YT YouTube2Text

YouTube2Text → Transcripts → New BEST 2026 Meta Ads Course After Andromeda

New BEST 2026 Meta Ads Course After Andromeda — Transcript

by Professor Charley T · 14,628 words · 2,127 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Everything you've learned about meta ads
0:02before Andromeda is now working against
0:04you. If your costs are climbing or your
0:06results feel unstable, the algorithm
0:08shifted and most people haven't caught
0:10up yet. This video is a full course and
0:14I'm breaking down the five disciplines
0:16you need right now. Account setup, the
0:19best ad to run, how to read your data,
0:23creative testing, and scaling. All built
0:26for 2026. It's timestamped below. So
0:29save it and come back to it.
0:32I'm opening up my ad account where I
0:35spend my money. I want to share my
0:38screen real quick and show you the exact
0:40structure and more importantly why we
0:43structure it this way. This is the
0:45modern version of the one campaign
0:47method which we now call Andromeda 1. Ad
0:51set one is the control. This is our
0:53benchmark. A small set of proven ads and
0:55ideally there's a good mix of images and
0:57videos. These aren't winners forever.
0:59They're just the most stable ads that
1:02are good enough to scale from and rely
1:04on. Ads sets two and three are tests.
1:07Each one contains a single 322 ad. And
1:09these two ad sets should be thought of
1:11as an AB test. And these ads have one
1:13specific job to improve or even replace
1:15the worst ad inside the control ad set.
1:17Also, in a little bit, I'm going to show
1:18you exactly how to identify the worst ad
1:20because I think most people pick the
1:21wrong one.
1:26We're not testing to find the next
1:29winner to scale and abuse. Remember,
1:32Facebook ads are a team sport. We want
1:34to find the worst player on the team and
1:36replace them so that we can win more
1:39often. Now, let's talk about how to
1:41judge a test. This is the part most
1:44experts get wrong and this is the most
1:46important part of scaling any account.
1:49We're not judging the test by any
1:51individual ad metric, CPA, rorowaz, CTR,
1:54hook rate, none of that. Those are
1:55diagnostics. The goal is money. We only
1:59need to ask one stupid simple question.
2:02Did the campaign's profit volume go up?
2:04And by that, I mean total revenue minus
2:07total ad spend. If yes, we just created
2:10more money to reinvest so we can scale
2:13the budget. If no, the test is a loser.
2:16That's it. Real quick, if you like this
2:17video, please go ahead and like this
2:19video and subscribe so you don't miss
2:21anything else and share it with all of
2:23your friends. YouTube loves that and
2:24it's how we help put fact to fiction so
2:26that everyone can see more success and
2:28less stress. Also, check out Disruptor
2:30Academy. We just launched a brand new
2:32course on Andromeda 1. Literally soup to
2:35nuts. Everything you need from setup to
2:37scale, automated rules for low budgets,
2:39high budgets, literally everything you
2:42need to see more success and less
2:45stress. The link is right there. All
2:46right, that being said, let's get back
2:48to it. budget behavior is the signal.
2:51And the three-step testing process I'm
2:53about to show you is where Andromeda 1
2:56gets really powerful. After we launch
2:59the test ads, we wait. And then step
3:02one, delivery. Did the ads earn spend?
3:05If the answer is no, the ads aren't
3:07competitive and the test is basically
3:09dead on arrival and we start over. Step
3:12two, evaluation. What happens to the
3:14campaign if the answer is yes? If the
3:16test earns spend and the profit volume
3:18goes up, the test is winning regardless
3:20of what any individual ad metric says.
3:23And if the test earns spend and the
3:25profit volume drops, it's a loser and we
3:27start over. And then step three, scale.
3:31If profit volume went up, we have more
3:33money to invest, so we increase the
3:35budget. And you continue to scale until
3:37profit volume stops improving. And then
3:40you're back to testing. It doesn't
3:42matter if any of your new ads are
3:44winners or if your rorowaz goes from 2
3:46to 20. If profit didn't go up, it's a
3:49loss. Only the business outcome matters
3:53because it's not about any individual
3:55ad. It's about the whole team. And if we
3:57can't scale the budget, we need to make
3:59the team better. Now, a little earlier,
4:02I promised I would show you exactly how
4:04to identify the worst ad. And it's also
4:07stupid simple. When I evaluate ads, I
4:09don't start with rorowaz. ROAZ is a
4:11ratio. It can look amazing while you
4:13make less money. We want to make more
4:16money. So, what we care about is gross
4:18profit per transaction. I call it GPT.
4:20And yes, you can build it as a custom
4:22metric in your dashboard. Here's how to
4:24do it. Here's why this matters. Ad A
4:26could be a 4x rorowaz on a $10 CPA with
4:28a $40 AOV. That's a $30 gross profit. Ad
4:32B could be a 2x rorowaz on a $60 CPA.
4:35With a $120 AOV, that's a $60 gross
4:38profit. half the row as double the
4:41profit. So when we evaluate an ad, the
4:43stupid simple question is, does the ad
4:46have a GPT above campaign average? If
4:49not, that's one we want to try to
4:50replace. This stupid simple structure
4:54gives you something most advertisers
4:56never get.
4:58Certainty. Because we know exactly what
5:01changed and when it changed and what it
5:04did to the bottom line of the business.
5:07We can say things like that new hook
5:09increased profit volume or that angle
5:11brought in a higher AOV customer. No
5:13guessing, no emotional attachment, no
5:15subjective view based on attribution.
5:18And here's the hidden upside, the cross
5:20channel effect.
5:25When you improve the quality of who you
5:27bring in and keep your traffic
5:29consistent, everything downstream gets
5:32better. Let's take a step back to look
5:35at what's really happening. Our creative
5:37testing upgrades the customer quality
5:39coming into the store. And because all
5:42spend runs through one system, Meta's
5:44Andromeda gets better at matching the
5:46right ad to the right buyer. And because
5:49we're not launching hundreds of random
5:51ads, all of our customers begin to look
5:53and feel and behave the same. Then we
5:56scale the budget to buy more of those
5:58higher quality customers. And all of
6:00that means that our site converts
6:01better. search and email pick up more
6:04revenue and even returning customer
6:06revenue also climbs. And that's the
6:08point. We're making the business better
6:11by buying better customers. That's why
6:14we judge success by the bank account,
6:17not just by platform rorowaz. I don't
6:19really give a [ __ ] about rorowaz. And
6:21this is the difference between
6:23optimizing ads like a performance
6:24marketer and being an operator scaling a
6:28business. I think about it like this.
6:30Hunters versus farmers. Hunters lived in
6:33caves for a 100,000 years, living and
6:36dying by their ability to feed
6:38themselves today. Hunters measure
6:40success by rorowaz and daily CPA. They
6:42ask, "Did this add work today? What's my
6:45cost right now? Did it hit immediately?
6:47They live and die by short-term swings."
6:50Once humanity learned how to farm, we
6:52went from living in caves to building
6:54empires and putting a man on the moon.
6:56We are in the empire building business.
6:59If you want to scale to the moon, you
7:01have to start thinking like a farmer.
7:03Farmers focus on feeding as many people
7:06as possible. One corn plant can produce
7:08three or four pounds of food and you can
7:10plant 10 other corn plants from what you
7:12don't eat. So for the same amount of
7:14work over time, you go from four pounds
7:16of food to 400 because you're focusing
7:18on building a better system. And that's
7:20what farming is. Farmers focus on total
7:23profit and growth. Their job isn't to
7:25chase win after win. is to build a
7:27system that works and to protect it and
7:29to scale it with the resources that
7:31system builds for itself. And that's
7:33what Andromeda 1 is actually optimizing
7:35for similarity to produce scale. We're
7:38increasing the volume of high quality
7:41data flowing into the platform so it
7:44gets better at matching the right
7:45message to the right buyer and finding
7:47more of your best customers. Now, before
7:51you copy this structure, we have to talk
7:53about budget because a lower budget
7:56changes the rules just a little bit.
8:04If your budget isn't high enough to
8:06reliably get more than one adset out of
8:08the learning phase, then everything
8:10lives in one ad set. Your control and
8:13your tests are just ads inside of the
8:15same adset. Then you follow a simple
8:18ladder. one adset until you can afford
8:20to get one testing and one control adset
8:22and then you scale until you can afford
8:24two testing adset. And even when I'm
8:26spending a million a month like I showed
8:29you before, I don't have more than two
8:31testing adsets. Once you get there,
8:34you're done. You're ready to scale to
8:36the moon. So now you understand what
8:38Andromeda 1 really is. It's not a tactic
8:41or a hack. And it's definitely not a
8:43structure you copy and paste from a
8:45screenshot. It's a decision to stop
8:48being a hunter and to start being a
8:51farmer. It's choosing a system you can
8:54run every week, not a lucky ad you hope
8:57doesn't die. Now that you understand
8:59that, next we need to talk about the ad
9:02itself, I got to show you the exact ad
9:05format I use, how to build it, and how
9:07to feed the algorithm the kind of data
9:09it needs to actually scale. Because the
9:11best structure in the world won't scale
9:13if your ads aren't teaching meta to win.
9:18These three ads all spent over a million
9:20dollars and they're the exact same
9:23format. After Andromeda, it's the best
9:25kind of ad you can run on Facebook and
9:28almost no one is using it properly. I'm
9:30going to show you exactly what it is,
9:31how to build one from scratch, and how
9:33to scale your campaigns with it.
