Episode 789: Why Pausing Meta Ads Based on CPA Is a BIG Mistake!

April 3, 2026

Has your business growth stalled? You need to audit all aspects of your marketing strategy to uncover the reasons. We are giving away our $10,000 audit process to a few brands.
Book your spot: https://www.tiereleven.com/audit

Ever been told to cut your losers and let your winners run? It sounds simple enough, but what if this popular advice is actually hurting your ad performance? In today’s episode, we’ll uncover why focusing on CPA (Cost Per Acquisition) might be misleading and what you should do instead to optimize your ad strategy.

Tier 11’s John Moran walks us through an experiment where we tested ad sequencing and creative strategies. The results prove that cutting ads based purely on CPA could cost you more than you think.

We’ll also break down the right approach to assessing ad performance in the Andromeda era and how understanding traffic stages can completely change the way you think about your ad spend.

In this episode you’ll learn:

Why focusing only on CPA is the wrong way to judge ad success
The importance of ad sequencing and targeting different user stages
How cold traffic interacts with ads and why it matters
The value of ads that don’t immediately convert
Why last-click attribution doesn’t always tell the full story
Analyzing the true impact of UGC (user-generated content) ads
The role of your creative in optimizing ad performance
How to spot high-potential ads even when they’re underperforming at first

Thanks so much for joining us this week. Want to subscribe to Perpetual Traffic?

Have some feedback you’d like to share? Connect with us on iTunes and leave us a review!

READ THE TRANSCRIPT:

Why Pausing Meta Ads Based on CPA Is a BIG Mistake!

00;00;00;02 – 00;00;18;29
John Moran
The Meta Andromeda engine is last click for importing tho first click conversions. So when you import first click and it says hey, this is the amount of people that did the last click. It means that this is getting the highest amount of add to carts per click. I mean, this is a crazy, crazy, crazy and $857. That is a winner for sure.

00;00;19;00 – 00;00;27;28
John Moran
You know what we’re going to do? We’re going to create five new versions of these. What we’ve identified is that Andromeda has a very specific way that it likes to optimize, which is.

00;00;28;00 – 00;00;31;14
Intro
You’re listening to perpetual traffic.

00;00;31;16 – 00;01;06;04
Ralph Burns
Hello, and welcome to the Perpetual Traffic Podcast. This is your host, Ralph Burns, founder and CEO of tier 11. And I am super excited for today’s show because this is one of the most popular shows that we’ve done in the last six months to a year here at Perpetual Traffic. But because it’s one of the most popular shows, I think it’s really important for us to tell you a little bit about how things have evolved since we first recorded this episode back in November of 2025, which doesn’t really seem like that long ago, but in actuality it was.

00;01;06;04 – 00;01;28;19
Ralph Burns
It was recorded in October. It was aired in November, and so much has changed since then. So when you’re listening to today’s episode, there’s a couple of things I want you to start thinking about and keep in mind, once again, Andromeda and Meta and everything going on right now with all of the individual ad platforms are changing. So lightning fast.

00;01;28;21 – 00;01;52;22
Ralph Burns
You’ve got to keep up with these changes, or have your team listened to these shows and have them keep up with it. So if you’re a director of marketing or VP of marketing, this is the kind of stuff that you should be talking to your team about. So the title of this show was Stop Pausing winning Ads. And This now Changes everything with Andromeda.

00;01;52;25 – 00;02;23;26
Ralph Burns
So was a great episode. And there’s a lot of things in this episode that we got right. So the core insight still holds that CPA in-app, a CPA is a poor judge of individual ad performance inside Andromeda’s sequencing model. Same today as it was when we recorded this in late 2025. So cut the losers. Let the winners run was already dead in November of 2025, and has been proven wrong many times over since.

00;02;23;28 – 00;02;48;17
Ralph Burns
In subsequent tests that we’ve done, we’ve now tested upwards of $70 million of our own money on a lot of these strategies here. So this is stuff that we’re constantly evolving and constantly updating. So tier 11 data suite or even Wicked Reports or an attribution software that uses first click Copy Imports for identifying true ad contribution, not just attribution but contribution.

00;02;48;17 – 00;03;14;28
Ralph Burns
Now what does it contribute to the bottom line? The AD itself may or may not get the last click attribution, but how is it contributing? So you have to look at everything as a continuum. Everything as an entire environment, everything as a universe, which includes your meta paid ads, your Google paid ads, your search, your email, your TikTok, your programmatic spend.

