretend now, but we have a different type of setup. We have, whatever, non-shop provides side, and we got a pixel, but we have no copy, okay? So we have no server events.

In this particular scenario, Shopify sees the sale, the meta pixel picks up the conversion event as well, but for whatever reason, the identity piece of that transaction is fucked, is murky, the email, whatever, it gets messed up.

And there's no connection of identity.

So now what's gonna happen in this scenario is meta

is going to see the sale, but it's not gonna be able to match it back to a specific ad. How is that sound to you? Yes, or a specific person individual,

“because then that's also important on the feedback loop”

to determine who is a good customer. So this would lean over into like less, and so in the beginning, we sort of talked about two kind of dimensions of how this could affect you. One dimension is reporting on conversion value.

Like how many sales happen and what the value was. Another dimension is informing targeting to the future. That example I gave you would be, like the sale would still get reported, meta's rawness would be sort of representative,

but the ability for meta to sort of recognize that specific user to architect them, to profile them, to expand the audience would be limited. Yes, yes, okay, all right, which could potentially lead to sort of kind of what you're

getting at, which is maybe the ultimate point, aka lower CVR fundamentally, right? Okay, cool, all right, I got it. So one particular failure mode is in core no server events, let's put it that way, right?

Yes, no server events or in some way like poor server events. Yeah, and this can sometimes manifest in different ways, especially if you have subscription or particular purchase or conversion events that you're looking for.

So that complicates it when it's not just a simple one-time purchase, yeah, yeah, for sure, yeah. We've had all sorts of fun puzzles around conversion, can we be tracking with the subscriptions?

“Yeah, so that's kind of, I'm sorry, that's what you have”

to deal with on a regular basis.

Well, yeah, yeah, well, signal has always been like,

I've always been kind of like a ads nerd. I don't know, I like signal and catalog and all this kind of stuff, it speaks to me, but can we be tracking with subscription and topic for a different day? I could go on about that brush.

But I wanted to kind of come back to something that you said, because I think this is one of the core takeaways for us for CDC, like what we're doing in our portfolio right now is this is kind of like best practice hygiene stuff, like, always, but it's like, I think particularly important

when we have these big moments in time where there's technology shifts in this ecosystem, it is worthwhile to talk about Shopify has, in the Shopify meta sales channel app, there's different levels of integration.

So different settings you can choose all the way up from maximum data sharing, which essentially, you can think about that, it's like sending email, phone number, address, all of it hashed, right? So all of it, like, no PI, but hash down to,

there's three different settings. And so our position on this is where it is applicable and appropriate for you as a brand. We want to share the maximum amount of information between Shopify and the meta pixel.

And we use the Shopify app to do so. And that this whole point of this is to ensure that this matching can happen so that we can connect a person, a purchase to a person, and thereby have really good attribution reporting in Shopify, but also a really good at targeting.

So the point being is, if one of the failure pathways for this iOS 27 thing to potentially have an impact on your advertising program, one thing to look at is your browser, your pixel, and your server events. And then if you're on Shopify, I would go have a double click

and do that specific data sharing setting and making sure that it's my recommendation set as high as you are comfortable with. And he saw it's on that, that sounds right. And your bottom work experience on the media side

and how that should be set up. Ever say, turn on, ever say. Yeah.

“Yeah, so I think, okay, so that's one particular pathway.”

I think another thing that we do frequently is like a housekeeping thing, simple thing to do, find surprising things all the time, and this is something I would recommend. Everyone does, that's listening, very simple.

Look at the number of purchases that you see in Shopify.

Those numbers should be, they don't have to equal exactly,

“but they shouldn't be, here's an example of a situation”

that is obviously wrong. You have a hundred purchases in Shopify. You have 200 in meta, you see double. It's a clue, something's, something's goofbally, all right. So the point being is, like, one of the things

that you want to, like, just stress tests. In this moment in times, good practice, just generally going into a Q4, Black Friday, September, Monday, but specifically, post a big iOS change.