9:39Let's actually build one of these
9:40together. This is a 322 ad. Three
9:44creatives, two headlines, two primary
9:46text. First, three creatives. Three
9:48videos or three images, not a mix. We
9:51want meta comparing like for like. If
9:53you want to test images versus videos,
9:55run those as two separate ads against
9:58each other. Next, two primary texts.
10:00These should be genuinely different
10:02angles, different hooks, different
10:04reasons to care. I usually like to test
10:07shorter copy, 180 characters or so or
10:10less and then also much longer copy that
10:13has the see more function. And if you do
10:16use the see more, remember to treat it
10:19like above and below the fold. You tease
10:21up top, then continue the conversation
10:23after the click. It is two parts of the
10:27message. And then you have the two
10:29headlines, two distinct frames of value
10:31like buy now and save or it's the last
10:34you'll ever need. Three creatives, two
10:36headlines, two primary text, the 322.
10:40That's 12 possible combination. We're
10:42trying to let Andromeda identify a
10:44winner fast so it can scale what's
10:47working. Now, if your ads manager
10:48happens to look a little bit different
10:49than mine, there is an update going on
10:51right now. There's a link in the
10:53description down below where I walk
10:54through the updated interface.
10:56Everything is the same. It's just some
10:58of the buttons are in different places.
11:03Ad spend is earned. Meta is a
11:06meritocracy. It pushes budget towards
11:08the best performing combination based on
11:10the highest value signals. What has the
11:13best estimated action rate, clicks,
11:15engagements, watch time, all of that.
11:16That's exactly what we want. So, why
11:19does this work so much better than
11:21launching separate post IDs? It all
11:24comes down to separate buckets versus
11:26shared learning. When you run ads one by
11:28one, side by side, each post ID learns
11:32in isolation. Each ad has its own data,
11:34its own history, and none of it carries
11:36over. So if you test a brand new
11:39creative pack, say a dozen new ads, they
11:42don't learn together. And that testing
11:44is funded by pulling spend away from the
11:46ads that were already working. And
11:48during that testing, the new ads that
11:50get delivery first are the ones that
11:52look best closest to the conversion,
11:55which is also why all of your new ads
11:57look great for a couple of days and then
11:59just tank. And it's why your old winners
12:02suddenly look worse right after you
12:04launch new tests. A 322 ad flips that.
12:07Instead of separate buckets, everything
12:09feeds into one shared pool. Creative
12:12headline and primary text, all of that
12:14performance data compounds together. So
12:17when something starts working, that
12:19signal isn't trapped in a single post
12:21ID. It's shared across all the elements
12:24of the ad, which means Meta can build on
12:26what it learned. And that's the real
12:28win. More learning continuity, right?
12:30There's less guessing and more stable
12:32performance. It's one ad learning from
12:34every win and every mistake to get
12:36better and better and better. It's a
12:37self optimizing creative testing device.
12:40Next, I'm going to show you why this
12:42stupid simple structure is what makes
12:45performance more stable. Your ads don't
12:47fatigue, they improve.
12:53Creative moves the needle more than
12:55anything else on Meta. So, our job isn't
12:57to overwhelm the machine with options.
12:59With three creatives, Meta can make a
13:01real decision quickly. That spend
13:04consolidates around the best creative
13:06that you've got, which means all of your
13:08spend is making that choice better. Now,
13:10remember that a 322 is already 12 ads.
13:12So if you load five or 10 creatives,
13:14that could potentially be hundreds.
13:16Whatever gets lucky early is going to
13:18dominate everything only because at your
13:21warmest audience, it was the thing
13:23people clicked on first. And this is why
13:25a lot of folks that say flex ads don't
13:26work. Dynamic creative is stupid is
13:28because they overwhelm the machine. So
13:30they're basically working against the
13:33data instead of leveraging it. You want
13:35Andromeda to do all the hard work so
13:36that you don't have to. And this is how
13:38we do it. Three Creatives keeps it
13:40stupid simple. Either you have a winner
13:43or you kill it and you move on. So if
13:46you want to test more than three
13:47creatives, you don't stack them into one
13:50ad. You build multiple 322s. If you have
13:52six creatives, that's just two 322 ads,
13:55which is already 24 ads. And each one of
13:59them gets its own chance to earn spend.
14:02And this is really useful for people
14:04that want to do copy testing, something
14:06I don't see anybody else talking about.
14:08If both of those 322 ads have the same
14:12headline and primary text, you can watch
14:14what happens to the spend across both.
14:16If the same copy dominates in both ads,
14:19that's a real signal the copy is doing
14:21the work. But if the performance of that
14:24copy is different amongst those two ads,
14:26you know the copy isn't the reason the
14:28ad is winning or losing. Real
14:30multivariant AB testing and it's stupid
14:33simple for you to read the results and
14:36move on. And this is the difference
14:38between chasing performance, launching
14:40ad after ad after ad, versus building
14:43something you can actually rely on, a
14:45foundation that will let you scale. I
14:47hope you're enjoying watching this video
14:49as much as I enjoyed making it. And I
14:51want to take a moment and just say thank
14:52you for watching it. Now, go ahead and
14:54smash that like button, subscribe to the
14:56channel if you haven't already,
14:57[clears throat] and ring the bell so you
14:58don't miss any future videos. And take a
15:00moment to join Disruptor Academy where
15:03you can get for free all the assistance
15:05you need. all of the things that we talk
15:07about in this video. There's a link
15:09right here. All right, with that being
15:11said, let's get back to it.
15:16Once the 322 ad has been running for a
15:18while, the most important KPI for you to
15:21watch is still spend. Spend tells you
15:24what the machine believes. And when a
15:26test earns spend and the overall
15:28business gets better, that's a real
15:30winner. Most people have been trained to
15:32think the goal is winner after winner
15:34after winner. They're focusing on like
15:36creative hit rate and how many winning
15:37ads can I get, but the worst thing you
15:39can do is interrupt something that's
15:41already winning. Your 322 ad is a
15:43self-contained creative test. And when
15:46one version starts to dominate, that is
15:48your winner. That one single post ID is
15:51probably getting most of the spend. It's
15:53the biggest contributor to the outcome
15:55that is good for you. So, you pull it
15:57not by shutting the ad down, but by
15:59taking the winning post ID and
16:01introducing it into your control ad set.
16:04the ad set where all of your other
16:06proven winners live. And you're not
16:08immediately replacing anything. You're
16:10just testing the ad in the control and
16:13then watching what happens. And that
16:15makes creative testing stupid simple.
16:17Does the new ad in the control ad set
16:20make the campaign better? If yes, keep
16:23it. If no, turn it off. Remember that
16:25the purpose of creative testing is not
16:27to find a winner. It's to build the best
16:29team. And this is how you do it. There's
16:32no gray area. There's no subjectivity.
16:35There's no uncertainty. You can be fully
16:37confident in every single choice.
16:43Before we move on, let me zoom out and
16:45tie this all together because once you
16:46see it as a system, everything becomes
16:49much clearer. Okay, so step one is
16:51testing. The 322 ad is your simple self
16:54optimizing creative testing device. Step
16:57two is harvesting. You find a winner,
16:59you grab that post ID and you bring it
17:01into your control ad set alongside your
17:03other proven ads. And step three is
17:06scale. If this makes the whole campaign
17:08better, we can increase the budget. So
17:09that's how you build an ad that can
17:11spend a million dollars. But knowing how
17:13to build one is only half of it.
17:16Meta's algorithm isn't actually a black
17:19box trying to take your money. And once
17:21you know what to look for, everything
17:23changes.
17:27Spend isn't just a budget setting. It's
17:30the clearest way to see what decisions
17:32Meta is making about your ad. Meta sells
17:36attention, and attention is precious and
17:38fragile. Before your ad earns real
17:40spend, it has to pass through two gates.
17:42Lattice balances the user experience
17:44with advertiser objectives. Basically,
17:46should this person see an ad at all? and
17:48Andromeda is the filter choosing which
17:51of the millions of ads competing for
17:53that impression deserve a real chance to
17:55earn distribution. So when your ad
17:57starts earning spend, it's usually
17:59because those systems are seeing signals
18:02like people stopping their scroll, they
18:04watch, they click, they convert. Meta is
18:07saying, "Users like this experience.
18:10We're going to show it to more people."
18:12And that is why ads getting the most
18:14spent can sometimes have a slightly
18:16worse CPA. Because when meta is spending
18:18higher, it's pushing your ad past the
18:20easy conversions and into new colder
18:22people that fill the funnel. Spend is a
18:24measure of preference. Ranking the ads
18:26by how well they meet both your business
18:28objectives and metas. Which brings us to
18:31the most misunderstood metric in meta
18:33ads.
18:38It's easy to see frequency rising and
18:40panic and call it fatigue, but frequency
18:42doesn't really show fatigue. Frequency
18:44is telling you how Meta is using an ad
18:47in the funnel. Here's a simple way I
18:49want you to think about it. When you're
18:50looking at daily frequency, when it's at
18:52a 1.05, that means 5% of people saw that
18:56ad twice. When your daily frequency is a
18:591.95,
19:00that means about 95% of people have seen
19:03that ad twice. Lower frequency usually
19:06means the ad is being used to find new
19:09people. Higher frequency usually means
19:12Meta is using that to retarget
19:13individuals. There are three common
19:16reasons your frequency will climb.