00;03;14;28 – 00;03;44;22
Ralph Burns
Even just some of those people that just might find you just directly, which we can track that through tier 11 data suite very accurately. But all of that still holds true in today’s episode. Creative diversification and UGC in particular user generated content. If you’re not familiar with that phrase as a sequencing entry point was exactly right, we still are using a lot of video at the very top of our funnels inside our meta ad accounts.

00;03;44;22 – 00;04;09;17
Ralph Burns
It’s very, very consistent. Top of funnel, middle funnel bottom of funnel. There’s an episode that we just did on the ad lib, which I’ll leave links in the show notes to this, where we’re starting to separate some of these out. Now, this episode doesn’t really talk about that quite as much, but think about how each individual component of your ads contributes to not attributes, but contributes to the bottom line.

00;04;09;19 – 00;04;31;05
Ralph Burns
Because the image ad might be the bottom of the funnel where it’s here’s this thing, buy it now for 20% off. That is not a top of funnel. That top funnel ads are usually ones where there is UGC content, founder videos, stories about the company, something that introduces you to the brand, and you only need one impression inside meta in order to get familiar with a brand.

00;04;31;05 – 00;04;58;19
Ralph Burns
And then it’s the middle of funnel ads that really start doing the heavy work. Oftentimes, those are not the ones that get the last click attribution, though, so you shouldn’t be turning off your ads that are getting lots of spend. As long as you’re looking at everything as a contribution model, which is how it’s all behaving within the total environment of all of your media and non paid media spend.

00;04;58;19 – 00;05;25;04
Ralph Burns
This includes your content marketing, which includes your email, everything else as well as your organic search. So all of that still holds true now. What’s changed since we recorded this is that Andromeda has actually updated again. There was a global rollout in late October 2025, right around the time that we recorded this, which is the Gem update, which we did do a show about, which is generative adds recommendation model.

00;05;25;07 – 00;05;54;00
Ralph Burns
We did a show with, Andrew Foxwell on that. This is sort of the second brain that ranks which ads actually get shown after Andromeda selects candidates for being shown in the newsfeed. So this changes how you think about creative scoring, changes how you think about which ads are your most effective ads. Everyone says just shut off you like my at least effective ads, but sometimes your least effective ads are actually your most effective ads.

00;05;54;00 – 00;06;19;08
Ralph Burns
They might be the ones that are getting the most ad spend. And John talks about that here. Pretty in-depth. Andromeda now analyzes creative at a far more granular level. We have now realized this in the six months since we aired this show for the first time. Hook quality in the first three seconds or hook rate first three seconds is an extremely important secondary in not really primary in many cases.

00;06;19;08 – 00;06;40;07
Ralph Burns
Metric on screen text audio signals creative fatigue prediction hold rate, which is how long people are actually staying after you hook them in. Are you getting your message across in the right way? None of this was fully live when we first aired this episode, and for anyone who has been watching it ever since then, these are some of the updates that you can expect.

00;06;40;07 – 00;07;04;08
Ralph Burns
So in this episode, we also talked about running 25 UG ads simultaneously for diversity. This is when we were still sort of trying to figure out this creative diversification thing. We now just call it creative. That’s our creative strategy. Our first strategy really is creative diversification. It’s not creative diversification unto itself is not a strategy. It is a creative system that we now follow.

00;07;04;10 – 00;07;32;03
Ralph Burns
And we look at these things, look at each individual creative and each individual ad as just doing an individual part of the job to ultimately lead to the sale. It might be one impression on a video ad for someone right to the bottom of the funnel where they just buy the thing, or it could be one impression at, top of funnel for a video ad, but then 17 other impressions or clicks or touches or views in the middle before they actually do that last click.

00;07;32;03 – 00;07;52;28
Ralph Burns
And it might be over on Google, it might be over on Amazon. So as long as you’re looking at everything globally and how all the channels interact with each other, that’s how you win in today’s digital marketing age. And last but not least, the feeder strategy, which was originally a Google strategy that we talked about many shows. We’ll leave links in the show notes on that has now been adapted for meta.

00;07;52;28 – 00;08;16;08
Ralph Burns
And we talked about this in episode 757, using separate creative campaigns. Now to train the algorithm using specific SKUs and not just the product it defaults to here actually. So we’re starting to run campaigns, and we’ll leave another link in the show notes for this one. For specific SKUs we have specific campaigns. So we didn’t really know this back in October and November.

00;08;16;08 – 00;08;46;02
Ralph Burns
That’s how fast this stuff is evolving. So still pausing is still true. The news structure matters as well. Still true. Don’t pause ads based upon in-app CPA alone. Yes, look at your in-app CPA, but also see how it’s affecting the entire ecosystem. But in 2025 and 2026, the, radical change is that we are really starting to look at everything in the top funnel, middle funnel, bottom, funnel way in which we’ll be doing more shows on that coming up.