Going to Shopify, count your purchases for the last 30 days. Going to meta, to the meta pixel, and look at the number of purchases that are being counted for both web, pixel events, and server. The number should be pretty similar, yeah?

Yeah. I think that's great. So, big pitch, zooming out. This is not iOS 14, as best as I can tell right now. Thank God.

This is a change to the ecosystem,

specifically with Apple, which always gets our attention.

It's a very specific browser-based third-party, third-party identity graph ecosystem kind of thing. If you are using what we would consider to be the best-in-class top-tier setup, which is Shopify, sales channel, pixel and server events, maximize data sharing,

you're going to mean pretty good shape. Would you agree with that, Steve? I would agree. But I'd also agree with your first kind of point in that it's probably a good time to check those numbers,

to audit yourself, to give in that this change in iOS, and we're in Q4 heading towards the biggest time a year for most brands. Yes, please do an audit of your signal up and down. So, do that.

And if you're in this setup, then I described Shopify, sales channel apps, pixel and copy firing through the sales channel apps, maximum data sharing, you're going to be in pretty good shape. Every sort of step away from that sort of ideal setup

creates an opportunity for this particular security update to have more of an impact on your brands specifically. And I would investigate every piece of that chain that is outside of that colleague V1 setup. The places that you're going to want to investigate,

the things that you're going to want to look at as count of purchases against count of track events. In meta, I would also look at your event match quality as another indicator for up and down the event stack, how sharp is your signal?

Do the audit.

“And I think the sort of fundamental takeaway”

for me at this moment in time is check the plumbing before blaming everything on Apple. We have something worse in best game. We're going to keep pulling on it. But I don't think at this point in time,

I don't feel like this is an explanation for like global falling conversion rates across the media stack. It is something that we want to keep pulling on. But yeah, that's my two cents.

What's your takeaway? Yeah, I mean, we are also just after Amazon Prime Day. So I kind of almost expect a little bit of drop off.

And October always has its rough bumps.

It's a tougher time of year for businesses. But it's a time to kind of straighten this out. Make sure everything is correct. We could potentially run an incrementality test. That was the other thing we talked about.

Right, but there might be a strong impetus to hold off until January, because we're at like 26% adoption of the end of September. Maybe we're a little bit higher now, but I don't think full adoption has happened yet.

So if you do come up with a number now, it's not indicative of how it would look with everyone on Apple being on OLS 27. That'll probably happen by early January. Yes.

Early next year is what Apple is talking about. Yes. Yeah, and the Achilles seal of all incrementality testing, which is something I believe in deeply, is it's a point in time measurement.

Right. So I mean, I think my philosophy on this is like at all times,

“you should be running a geohold out test.”

I think you and I are having this friendly debate on going, like I want to run sequential, not what's the opposite of sequential. Kind of her, it couldn't current tests. And Steve's like, no, you can't do that.

So I'll be we can, it just requires a lot more work. And a lot more thoroughness to make sure that there's not kind of crossover impact. Yeah, I think from one test to another. Yeah, yeah, yeah, yeah, yeah.

Thank you, Steve, for keeping my tests clean. But let me say back to you what you're saying here.

So which is we can and we're always eye testing.

We just had a conversation about this yesterday.

we have is our own meta and that testing is concentrated

“into a seven day click attribution, roast bidding.”

And we're always looking at expanding and deepening.

I would say broadening and deepening are our benchmarks, our eye test factors around meta.

“And I think this eye list, eye list changes”

are a really good point for us to bring that topic up again.

So right, we'll be off to the races and resetting that will come back to the pod here and let everybody know if we find anything material.

But yeah, other than that, listen, check your signal.

I know it's boring, okay, but go to go do this. This is a simple thing. Anybody can do this. Everyone can do this. Go counter Shopify purchases for the last three days.

You got that, go into meta, go into the event, page, look at the purchase count pixel and server. Make sure those numbers are pretty close. Anything else?

“I mean, that's what our teams are gonna go.”

That's what we're up to. All right, happy hunting y'all, fourth quarter, carry on.