19:18Number one is that cold audiences don't
19:20like the ad. The engagement and
19:22conversion signals aren't strong enough,
19:24so Facebook just stops expanding to new
19:27people. Number two, you keep resetting
19:29the machine if you're launching too many
19:32ads, changing budgets, bids, or
19:34audiences. So nothing gets enough stable
19:36data. Meta can't confidently find new
19:38people. So it leans on the people it
19:40already understands. Those two usually
19:43show up as high frequency and low spend
19:45because the system is protecting the
19:46user experience. Our third option is an
19:49ad that converts really well on warm
19:51audiences. In that case, it can still
19:53earn spend even with the high frequency
19:55because it's doing a great job later in
19:58the journey. Remember that creative does
20:00the targeting. Frequency is how you can
20:03see who's being targeted. So, now that
20:06we know what the algorithm wants to do
20:07with every impression we buy, we have to
20:09know how much those impressions cost and
20:12how good they are. And that brings us to
20:15CPM.
20:19Your ads aren't only competing against
20:21other ads, they're competing against
20:23organic content, too. Your friends,
20:25creators, memes, whatever. In Meta's
20:27business model, where attention is the
20:29product, it's in their best interest to
20:31show people content they'll engage with
20:34and ads for products they want from
20:35brands they can actually trust. There's
20:38a second factor, too. Supply and demand.
20:40The audiences closer to buying are often
20:42the same audiences your competitors
20:44want. So, there's less supply and more
20:46demand. That equals higher CPM. But Meta
20:49doesn't just sell impressions to the
20:50highest bidder. It's optimizing for the
20:52outcome and the experience. Ads with
20:56higher quality experiences can earn
20:58lower CPMs. And as users get closer to
21:02buying, Meta gets more selective about
21:04what ads it will show them. It cares
21:07about what it can predict will happen
21:09after the impression. If your ads create
21:12a great user experience, you can often
21:15be charged less to reach people more
21:17likely to convert, even in competitive
21:20audiences. And if your ads create a bad
21:22experience, you often have to pay more
21:24just to get delivery. Yes, CPM is the
21:28cost of attention, but more importantly,
21:30it's a read on the quality and
21:32competitiveness of that attention.
21:38And that brings us to the other really
21:40important cost metric, cost per result.
21:42A lot of people use it to decide whether
21:44an ad is good or bad, and honestly,
21:46whether they're actually good or bad at
21:48the job of running ads. Lowerfunnel ads
21:51can look incredible. Meanwhile,
21:52upperfunnel ads almost always look worse
21:55on a CPA basis because they're doing
21:57heavier work earlier. They grab
21:59attention. They build intent and earn
22:01trust and prove authority. A lot of
22:03teams optimize by chasing a better CPA
22:05and a higher rorowaz. What they're
22:07really doing is trying to spend as much
22:08of their budget as possible at the
22:10bottom of the funnel, not by audience,
22:12but by how Meta is using their ads,
22:16which is a very important distinction.
22:18They effectively cut prospecting because
22:20it looks expensive. They keep only the
22:22ads that close. And so your CPA drops
22:25and your rorowaz goes up and everybody
22:27celebrates. But then the funnel starts
22:29shrinking. Your frequency climbs, your
22:31CPMs climb, your CAC climbs, and
22:34eventually revenue stalls because you've
22:36optimized for efficiency instead of
22:38growth. Cost per result is not a measure
22:40of how good an ad is. It's a measure of
22:42how much that ad contributed to the
22:44conversion actually happening. The lower
22:46the cost, the less work it probably did
22:48because it showed up at the end. And the
22:51higher the cost, the more work it might
22:53be doing because it's creating demand
22:56earlier in the journey. So now that
22:57we've covered all the metrics, let's
23:00talk about how to put them to work.
23:06When you use these four metric, spend,
23:07frequency, CPM, and cost per result to
23:09understand how the machine is using your
23:11ads, you are conducting a 4PI analysis.
23:15Let me give you three of the most common
23:17patterns that I look for. Two examples
23:19of an ad that is doing its job well and
23:22one huge red flag that tells you what to
23:24pause or replace with creative testing.
23:27Example number one, a great upperfunnel
23:30ad prospecting. The spend is above
23:33average. It's spending more than the
23:34other ads. The frequency is below
23:36average. Your CPM is below average
23:38compared to your other ads. It's a good
23:40attention experience. And the cost per
23:42result is probably above average. and
23:44that's normal. Cost per result can be
23:46higher as long as it's not so high that
23:48the ad becomes a liability relative to
23:50the spend that it's taking. This is the
23:52ad that creates a great first touch, and
23:54it should set up the ads that come later
23:57in the funnel. Example number two is a
23:59great lowerfunnel ad, retargeting and
24:01closing, a frequency that's higher than
24:03average, a CPM that's higher than
24:05average, and a cost per result that's
24:07better than average. Spend should be
24:10meaningful enough that it actually
24:11contributes. We don't want it to spend
24:13just enough to get one or two sales a
24:15week. It should be the highest spending
24:17thing that has a greater than average
24:19CPM and frequency. When these ads are a
24:21logical second touch to best prospecting
24:23ad, you'll often see them get even more
24:25efficient as the funnel fills. Let's
24:27talk about the red flag expensive
24:30results and the machine still loves it.
24:32If an ad has a bad cost per result, but
24:34it's still earning a ton of spend, it
24:36might be amazing at attracting the wrong
24:39kind of attention. It's getting
24:41distribution, but it's pulling in people
24:42with no intent to buy. So, it becomes a
24:45liability to the bottom line. And
24:48another version of this same red flag is
24:50frequency above average, CPM above
24:52average, but the cost per result is no
24:54better than the prospecting ads. That
24:56usually means the ad appeals to a warm
24:58audience, but doesn't convert them. So,
25:01it soaks up budget that could be going
25:03to an ad that actually improves
25:05efficiency and drives sales. Now that
25:08you can spot these patterns, the next
25:09step is simple. What do you turn off?
25:11What do you keep? And what do you test
25:14next based on your 4PI and that is what
25:17we're doing in the action item. Now, I
25:19hope you're enjoying watching this video
25:20as much as I enjoyed making it. And
25:21please go ahead and like and subscribe
25:23so you don't miss anything else. And if
25:25you want to learn more about the 4PI
25:26analysis, if you want to dive all the
25:28way into the deep end, we just dropped a
25:30brand new course in Disruptor Academy.
25:32There's a link right there so you can
25:33join. All right, with that being said,
25:35let's get back to it.
25:40Data is only ever as valuable as it is
25:43actionable. So what actions does a 4PI
25:45analysis tell us to do? There are really
25:48only two logical outcomes if you want to
25:50make the account better. Outcome number
25:51one is optimization by subtraction. This
25:54is the simplest way and honestly it's
25:56often the fastest way to scale results.
25:58Your goal is to remove an ad that's
26:00hurting the system so the budget
26:01naturally flows to the ads that are
26:03actually doing the job. You're scaling
26:05the winners without touching the budget.
26:06Because remember, we're not trying to
26:08find one winning ad to isolate, scale,
26:11and abuse until it breaks. We're trying
26:13to build the best team of ads that
26:15produces a repeatable, stable, and
26:18scalable outcome. So, if you're running
26:19a lot of ads, maybe more than eight.
26:22Start here. Find the ads that are taking
26:25meaningful spend and have clear red
26:27flags, and start turning those off.
26:29Eliminating the liability is the easiest
26:32way to get better. And outcome number
26:35two, improve the team by filling the
26:37gaps. If you don't have a clear red
26:39flag, you don't have a pause stuff
26:42problem. You have a coverage problem in
26:44the funnel. Here are the most common
26:46ways it shows up. First, everything
26:49looks fine until you raise budget. And
26:51when you scale, CPM, frequency, and cost
26:54per results all get worse across the
26:56board. That usually means you're not
26:58filling the funnel effectively enough.
27:00So, your creative testing needs to focus
27:02on stronger upperfunnel ads. Number two,
27:06when you scale, one prospecting ad soaks
27:08up all the spend, and CPR gets worse and
27:10worse. And that's usually an ad people
27:12love to watch, but it's attracting the
27:14wrong kind of attention at scale. So,
27:16the fixed isn't more budget tweaks and
27:18bidding modifications. The fix is to
27:21replace or iterate on that prospecting
27:23ad to give the machine better choices.
27:25And number three, nothing behaves like a
27:27lowerfunnel ad that can earn spend. So
27:29you don't have a closer. So you need to
27:30test for an ad with a higher frequency,
27:32higher CPM, and a better CPR, something
27:34that can convert the warm traffic that
27:36you're buying more efficiently. The
27:38easiest way to do that is to take a
27:40strong prospecting idea, and make a
27:42second touch version, same core message,
27:44but built for someone who already saw
27:47the first ad.
27:52Your job isn't to find one winning ad
27:55and abuse it until it breaks. Ad fatigue
27:59is almost always operator error. Your
28:02job is to get ads that are working
28:04together so the system stays balanced,
28:06profitable, and scalable because that
28:08balance is what lets you answer the only
28:10question that matters. Can I spend more
28:12money? With a yes. A 4PI analysis turns
28:16your ads manager into a decision engine.
28:19So now you know how to look at all the
28:20ads. I got to show you how to test more
28:22effectively so you can cover those gaps
28:24and scale your account.
28:26Every expert says to test more and more,
28:28but that only wastes more money, creates
28:30more ad fatigue, and makes scaling
28:33almost impossible. I'll show you the
28:35Olympic rings method, my five concept
28:37system built around one simple question,
28:40and how it's taken dozens of accounts
28:42from hundreds a day to thousands. Before
28:45I walk you through how all of this
28:46works, I need to show you what an ad
28:48really is. Because it's not what you
28:50think. On Instagram and Facebook, every
28:53post, paid or organic, is a web page on
28:57their site that competes for attention
29:00in the feed. Every social media feed is
29:04basically just a Google search results
29:07page where Andromeda is showing you
29:09content it thinks you'll engage with.