00;08;46;02 – 00;09;17;02
Ralph Burns
So the bottom line is this is fewer campaigns, more ad diversity within them. And that includes creative diversification. So without further ado, let’s get into today’s episode with myself and John and sort of keep some of these things in mind. What we got right, what we didn’t get right, we didn’t really know at that point in time. Like I said, everything is evolving and we’re going to continue to go back to these shows, which have gotten lots of views and lots of downloads and revise them as this new information takes hold and we continue to test it with our own money.

00;09;17;02 – 00;09;18;07
Ralph Burns
So take it away.

00;09;18;07 – 00;09;39;21
John Moran
Boys were testing different ads, different creatives, different users, different targeting, and we’re getting the sequence more streamlined. So when it says cut your losers, let your winners run. The losers are not CPA defined like in my opinion. And you’d correct me if I’m wrong when they say cut the losers and let the winners run. You probably heard that from a CPM perspective.

00;09;39;27 – 00;10;05;07
John Moran
It’s like a CPA perspective is typically how people say, well, a winner must be in CPA. But again, in a hierarchical model where you have, you know, sometimes one I mean, if you follow this one, two, three, four, five, six, I missed one, but, you know, seven and eight, basically how those users are moving through the ad sequencing, if they click and buy, it could be on the first one before they retarget them.

00;10;05;07 – 00;10;22;02
John Moran
It could be on the second one before or like after they got hit with the first one. I mean, you all get one checkbox out of all of these here that says that this is right. This is why I’ve always said since day one, focus on spend. Don’t focus on CPA attribution model is poor. Spend model is actually 100% accurate because it’s being spent on.

00;10;22;04 – 00;10;42;08
John Moran
If you look at the spend descending and you look at the CPA, it’s like, well, John, that one’s better than that one. That one’s better than that one, that one’s better than that one. Like, you know, this one here, this one here and this one here. John, these are doing better. Like if you look at like they got a 263 compared to a two and you got a 251 compared to 279.

00;10;42;08 – 00;10;58;13
John Moran
This one actually is something that we’re reiterating. The same would read boost. And it’s like, man you got a 661. That’s pretty terrible. So why are you not spending more on those. Why are you still spending on that? And this is where CPA is a poor judge of performance. We don’t think what.

00;10;58;13 – 00;10;59;12
Ralph Burns
Exactly what we’re talking.

00;10;59;12 – 00;11;17;29
John Moran
About. The green group is. UGC the blue group is non. UGC their product highlights. What we’ve identified is that Andromeda has a very specific way that it likes to optimize, which is showing all ads to all users and developing a sequence of what needs to see first, second, third, fourth. On top of that matrix, you’re layering in your pain points, your hooks, your offer.

00;11;17;29 – 00;11;35;18
John Moran
So you’re you’re trying to basically have product market fit. That’s all you’re doing when you’re doing that with creative. Right. So but not everyone is going to see everything first, second, third, fourth. Now there wasn’t a click the first, second, third, fourth. Now everyone’s going to buy from the first, first, second, third, fourth and click. So we have to identify where the trends analysis that we’re doing on the users.

00;11;35;21 – 00;11;50;29
John Moran
And you’re going to notice that in the green. And I’m not gonna use the same color green because I’m not that good in selecting RGB wheels. But you’re going to be able to see that this has some fairly good amount of cold traffic. He got 63, 77, 71 now and then in the bottom row about the same 7074.

00;11;51;00 – 00;11;54;01
John Moran
Okay. Good. Now you’re gonna notice do visit percentage.

00;11;54;01 – 00;11;54;21
Ralph Burns
By the way.

00;11;54;23 – 00;11;56;29
John Moran
Yes. Really high is a percentage exactly.

00;11;57;00 – 00;11;59;29
Ralph Burns
60 to 70%. Like that’s really good.

00;12;00;04 – 00;12;21;02
John Moran
Oh exactly. Yeah. That’s that’s actually very good. That’s a that’s a hefty amount of cold traffic in terms of meadowland. Correct. But you see the blue is in the 40s. So now you’re seeing okay more first time visitors on UGC. More second time visitors on the product ads okay. So we’re starting to establish a trend. Now let’s dive a little bit deeper into that trend.