29:12Every single piece of content has a
29:14score called the estimated action rate
29:17based on how likely people are to do
29:20things like engage and click. And every
29:23new ad starts with a score of zero.
29:30This is where testing tons of ads begins
29:32to make everything worse. For the first
29:35few days of any new ad, your ads are
29:38always shown first to the warmest
29:40audience because those are the people
29:42most likely to engage and click and take
29:45actions on your site that you want them
29:46to. Let's take a step back real quick.
29:49The more new ads you launch, the more
29:51you invest in competition against your
29:54warmest audience. The new ads take spend
29:58and credit from your existing ones. So,
30:00the older ads look worse, and you're not
30:03filling the funnel. This means it gets
30:05more and more difficult to reach new
30:07customers, and all of your costs go up,
30:10and then you get trapped in the cycle.
30:12Results look worse, so you launch more
30:15ads, performance gets less stable, and
30:17you become more dependent on whatever
30:19new ad is getting the warm conversions
30:22today. You end up day trading the ad
30:24account. just fighting to hold the line
30:26and that's not scaling the business. If
30:28you want to make a lot more money, that
30:31doesn't make a lot of sense.
30:36Now, I know what you're thinking,
30:39Charlie.
30:40>> What about creative diversity?
30:42>> Let's clear it up because if you don't
30:45understand this, nothing else is going
30:47to fix your problem. Creative diversity
30:50is not about the volume of ads you make.
30:53Meta specifically states that Andromeda
30:56was built because so many people were
30:58launching so many ads and now it
31:00penalizes that. What Meta wants is
31:03simpler, a few ads that do different
31:05jobs.
31:10Which brings us to the method of the
31:12Olympic rings. A framework to analyze,
31:15test, and scale without undermining the
31:17ads that are already working. The
31:19Olympic rings is a fiveconcept
31:23process. Each ring is a concept. Each
31:25ring has different ads and all rings are
31:28purposefully connected. The three rings
31:32on top are your prospecting ads. The two
31:34rings on the bottom are your retargeting
31:36ads. Ring number four should be designed
31:38as the second touch for the ads in ring
31:40number one and number two. Ring number
31:43five should be designed as the second
31:44touch for ads in ring number two and
31:46number three. This structure creates a
31:49well organized framework of prospecting
31:51ads and retargeting ads. To get started,
31:53you just need five core ads, not
31:56hundreds, not spammy chaos. A small team
31:58or every ad has a job. And none of these
32:01ads have to be perfect. They just have
32:03to be good enough at their role in the
32:04system. Because when they are, Andromeda
32:07can match the right message to the right
32:09person at the right time. And that's
32:10what this is all about. Now, I hope
32:12you're enjoying watching this video as
32:14much as I enjoyed making it. And I want
32:16to take a moment and just say thank you
32:17for watching it. Go ahead and smash the
32:19like button. Subscribe to the channel if
32:20you haven't already. Ring the bell so
32:22you don't miss any future videos. And
32:23take a moment to join the Disruptor
32:25Academy where you can get for free all
32:27the assistance that you need. All the
32:29things that we talk about in this video.
32:30There's a link right there. All right,
32:33with that being said, let's get back to
32:34it.
32:39When you build your account like this,
32:41everything changes. These ads don't
32:43magically become profitable forever, but
32:45they stay stable for much longer and
32:48actually improve over time because they
32:50aren't competing for the same
32:51conversions. Every impression teaches
32:54Andromeda something useful. Which
32:56sequence of ads works on which person,
33:00when to show the problem ad versus the
33:01one with proof, and who just needs a
33:03reminder, and in what order to tell the
33:05story to get the best customer for the
33:07cheapest price. Most accounts run dozens
33:10of new ads every week, competing to be
33:12the last touch on any sale that happens.
33:14The Olympic rings is a five concept
33:16system of ads that work together to make
33:19more sales happen. That's why results
33:21get more stable. Scaling stops feeling
33:23like gambling, and you stop needing to
33:26save the account every week with a new
33:28win from a fresh creative pack. And when
33:31your traffic is more consistent in
33:33intent and quality, everything
33:35downstream gets easier. landing pages,
33:38email, even search because you can
33:40finally focus on improving the business,
33:43not day trading the ad account. Less
33:46wasted spend, higher quality customers,
33:48and more scale that finally feels
33:50inevitable instead of impossible. Alex
33:53Schultz, met a CMO in his last book,
33:55explains that when Zuckerberg talks
33:57about personalization, he's not talking
33:59about making a different ad for every
34:01single person. He's talking about the
34:03overall experience somebody gets across
34:06the feed, across all of the touch
34:08points, paid and organic. And that is
34:11the whole point of the Olympic rings.
34:13You're not trying to spam the system to
34:14hack your way into the highest rorowaz.
34:17You're giving it a small set of distinct
34:19and proven messages so it can
34:21personalize who sees what and when.
34:24Personalization is a delivery sequence
34:27problem, not a creative volume problem.
34:30So, all of this forces one simple
34:33question. When should you test new ads
34:36and how should you do it? Creative
34:39testing isn't about the next win or the
34:41cheapest sale. It's about being able to
34:44get more sales. If the balance between
34:47the rings is stable, and every time you
34:49increase the budget, the performance is
34:51consistent. You already solve the
34:53hardest problem in any ad account. The
34:56real question is, can I spend more
34:59money? If the answer is yes, I can,
35:01don't try to fix what ain't broken.
35:03You've already won. The single worst
35:05thing you could do is launch new ads.
35:08Just keep scaling. But if the answer is
35:11no, that is when creative testing
35:13matters.
35:18That is when you use a simple version of
35:20the scientific method to solve the
35:23problem. You start with a control, the
35:25system you already know works, even if
35:28it's not working well enough. Then you
35:30test one variable at a time until it is.
35:33One variable could be a new hook or a
35:35new testimonial or handling a different
35:37objection, but don't change three things
35:40at once because you'll never know what
35:42actually fixed the problem. That is what
35:44creative testing is for. Not finding the
35:47perfect ad. It's finding one small
35:49improvement that gets you back to a yes,
35:51I can spend more money without
35:53performance breaking. So instead of
35:55launching random new ideas, you look at
35:58your rings and ask one simple question.
36:01Which one of these isn't doing its job?
36:03Remember when I said creative testing
36:05isn't about finding more winners? This
36:07is why. Any winning ad can look
36:09unstoppable, but it can only carry so
36:11much weight. And if your winners don't
36:14work together, you don't have a system.
36:16You just have a bunch of isolated
36:18journeys and you've effectively built a
36:20more complicated house of cards and
36:22you've amplified the problem instead of
36:24solving it. And I've made that mistake.
36:26And this is why so many accounts
36:29struggle to scale. Testing as much as
36:31possible to find the next winning ad.
36:34But Facebook ads is a team sport. Messi
36:37is incredible, but he's not winning
36:39matches if he's on the pitch by himself.
36:41The way you win is by building a team of
36:44complimentary players, each designed for
36:47a specific job, all pushing towards the
36:50same goal, consistent, profitable
36:53growth.
36:58That's what meta rewards, that's what
37:00real creative diversity actually looks
37:03like in practice. And that's what
37:04Andromeda was built for, a system that
37:07can win again and again. And that makes
37:10your job incredibly easy. If you're not
37:13winning, improve the worst player on the
37:15team. Meta doesn't care how clever your
37:17ads are. It cares whether your system
37:20gives people a good enough experience.
37:22And if it does, your performance will be
37:24good enough to increase the budget. That
37:26is how you scale an account. The real
37:28definition of success isn't vanity
37:30metrics like rorowaz or CTR or hook
37:33rate. It's being able to say yes to the
37:35one simple question of, "Can I spend
37:38more money tomorrow?" If the answer is
37:40yes, you're winning. And if the answer
37:42is no, you do what we talked about here
37:44in this video until you get back to a
37:46yes. You don't try to hack your way to
37:49the next win. You make Andromeda your
37:51sales team. You make Meta work for you,
37:53not the other way around.
37:58How do they scale under outfit from 50K
38:00a month to a million a week in a year
38:02and a half? and how do they take 310
38:04nutrition from 15 million a year to 98
38:07million a year in 20 months? I'm going
38:08to show you the stupid simple
38:10three-stage system I use to scale any
38:13business. This is the new best way to
38:16scale Facebook ads. So, I'm going to
38:17share my screen here real quick and
38:19we're going to look at this ad account.
38:20We can see that we're profitable on GPT
38:23and when we look at total profit,
38:25including subscriptions and purchases,
38:27we're in a great place. We've been
38:29pushing the budget up more and more and
38:31it looks solid. So, let's just go here
38:33into the campaign level, hit edit, go to
38:37the budget, and let's triple it right
38:39there. All right, that looks good. Let's
38:41hit publish. Actually, that's a terrible
38:44idea. I'm not going to do that. If it
38:47was that easy, you'd be a millionaire
38:49right now, and you wouldn't be watching
38:50this video. But why is it that every
38:53time you increase the budget, your
38:54performance tanks?
38:59It's because scaling isn't one move.
39:01It's a system and there's a specific
39:04order it has to happen in. Think of it
39:06like building a house. You don't put the
39:08roof on before the walls are up. But
39:09that's exactly what most advertisers do.
39:12They go straight to the budget before
39:13the foundation is even there. There are
39:16three stages and they have to happen in
39:18order. Most people skip straight to
39:21stage three, budget. And that's where
39:23everything falls apart. So, we're not
39:25starting there. We're starting at the
39:27foundation. And that is stage one.
39:34Most people think growth comes from
39:36spending more money. Let's say you're
39:38getting good results, so you increase
39:40the budget and revenue goes up, but
39:42profit barely moves. Added spend without
39:45added profit is just a waste of time and
39:48money. You're not creating leverage.
39:52Scaling efficiency is about answering
39:54one stupid simple question. How do we
39:57make every dollar produce more? Because
39:59when you improve output per dollar, you
40:01make more money without spending more
40:02money. That's leverage. So, here's the
40:04three-step process to actually do it.
40:07And most people get this completely
40:10wrong. If you want to make more money,
40:12you have to track profit. Most people
40:15obsess over platform metrics or they
40:17cling to rorowaz. And that is where
40:19things break because you can have a
40:20strong rorowaz and be way less
40:22profitable. And I'll prove that in just
40:25a minute. But first, you only need four
40:28numbers. What you spent, what it cost to
40:31get a sale, what that sale was worth,
40:33and how much profit did you make. That's
40:35it. And I call that last one GPT, gross
40:38profit per transaction. For every ad you
40:40look at, stop asking what's the rorowaz,
40:43and start asking how much profit did
40:45this make. To make this stupid simple,
40:49just add GPT to your dashboard. Here's
40:52how to set it up, which brings us to
40:54step two. Look at the ads spending the
40:56most money and identify the ones making
40:59you the least profit per sale. That's
41:01where the immediate leverage is. And
41:03that brings us to step three. The
41:04easiest way to scale profits today
41:06without doing any extra work is simply
41:09to pause the worst performers first.
41:11Turning off ads is the same thing as
41:13changing the budget. So never turn off
41:15more than 20% of the spend on any day.
41:18Now your money is working harder every
41:20day so that you don't have to.
41:26So before we move on, I promised that we
41:28would bust the rorowaz math. Here are
41:31two ads selling the same product. Ad A
41:33has a CPA of 60 and an AOV of 110, so
41:35the GPT is 50. Ad B has a CPA of 40 and
41:39an AOV of 80, so the GPT is 40. One ad
41:43looks worse in rorowaz, but it's 25%
41:45more profitable on every single sale.
41:47The rorowaz doesn't matter. The business
41:50outcome is all that matters and that's
41:52the real job of scaling efficiency. I
41:54hope you're enjoying watching this video
41:55as much as I enjoyed making it. Please
41:57go ahead and hit the like button if you
41:59like the video and subscribe to the
42:00channel if you haven't already. Also,
42:01check out Disruptor Academy. There's a
42:03link right there. We just launched a
42:05brand new course on infinite scaling
42:07every single step of the way of
42:09everything that you need to know so that
42:11you can see the success that you
42:12deserve. All right, let's get back to
42:15it.
42:19Most people increase spend because they
42:20think things are working, but feeling
42:22profitable and being ready to scale are
42:24two very different things. Budget
42:26amplifies what exists. So, you need to
42:28know exactly what you're amplifying
42:29before you touch it. First, is your CPA
42:32and GPT relatively stable. If the answer
42:34is yes, we move on. Second, are there
42:38more bad ads you could get rid of? If
42:40the answer is yes, keep optimizing so
42:43that you can create as much leverage as
42:45possible. We don't want to scale bad
42:47ads. And if the answer is no, then we
42:50move on one more time. You ask yourself
42:53one stupid simple question. If my CPA
42:56increases by 10 or 20% tomorrow, am I
42:59still profitable? If the answer is no,
43:01then we have some more work to do. But
43:03if the answer is yes, we can begin to
43:05scale the budget with confidence.
43:11Now that you're ready to scale, we have
43:12to choose the right strategy. There are
43:14three ways to scale your budget and if
43:16you choose the wrong one, things could
43:18get ugly quick. So to choose the right
43:20method for your account, we have to ask
43:22three simple questions. How stable is
43:24your system? How confident are you in
43:25the performance? And how aggressive do
43:27you want to be? If the answer is
43:29performance is stable, I'm confident in
43:31the outcome and I just want to grow
43:32without breaking anything. The right
43:34method is linear scaling. And that's
43:36where we're going to start. Linear
43:37scaling is the simplest method. You add
43:40a fixed amount to your budget on a
43:42regular schedule. I love using automated
43:44rules for this. Here's how to do it. So,
43:46instead of doubling your spend
43:47overnight, you might add $10 to $50 a
43:50day. The advantage is stability, and
43:52over time, each increase becomes a
43:54smaller and smaller percentage of your
43:56overall budget. So, the system gets more
43:58and more stable and easier to manage.
44:00Your entire job becomes as simple as
44:02making sure that $10 more tomorrow is
44:04totally fine. The downside is that it's
44:08slow by design. Linear scaling is right
44:10when your system is stable and
44:12protecting that stability matters more
44:14than growing fast. And while it might
44:16not sound like a lot, adding just $10 a
44:19day means that one year from now you'll
44:22be easily spending over $3,000 more
44:25every single day. Now, fractional
44:28scaling. If you're willing to be more
44:30aggressive, fractional scaling is the
44:32faster method. Instead of adding a fixed
44:34amount, you increase your budget by a
44:36percentage on a regular schedule. Again,
44:38here's how to do it. So, instead of
44:40adding $10 every day, you might increase
44:42by 2% daily. The advantage is speed.
44:46Because you're using percentages, every
44:48increase is bigger than the last one.
44:50And the bigger your budget, the faster
44:52it compounds. The downside is
44:54volatility. Fractional scaling moves
44:56much faster than linear, which means the
44:58system has way less time to absorb each
45:00increase. The faster you grow, the less
45:03stable things get. Fractional scaling is
45:05right when you are dramatically beating
45:07your goal. And while 2% doesn't sound
45:09like a lot, you could 6x your daily
45:11budget in 90 days while protecting
45:13profitability, and that is incredible.
45:15Linear scaling and fractional scaling
45:18are excellent to get you to that next
45:20level, but eventually results plateau,
45:22and you have to turn those rules off.
45:24And that brings us to my favorite
45:26method, marginal scaling. Instead of
45:28adding a fixed amount or fixed
45:30percentage daily, you scale based on
45:32whether your current performance has
45:34earned it. So instead of increasing on a
45:36schedule, you set a rule like if my CPA
45:38over the last seven days is below my
45:40target, increase the budget by a small
45:42percentage, but never by more than the
45:44profit margin allows. So for instance,
45:47if my target CPA is 50 bucks, I'm
45:49currently coming in at a 45, I can scale
45:51by 5%. And if I get no additional sales,
45:54I'm still profitable. Marginal scaling
45:56is right. If your system is mature and
45:59your target CPA is clearly defined and
46:01you want the account to scale itself
46:03when performance is good enough to
46:05support it.
46:11Scaling up is a lot of fun, but what do
46:13you do when your CPA jumps 30% overnight
46:15and your every instinct is to just start
46:18fixing everything? So, now I'm going to
46:20show you a stupid simple three-step way
46:22to scale down without killing your
46:23winners, so you can stabilize before
46:26sprinting to the next plateau. Step one
46:29is always optimization, not budget. Go
46:31back to your ads, find the ones with the
46:33worst GPT, and turn them off. Let the
46:36budget reallocate to your strongest
46:37performers, and give this system a
46:39chance to fix itself before you touch
46:41spend. Nine times out of 10, the problem
46:44isn't the budget, it's how Andromeda
46:46decided to use it. Fix that first. Step
46:49two is evaluation. If you've optimized
46:51and performance is still dropping,
46:52that's when you look at budget, but you
46:54only act when performance is clearly and
46:56consistently outside your acceptable
46:59range. You don't react to a bad day.
47:01Those happen. If every week is GETTING
47:03BETTER, THE BAD DAY IS JUST NOISE. Don't
47:05forget how far you've already come. And
47:07step three is reduction. Scale down at
47:10half the speed you scale up. If you were
47:12increasing by 10%, reduce by five. If
47:14you were adding 50 bucks a day, you pull
47:16back by 25. And then you wait. Don't
47:18touch anything for at least 48 to 72
47:21hours. Let the spend redistribute. Let
47:23the strong ads take over. Let
47:25performance tell you what it needs
47:27before you make another move.
47:33Facebook made a massive algorithm update
47:35in April, and I'll tell you what, it has
47:37been wrecking havoc with business owners
47:39who advertise on Meta, including
47:41Facebook and Instagram. I'm going to
47:43break down exactly what the change was
47:45and how to combat this so that you're
47:47not left behind. All right, so let's go
47:50ahead and get started. I'm going to
47:52share my screen real quick. Okay, so the
47:55update was that Meta killed the flexible
47:58ad. You're already seeing the alert for
48:00this inside of your own dashboard. When
48:02this update was announced, I reached out
48:04to my friends on the Meta Engineering
48:06and Product teams, and this is the email
48:09they sent me. And then they gave me all
48:12of the internal training docs that Metal
48:14Reps will be taught in May. Here they
48:17are. And I got to tell you, it's really
48:19exciting. The first mistake that people
48:21make here, this workflow only appears on
48:24new ads. So, you can't edit old ads. So,
48:27you need to create a new ad to use it.
48:30Now, once you're at the ad level, this
48:33is where things start to look a little
48:35different. Here, you're going to click
48:37set up creative. Then you choose image
48:41or video. Like flex ads, dco, and
48:44dynamic creative beforehand. You can
48:46upload multiple images and videos all
48:50inside of one ad. Everything now just
48:52lives inside of this one system. Now,
48:56here's where most people go wrong. They
48:57upload random creative packs after pack
49:00after pack built around some general
49:03idea. Instead, you want to be
49:05intentional. You'll notice that you can
49:08upload up to 10 creatives here. Don't
49:11upload more than three. You'll see image
49:14tab and video tab. And while you can
49:16toggle between them, these ads should
49:19only be of one format. Each ad is a
49:22test. Every test should answer a
49:25question like which hook is the best,
49:27which image format converts best. If you
49:30mix formats, you're testing to see if
49:32you can get lucky. Luck is not a
49:35business model. Mixed ads will fatigue.