00;12;21;07 – 00;12;42;29
John Moran
Now the first trend that we’re going to notice is okay what how many clicks before how many add the cards. So we got 10,000 clicks 189. Add the cards okay 63,000 clicks 62 add the cards okay 24,000 clicks 79 add the cards. All right. So kind of, you know, all over the board, but lot of clicks, tens of thousands for not a whole lot of conversion.

00;12;42;29 – 00;12;58;21
John Moran
Not a lot of add cards. Right. That’s because that’s the first interaction that they’ve had with that product that add our brand, etc.. It is like hi, I’m John, next thing out of the customer’s mouth is going to be like, I’ll buy one today, right? We’re skipping the middle part. So next we have to look at. Now how did the blue do.

00;12;58;22 – 00;13;00;17
John Moran
The blue 1%.

00;13;00;19 – 00;13;14;00
Ralph Burns
Thereabouts. It’s about 1%. Well maybe not the 24,000, but it’s about 1%, which is about that makes sense. Like we usually say like 1% of your traffic is going to click and buy. In most cases this is an add to cart. So it’s not entirely accurate. But yeah.

00;13;14;03 – 00;13;29;14
John Moran
Yeah, it’s like we’re starting to see that the trends analysis. Now let’s go down to the blues. Now after they click they look look at the product. You know they’ve maybe out of the car. Most of them didn’t. But you got a lot of understanding about what this is. So actually let me let me I did everyone a disservice.

00;13;29;14 – 00;13;41;29
John Moran
Let me back up one step. What has happened so far that we discussed is we’ve shown a UDC ad to fairly cold traffic. We say, hey, look at this dog. This dog was sick and now they’re on the mend. And look at the before and after that we’re kind of like, you know, not that great. Now they’re happy, healthy.

00;13;41;29 – 00;13;57;29
John Moran
When they said, wow, I love success stories about dogs being healthy. I wonder how that happened. I’m going to click the other one. Say, you know what? Great for the dog. Happy fun story I’m scrolling on. Great. So the people like the people that click, they say, that’s the product. They’re learning about it. And they say, that’s interesting.

00;13;58;00 – 00;14;19;15
John Moran
You know, maybe I’ll come back to this. I’ll look at this later. They leave. Now, the second thing that happens most often based on percentiles of user journeys is we identify that the product that has most often the second click, because we have a lot more, more traffic. We also have a lot fewer clicks. Now these the these fewer clicks will highlight these the same.

00;14;19;15 – 00;14;46;19
John Moran
The highlighting the green. Now we have 303,752 compared to 76. That’s a vast difference because if we look at 24,000 got a 79, only 3000 got a 79. So you can see that these people are much more warmer. Right. And there’s a lot fewer visits but a lot higher order cards. Look at the next one down 3900. I just don’t think that seven red boost is working at 664.

00;14;46;19 – 00;15;05;14
John Moran
Remember the Meta Andromeda engine is last click. Yeah we’re importing though. First click conversions. So when you import first click and it says hey this is the amount of people that did the last click. It means that this is getting the highest amount of ad to cards per click. But that is not where they started or where they necessarily ended.

00;15;05;14 – 00;15;15;17
John Moran
Okay, that’s fine, but I’m not going to take one of my my biggest add to cart performers and shut it off because the CPA isn’t there.

00;15;15;19 – 00;15;19;04
Ralph Burns
If you didn’t have tier 11 data suite plug pug, tier 11 data.

00;15;19;04 – 00;15;20;27
John Moran
Suite, of course,

00;15;20;29 – 00;15;24;01
Ralph Burns
You would have made a very large error.

00;15;24;07 – 00;15;36;17
John Moran
We probably would have paused. Yeah. I mean, have you if all signs pointed to. Well, yes, we would, this would absolutely be something that we would end up pausing. But this is something that we’re going to be continuing on with further.

00;15;36;25 – 00;15;40;19
Ralph Burns
Right. So it’s early days anyway. Like the ads only been live what, a day or two.

00;15;40;23 – 00;15;55;06
John Moran
Yeah. Yeah. It’s been like three days now. What’s the first thing. I launched it in the first three days and it doesn’t look that good. So I pause it. Every frickin person I, you know, I’ve launched the campaign or launch that. And then the first week I start hacking off as they’re having low CPA or high CPA again, the old way of doing things.

00;15;55;06 – 00;16;17;05
John Moran
Exactly. You cannot do it that way. I mean, this is a crazy, crazy, crazy $857. I’m like, I don’t care. Us delivering the highest amount of ad cards we’ve ever seen. I mean, that is an I’m fantastic and majority of them are cold to three causes of a cold. I would have taken 200 cold, add to carts and just chucked out the window because Andromeda didn’t say last thing that they ever did ever to the website was there.