49:37You want results, not excuses. All
49:40right, moving on. You're going to want
49:42to check your crops. It's really easy to
49:45speedrun through this little one step
49:47and waste a ton of money on ads that
49:50will never work. So, to do it, hover
49:53over your media, click crop, and adjust
49:57square, vertical, and horizontal.
50:00Facebook will auto adjust, but you
50:03should always double check. And quick
50:05note, because believe it or not,
50:07Facebook has your back. You can do this
50:10both in the wizard and also within the
50:13ad after you publish. Next, we'll get to
50:17our text and image generation steps. For
50:20the text gen, if you have copy you want
50:23to use, just go ahead and ignore this.
50:25But if you just want to test an idea and
50:28don't really know what your copy should
50:29be, I fully recommend taking the text of
50:33your landing page and using it as a
50:35source and letting the meta AI write you
50:38five new sets of copy to use. I've been
50:41doing this for nearly a year now, and
50:43it's an incredible way to help brands
50:45get their first ads running, or to help
50:48test new products or go fishing for new
50:50angles that you might want to more fully
50:53explore later without dedicating a full
50:55workflow and team and a ton of resources
50:58to the project. And because the machine
51:01knows who likes what, the ability for
51:03Andromeda to match the right copy with
51:06the right person is staggering. This is
51:09also killer for brands with a lot of
51:11offers and SKs like print on demand,
51:14clothing outfits, supplements, general
51:16stores, or even drop shippers and
51:18affiliates just hunting for opportunity.
51:21And now for the image gen. And I'll be
51:24honest, the preview Meta gives don't do
51:27the tool any justice. You can use this
51:30to create a whole suite of options. I've
51:33been testing this with Meta support now
51:35for over a year and I'm very happy to
51:38report that these ads absolutely go
51:40head-to-head with any AI image maker
51:43I've used and a LOT OF DEDICATED SHOPS
51:45TOO. When I have a model in the shop,
51:47it'll even give me dozens more to choose
51:49from. It has contacts to the vertical
51:52and other advertisers. It's a cheat code
51:55and it's free. And like text gen because
51:59the machine knows who responds to what
52:01could be creative. The ability for
52:02Andromeda to match the right image to
52:04the right person again is staggering.
52:07Now, you do have to prune away the
52:09things that don't work. Don't just
52:10blindly trust artificial intelligence.
52:13Okay? And yes, sometimes the text gets
52:15weird, but I've absolutely used this
52:17with brands that only had product
52:19renders and built out complete campaigns
52:21that convert and scale using nothing but
52:23AI. And again, it was free and took 5
52:26minutes to create the text and the
52:27images and to prune the things and
52:30launch it. The last option here is the
52:32enhancements. There is a huge list of
52:36options that you might see depending on
52:38what you're doing and where you are. By
52:40the way, if you want me to do a video
52:41where I break down each one, let me know
52:43in the comments down below. But for now,
52:46here's a simple checklist of what to use
52:48and what to avoid. Yes to visual
52:51touch-ups, to relevant comments, to
52:54brightness and contrast, and yes to
52:56dynamic descriptions. And I highly
52:59highly recommend that you avoid enhanced
53:01CTAs, overlays, anything to do with
53:03image or background or text, music, and
53:06never the site links. Again, in the
53:08comments, if you have a question about a
53:10specific enhancement, I'll be happy to
53:12address it. And in Disruptor Academy,
53:14I'm finishing up a full cheat sheet for
53:16every single enhancement and their
53:18formats, use cases, and risk levels. The
53:21link is in the description, and we will
53:23be raising the price soon. So get in now
53:26if you want it. Now for the first update
53:29that really shivers my timbers. You can
53:32customize every combination. Crop,
53:35placement, copy, URL. Before all the ads
53:39in the Flexer Dynamic Creative shared
53:41the same settings. Now you can customize
53:45each iteration inside the same ad.
53:48They've opened up Pandora's box and you
53:51can control everything and the
53:53opportunities are endless. This is a
53:56massive upgrade. You can now control
53:58where each creative shows up. For
54:00example, you might want one creative in
54:02feeds but not in reels. You might want a
54:04specific variation for Instagram and
54:07another in Facebook. You might want one
54:09set of copy to show up in one place and
54:10another to show up someplace else. You
54:12can turn off placement at any level of
54:14the asset. And with the destination
54:17option, you could support multiple
54:19funnels in one ad. Perhaps you have
54:21three creators and each has their own
54:23landing page and give them all the same
54:24script. You could route each creative
54:27from each creator to their own dedicated
54:29funnel in a way that was never possible
54:30before. And [snorts] for the lead genen
54:32and SAS folks, you might notice that
54:35Instagram mobile traffic behaves very
54:37differently than a Facebook desktop. And
54:39you can control platform and placement
54:41and then route that traffic to a
54:43dedicated user experience.
54:46This gives you full control without
54:48needing separate ads. We have been
54:51begging for this since I don't know at
54:54least 2018. And also also no more of
55:00those annoying carousels.
55:03[laughter]
55:04The second update is a wet dream.
55:06Finally, you'll get reporting for each
55:09element. We haven't seen this since iOS
55:1114, and now it's standard. Now, this is
55:14all super exciting, but with great power
55:16comes great responsibility. Before you
55:19smash publish, you absolutely need to
55:24check your previews. So, use the preview
55:26option, turn on the ad preview, and hit
55:28advanced preview. Look at how each
55:30creative variation appears across all
55:32the placements in the way that you want
55:34it. Because something that looks good in
55:36one place might break in another. The
55:38more things that you've changed and
55:39customized, the more likely it is
55:41something's broken. Now, once everything
55:43looks good, you just hit publish. And
55:46this is where it gets complicated. And
55:48that's what I'm going to show you at the
55:49end. It's going to be super cool. By the
55:52way, if you want to get your hands on
55:53these materials, they're free in my
55:55community on School Disruptor Academy.
55:58Link is down below. But I'm going to
56:00share with you guys now on my screen,
56:03which I'm very excited about the
56:04campaign structure of how we're going to
56:07be testing this. So, we have three
56:11general trends. We have the alpha guru
56:14spammer style made famous by guys like
56:16Jeremy Haynes where you have one [ __ ]
56:19campaign with 50 ads in an ad set and
56:22then you isolate the winners and dupe
56:24the ad set and you launch three new ads
56:26at a time until you end up with hundreds
56:27of ads that all fatigue quickly.
56:29Performance is rarely stable and you're
56:31just hunting for any win hoping it works
56:33long enough for you to get your money
56:34back. Then we have the Sam Hilliero and
56:37Ben Heath style which Sam calls the M3
56:39or M4 method which is basically just my
56:41bucket based CBO method from 2018 and
56:44Tim Bird borrowed it and rebranded it as
56:46cloud CBO a year later. This is where
56:48you have a CBO campaign with new ad sets
56:51launched weekly or so with five to 20
56:53ads in each and spend minimums on
56:55everything where you then copy the
56:57winners to interest groups and
56:58retargeting and retention campaigns and
57:00you scale based on platform rorowaz and
57:02hope for the best while generating a ton
57:03of noise and giving the machine almost
57:05no data to truly stabilize and optimize
57:07and deliver consistent quality or volume
57:10of traffic. And this style absolutely
57:12works if the goal is to make the agency
57:15report look good right up until the
57:17whole funnel is depleted and the whole
57:19business tanks because this style
57:20usually bets on scaling mid and
57:23lowerfunnel ads that fatigue faster and
57:24faster without refilling the bucket. And
57:27lastly, we have the one campaign method
57:29that I invented like 5 years ago and
57:31it's championed by everyone from Neil
57:32Patel to Spencer, the goat of school.
57:35Now, the reason they all love it is
57:36because this style requires the fewest
57:38amount of ads and takes the least amount
57:39of work to manage. Instead of day
57:41trading the ad account, people can work
57:42on their business and enjoy their life.
57:45If you're in an operator of capacity,
57:46this is the easiest way to make a lot
57:48more money. And this is what we're going
57:49to use for these ads because the one
57:51campaign method, unlike the other two,
57:54was built on dynamic creative and flex
57:56ads, specifically 322 ads, which
58:00historically is the most stable and
58:02scalable ad type with the least ad
58:04fatigue. And this new workflow was built
58:06to replace that one best tool. In the
58:09modern version of this that we now call
58:10Andromeda 1, which is a CBO campaign, we
58:13have one ad set with our most consistent
58:15ads, no less than four and no more than
58:18eight at a time. And using a simple 4PI
58:21analysis, we can see which ads are good
58:23at their job and which ones aren't.
58:24Based on things like daily frequency,
58:26spend share, and CPM while optimizing
58:28for gross profit per transaction instead
58:31of rorowaz. Remember, rorowaz isn't
58:32money. You can't pay bills with ratio.
58:35You can't buy whiskey with fractions.