00;16;17;06 – 00;16;25;03
John Moran
Like if you just said, hey, what do we increase? We increased. We doubled this last two weeks. Cool. What happened? And Cat is still below goal. All right keep going.

00;16;25;03 – 00;16;26;11
Ralph Burns
Thanks everybody. Keep going.

00;16;26;11 – 00;16;42;20
John Moran
That’s that’s pretty much it. Like smart and cool. There are 72 more new customers. Now if I didn’t have data suite I would have killed my middle, which absolutely would have hurt. If I lose 200 new add to cards a week. I’m totally feeling that now. Could it be picked up by another ad? Perhaps? Could the other ad do better?

00;16;42;20 – 00;16;56;28
John Moran
Perhaps. But you know what this is telling me right here is follow Andromeda. Which means is it a winner? Yes, in my book, when I don’t measure by just last click CPA. Otherwise my dumb ass would be just scaling brand campaigns all the time because it’s last click. It looks like great.

00;16;56;28 – 00;16;57;20
Ralph Burns
For last click.

00;16;57;20 – 00;17;12;25
John Moran
Yeah, right? Yeah. This is where they ended. Let’s just end more cool. Should we start more. Not on. Don’t just kill the starters. Let’s just end more there. That’s what you’re doing. Yeah. So when we look at this is is it a winner. Absolutely. 4000 AD the cards 200 or 4000 clicks, 232 out of cards is the biggest add to cart.

00;17;12;25 – 00;17;18;16
John Moran
Is this literally more than a quarter of all of my add to carts? Yeah. Oh yeah. Yeah. That’s that’s your bottom.

00;17;18;16 – 00;17;19;25
Ralph Burns
Of funnel ad right there.

00;17;20;02 – 00;17;25;10
John Moran
That’s like kind of middle bottle that is a winner for sure. You know what we’re going to do. We’re going to create five new versions of these.

00;17;25;10 – 00;17;50;14
Ralph Burns
All right. Hope you enjoyed today’s show. We’re going to go back into the archives quite a bit and revise some of the episodes and give our own take, myself and John, in some cases on some of the newer nuances with all of this stuff that we’re doing right now with meta and with Google. So, hopefully when you listen to this episode or you might want to listen to again, keep those developments in mind.

00;17;50;17 – 00;18;22;24
Ralph Burns
The principles that we talk about here are not based upon last click CPA in app. So let Andromeda sequence the creative. Use your data to see the first click behavior in the last click behavior. All that is foundational and is still accurate. What’s changed is that the scale of what’s possible now we’re seeing this, just be able to scale up brands like we’ve never been able to experience in the past, and we’ll do some case studies on that in coming shows.

00;18;22;24 – 00;18;46;04
Ralph Burns
And you now have the precision as long as you have all that click data and a very reliable attribution platform. Our preferred attribution software right now for mid-level businesses is Wicked Reports. But we power that and make that information even more granular, that data even more granular. So we literally do not miss a single click so that we literally do not miss a single click.

00;18;46;04 – 00;19;06;26
Ralph Burns
By using edge servers. We combine that together, with our friends over a lot out to create the tier 11 data suite. And that’s how we’re able to get these types of results. But you can formulate these pretty closely yourself even without all of that additional tech. So like I said, all the links that I mentioned here on today’s show, as well as references to other shows are in the show notes over at Perpetual traffic.com.

00;19;06;28 – 00;19;24;04
Ralph Burns
Make sure you leave a rating and or review wherever you listen to podcasts. It allows us to get this show out to a wider audience and teach people how to do digital marketing the right way in 2026 and beyond, through metrics that matter and growth at scales. So till the next show, see you.

00;19;24;06 – 00;19;26;13
Outro
You’ve been listening to Perpetual Traffic.

Episode 789: Why Pausing Meta Ads Based on CPA Is a BIG Mistake!

Mentioned in the Episode:

Listen to this episode on your favorite podcast channel:

We appreciate your support!

Connect with John Moran:

Connect with Ralph Burns:

READ THE TRANSCRIPT:

Why Pausing Meta Ads Based on CPA Is a BIG Mistake!

More Episodes

Episode 793: Scaling Ads Is Killing Your Profit (Here’s the Fix That Works)

Episode 792: 156% More App Subscribers in Six Months! Here’s How We Did It

Episode 791: 5 Missed Forecasts. Then One Budget Shift. Then 4 Straight Hits.

Sign Up For News & Exclusive Opportunities