58:37You need currency. In this build, we
58:40have a control ad set of like four to
58:41eight ads and then a test ad set using
58:43multimedia creative workflow ads. And
58:46when we launch new ads in this test ad
58:47set within our CBO, we look to see if
58:49they earn spend and if they do, what ads
58:51from the control ad set get less spent
58:54and did the overall performance across
58:56the entire business get better or worse
58:58because of it. This way, we set a
59:01baseline. We introduce a change and we
59:03measure the impact. It's a simple
59:06scientific method. The only real change
59:09with this new multimedia creative
59:10workflow is that if we wanted to be more
59:12granular, we can. And if we don't have
59:15any idea what will work and we don't
59:17have four to eight good ads yet, we can
59:19use the test and image gen tools or the
59:21heavy customization options to truly
59:23constrain or massively widen the set of
59:26options that we are letting the machine
59:27make for us. If we know what works, take
59:30more control. If we don't, we let the
59:33machine do the work.
59:38Meta's new update just made every AI
59:40tool that you're paying for completely
59:42unnecessary. If you're still using
59:44ChatGpt or Cloud to build your ad
59:46creative, that changes today. I'm going
59:48to show you exactly what Meta just
59:50released. how Meta's new tool already
59:52knows what will convert before you spend
59:54a single dollar and how to use it live
59:57inside an ad account so you can make
59:59better ads faster without any other
1:00:02tools. This is really really exciting.
1:00:05Okay, let me share my screen real quick.
1:00:07Meta just launched a new end toend
1:00:10creative solution and it's actually
1:00:13seriously powerful. When Meta released
1:00:15this, I went straight to my contacts on
1:00:16the engineering team and this is what
1:00:19they sent me. This isn't just a basic AI
1:00:22ad creator. It lives inside the creation
1:00:26process. It learns what's working and it
1:00:28builds from that in real time. Here's
1:00:32[snorts] what's really interesting. It's
1:00:34not just learning from your own account.
1:00:36It's referencing what's already
1:00:38converting across your entire niche.
1:00:40Instead of guessing which angle is going
1:00:42to land or what design might work, the
1:00:46AI already has a read on what's actually
1:00:48working inside your market right now.
1:00:51That is a wildly unfair advantage.
1:00:53You're not testing in the dark. You're
1:00:55not guessing. You're leveraging what's
1:00:57already working for everybody else. The
1:01:00machine does all the hard work so that
1:01:02you don't have to. The other thing the
1:01:03brand memory solves is the number one
1:01:06complaint I hear about AI ads. It
1:01:09doesn't feel like my brand. Brand memory
1:01:12learns your rules, your tone, your
1:01:14voice, and even what to avoid, so every
1:01:18creative it generates stays consistent
1:01:22at scale without you having to police it
1:01:25manually. All right, let's jump into a
1:01:26live ad account. I'm going to show you
1:01:28exactly how this works using my friend
1:01:30Anita's brand, Apothecary Me. So, the
1:01:33first thing we need to do to create an
1:01:35ad is pull up her website and we're
1:01:37going to promote her five-star flagship
1:01:39product, the number one illuminate face
1:01:42oil. Now, we're going to grab the URL
1:01:44for this product page and we're going to
1:01:46bring it into our ads manager. So,
1:01:48here's a new ad that we're going to set
1:01:50up and we're going to put this into the
1:01:51source URL. Then, we're going to take
1:01:54the product shots from our Shopify page
1:01:56and we're going to add them into the ad
1:01:58account. These are going to be what we
1:02:00use for permutations. Today, I'll give
1:02:02you the example using the first one. And
1:02:04for this ad, for the primary text, we're
1:02:06just going to scrub the landing page.
1:02:08And I'm going to use the description
1:02:09below the add to cart button for our
1:02:11primary text. And I'm okay with leaving
1:02:13the headline blank because we're just
1:02:15going to send people to this landing
1:02:16page. And Facebook is going to be able
1:02:18to pull in a headline for us from the
1:02:20metadata. And that brings us to the
1:02:22image generation tab, which if you
1:02:24notice has a really great refined look.
1:02:26What's working for you and how can we
1:02:28make it better? You'll also see a
1:02:30popular in your niche section. So, this
1:02:32is popular in beauty. And you'll see
1:02:35something that's performance-based. When
1:02:37you're running conversion ads, look at
1:02:39that high return on ad spend. And these
1:02:41look pretty good. What's popular in
1:02:43beauty and high return on ad spend is
1:02:45based on what's working across the niche
1:02:47from everybody else running ads. If you
1:02:50look at it, these are essentially
1:02:52plugandplay. So, a best practice here is
1:02:54that you can make one variation from
1:02:57each one of these sections and
1:02:58essentially be able to print three 322
1:03:01ads from right here alone.
1:03:04The one default product image has made
1:03:07us way more than we need it to and it's
1:03:09right out of the box. This takes minutes
1:03:11and it's completely native. So, let's go
1:03:13ahead and select three high return on ad
1:03:16spend options. I also want to show you
1:03:17this edit function because you have
1:03:19prompts. You can absolutely change some
1:03:22of this stuff if you wanted. Now, when
1:03:24you select edit here, you'll see a
1:03:26high-res preview. If for any reason the
1:03:29product has a weird thing going on in
1:03:31the text, which is the most common
1:03:32mistake that we see with AI, you can
1:03:34download this and then simply upload
1:03:36these to Gemini, give it your default
1:03:38product shot, have it fixed, download
1:03:41it, throw it into Canva, and magic erase
1:03:43that little star and throw it back into
1:03:45the ads manager. Now, of course, that's
1:03:47a little bit of work, but it's so easy
1:03:49to do, and you're not going to have to
1:03:50test hundreds of ads. These are the ones
1:03:52you know will work. The other really
1:03:55cool thing here is that after you hit
1:03:56next, it can generate videos for you.
1:03:59It'll do so in styles that work for your
1:04:01niche, including UGC, which is crazy.
1:04:04Now, let's check out these videos
1:04:05because I really want you to see what is
1:04:07just coming out of the box, again, for
1:04:09free in seconds that you can do natively
1:04:12and is built on already proven concepts.
1:04:16Now, I got to be honest, the people in
1:04:17this AI look great. And of course, you
1:04:20have the edit function. Currently, the
1:04:22edit will allow you to reorder the clips
1:04:24in the video. So, if there's just
1:04:26something that doesn't work, you can get
1:04:27rid of it. Perhaps it's a product shot
1:04:29that immediately brings in some UGC or
1:04:31in this case skin care showing a person.
1:04:34We might not want to make that claim
1:04:36that that's why they look this way. So,
1:04:38we can just get rid of it. And you also
1:04:40have the option of changing the music.
1:04:42If you wanted to, you could swap it out.
1:04:44And just like the image generation, you
1:04:46can download these videos. Perhaps you
1:04:47want to do a voice over. Awesome. It's
1:04:49all right here for you. And if you don't
1:04:51like it, just hit generate more videos.
1:04:54Or perhaps you just hit generate more
1:04:55and make 10 more videos right inside the
1:04:57box in a matter of minutes. Every single
1:05:00one of which is built on proven ad
1:05:02concepts. I don't know how much easier
1:05:04you need it to be. Also, when you hit
1:05:06generate more videos, it shows you all
1:05:09the images that you can use to generate
1:05:11the videos because we've given it our
1:05:13website and our landing page. It's
1:05:15scrubbing that to find source material,
1:05:18which is amazing. Plus, if you have any
1:05:21reviews or any customers or anything
1:05:23where somebody has taken a picture of
1:05:24themselves with your product, you can
1:05:25upload it right here and it'll bake it
1:05:27into the video for you, which again is
1:05:30amazing. I've seen big success here just
1:05:33taking images of the product and also
1:05:35happy customers from the review section
1:05:36and making videos from it. It's insanely
1:05:39easy. And after that, you have your
1:05:40opportunity for enhancements. And I
1:05:42think everybody sort of has their own
1:05:44opinion. If you'd like me to do a video
1:05:45on the value of each enhancement and
1:05:47what you should use and what you
1:05:48shouldn't use and what they all mean,
1:05:50let me know. I've got a ton of internal
1:05:52data on this and I'd love to share it
1:05:53with you. But drop a comment if you want
1:05:54to see that video next. That took
1:05:57relatively speaking no time at all. If
1:05:59you were to go to Claude or Chat GPT or
1:06:01Higsfield or any other AI resource,
1:06:03instead of publishing multiple ads right
1:06:05now, you'd be working through a prompt
1:06:07to hopefully get one usable concept and
1:06:10you would have no idea if that thing
1:06:11would actually work. Here, we've already
1:06:13built images. We've built videos. We've
1:06:15leveraged market information. So our
1:06:17confidence level around what we're using
1:06:19can be so much higher. The truth is the
1:06:22machine is telling us what the people
1:06:24who use the machine want to see based
1:06:27around the people who are making money
1:06:29with ads and the ads that they're using.
1:06:31This is completely unfair and I love it.
1:06:38Facebook just quietly pushed an update
1:06:40that is the single GREATEST FIX TO
1:06:42RISING CUSTOMER ACQUISITION costs I have
1:06:44seen in years [snorts]
1:06:45and no ONE ELSE IS TALKING about it. I'm
1:06:48going to show you exactly what it is and
1:06:49how to take advantage of it before
1:06:51everyone else catches on. I'm going to
1:06:53share my screen real quick. Okay, so the
1:06:55update is that Meta launched the
1:06:57customer life cycle strategy. You're
1:07:00already seeing the alert for this inside
1:07:01of your own dashboard. When this update
1:07:03was announced, I reached out to my
1:07:05friends on the Meta engineering and
1:07:06product teams, and this is the email
1:07:08they sent me. Then they gave me all the
1:07:10internal training docs that Meta reps
1:07:11will be taught in August. Here's how it
1:07:13works. When you create a new sales
1:07:15campaign, you'll see a new option at the
1:07:17adset level called customer life cycle
1:07:19strategy. And the fact that that lives
1:07:22here is the whole point. This isn't a
1:07:25creative tweak. It's not something you
1:07:26fix later after performance takes. It's
1:07:28a delivery strategy decision. So, you
1:07:31open up the drop down and you'll see an
1:07:32option that says acquire new customers.
1:07:35This is what I've watched a lot of beta
1:07:37testers and experts click. But it's not
1:07:39that simple. There's a few extra things
1:07:41that you need to do. So, let's get into
1:07:43it. Once you select acquire new
1:07:45customers, the next place people mess
1:07:48this up is exclusions. And this part
1:07:51isn't optional if you actually care
1:07:54about the output. And it's way more in
1:07:56depth than you might think. The biggest
1:07:59mistake here is relying on one week
1:08:01signal. You want both sides of the
1:08:04exclusion set up. Customer list
1:08:06exclusions, that's your backend truth,
1:08:09and website exclusions based on the
1:08:11purchase event. That's your pixel truth.
1:08:14Under exclude audiences, open your
1:08:16existing customers and then add a
1:08:18customer list, custom audience. And this
1:08:20is where I want you to slow down.
1:08:22Depending on your business, you may need
1:08:24multiple lists. All customers are not
1:08:27the same as high-v valueue customers.
1:08:28Refunded, charged back, and disqualified
1:08:30buyers very different. One-time buyers
1:08:32versus subscribers and repeat purchases
1:08:34are also completely different lists, or
1:08:36they should be. If you treat all the
1:08:38customers the same, you're going to
1:08:39train the system with just terrible
1:08:41data. The better your customer
1:08:43definitions are, the better your
1:08:44exclusions can be and the better these
1:08:46campaigns will perform. Now, don't skip
1:08:48ahead, but when we hit the best
1:08:49practices chapter later, I'll show you a
1:08:52fun little exclusion trick that can make
1:08:54your new customer campaigns pull in
1:08:56higher quality first-time buyers, not
1:08:58just people who haven't bought before.
1:09:00And that is a cheat code, especially to
1:09:03your contribution margin and LTV. We'll
1:09:06get more on that in a little bit. Now,
1:09:08the other big update here is purchase
1:09:10event website audiences. And this is a
1:09:13really big deal that again nobody's
1:09:16talking about. Meta now lets purchase
1:09:18custom events support a retention window
1:09:20of up to 730 days. Meaning you can
1:09:24exclude purchasers going back 2 years.
1:09:27And that matters because existing
1:09:30customer [clears throat] isn't the same
1:09:32for every business. For a subscription
1:09:34or consumable brand, supplements,
1:09:36skincare, coffee, whatever, someone who
1:09:38bought 60 or 90 days ago is clearly
1:09:40still an existing customer. But for
1:09:42something with a long repurchase cycle
1:09:44like a mattress or a luxury watch or
1:09:47some high ticket service, someone who
1:09:49bought 12 to 18 months ago might
1:09:52basically be cold again. So don't
1:09:54blindly choose the longest window just
1:09:57because it feels safer. Choose the
1:10:00window that matches your actual customer
1:10:03life cycle. It's in the name of the
1:10:06product. Let's respect that because the
1:10:08goal here isn't longer is better. And
1:10:11later again in the best practice
1:10:12section, I'll show you how to use this
1:10:14same windowing idea for win back too. So
1:10:17you're not just excluding smarter,
1:10:19you're also bringing the right customers
1:10:21back at the right time. Now, here's the
1:10:24thing that's going to throw a lot of
1:10:25people off. Once you set this up,
1:10:28delivery is going to change. Audience
1:10:30sizes and performance, CPMs and CPA,
1:10:33even the types of people that Meta goes
1:10:35after. Before Meta basically had
1:10:37permission to take the path of least
1:10:39resistance, which often means people who
1:10:40already know you, people who already
1:10:42visited your sites, people already on
1:10:44your list, and sometimes people who
1:10:46already bought and are likely to buy
1:10:48again. When you select acquire new
1:10:51customers and you configure exclusions
1:10:53properly, you're making that easy path
1:10:56harder. So yes, the numbers will move,
1:11:01but that does not automatically mean
1:11:04that performance got worse. And this is
1:11:07where measurement becomes the whole
1:11:09game. Because if you don't have a clean
1:11:11way to answer, was this purchase a new
1:11:13customer? This feature gets confusing
1:11:16fast. And it's way more in-depth than
1:11:18just the customer segments on your
1:11:20breakdown. At a minimum, you need one
1:11:22source of truth that can separate new
1:11:24customer purchases from existing
1:11:26customer purchases. And if you're at
1:11:28scale, a measurement partner can help,
1:11:30too. Whether it's Ammerst or Popsicle,
1:11:32Elevar, Blotout, use whatever fits. I
1:11:34don't care which one. I care that you
1:11:36can answer the basic question. Did this
1:11:39campaign actually acquire a new
1:11:41customer? If you can't measure new
1:11:42customers today, don't turn this on yet.
1:11:44Fix measurement first. If you can't tell
1:11:47with a column inside of your dashboard
1:11:48whether you're getting new customers or
1:11:50not, you have no idea what's actually
1:11:52happening. A quick tip, if you're
1:11:54looking at new versus engaged versus
1:11:57existing customers, you're not actually
1:11:59measuring CAC. You're measuring
1:12:00something completely different. But
1:12:02that's another video for another day.
1:12:04Your job is to give the system better
1:12:06instructions and better truth and to be
1:12:10able to quantify the outcome of your
1:12:12actual business objectives. And that
1:12:15brings us to the best practices. I made
1:12:17a couple promises earlier. Let's
1:12:19deliver. Because once you understand
1:12:20what the setting really is, your finance
1:12:23department will appreciate the button
1:12:24you clicked [clears throat] inside the
1:12:26ads manager. So number one, our first
1:12:28best practice is to maximize prospecting
1:12:31accuracy with exclusions. Combine
1:12:33[clears throat] exclusions wherever you
1:12:34can. Website audiences, mobile app
1:12:37audiences, customer list, campaign level
1:12:38exclusions where it makes sense. You're
1:12:41trying to cover your bases to make sure
1:12:43that the machine isn't accidentally
1:12:45filling your new customer campaign with
1:12:47people who are very much not new. And
1:12:50here's the fun trick I teased earlier
1:12:52because this is where you can scale the
1:12:53value of your growth, not just the
1:12:55volume. Instead of one giant exclude
1:12:58customers bucket, use tiered exclusions.
1:13:01So you're not only filtering out
1:13:02existing customers, you're also
1:13:04protecting the system from the obvious
1:13:06edge cases that contaminate the data.
1:13:08The goal is your new customer campaigns
1:13:11pull in higher quality first-time
1:13:14buyers, not just people who haven't
1:13:16bought before. That is extremely
1:13:19important. Second best practice is
1:13:21signal quality and freshness. This is
1:13:24boring, but it matters. Get your match
1:13:26scores as high as possible. If it's not
1:13:28at least an eight and a half, you have
1:13:29work to do. Use strong identifiers. It's
1:13:31not just email, phone, and mobile ad ID.
1:13:34You need to also update your customer
1:13:35lists frequently, dynamically. A stale
1:13:38list is not a source of truth. It's a
1:13:40historical artifact. It's a retroactive
1:13:42report card. If your list is old,
1:13:44incomplete, or messy, Meta's
1:13:47understanding of your existing customers
1:13:48is going to be old, incomplete, and
1:13:50messy, too. Third best practice,
1:13:52structure your customer lists for
1:13:55success. This is where I think most
1:13:57brands have a huge opportunity. And
1:13:59again, something nobody else is talking
1:14:01about. Don't just have one giant list
1:14:03called customers that tells the system
1:14:05and you almost nothing. Use labels,
1:14:08customers versus leads, high value
1:14:10versus low value, one-time purchasers
1:14:12versus repeat buyers, refunded
1:14:15chargebacks, disqualified versus good
1:14:17buyers. Because over time, Meta's life
1:14:20cycle tools are only going to get more
1:14:21important. And if your back-end data is
1:14:24a junk drawer, you're going to be
1:14:26limited by your own mess. Plus, that
1:14:29means you can optimize for different
1:14:31itemized business objectives. Your
1:14:33fourth best practice is measurement. You
1:14:36need one strong source of truth for new
1:14:38customer actions. It needs to be a
1:14:40column in your ad account. Something
1:14:42that says this is a new customer
1:14:44purchase. And often I use browser and
1:14:46server signals to back each other up.
1:14:48And you need a thirdparty tool to do
1:14:50this even if it's Zapier. Because
1:14:52nothing is worse than launching a new
1:14:53customer campaign and realizing that the
1:14:56data you're looking at with new and
1:14:58engaged and existing customers isn't
1:15:00actually how you track any of this.
1:15:03And your fifth best practice is to
1:15:05optimize for your most valuable actions
1:15:09wherever possible. Don't just optimize
1:15:11for a purchase. Standard events are for
1:15:14entrylevel marketers. Optimize for
1:15:17custom conversion events that represent
1:15:18your most valuable customers. That might
1:15:21be new customer purchase. It might be
1:15:23new customer purchase over some dollar
1:15:25amount. Maybe it's new customer purchase
1:15:27with a specific product. Maybe it's new
1:15:30customer purchase that hits your
1:15:31contribution margin threshold. Because
1:15:32the goal isn't to get more net new
1:15:35customers on paper. The goal is to
1:15:37acquire the right net new customers for
1:15:39your business so you become somebody
1:15:41that shifts from trying to sell products
1:15:43to somebody who's buying cash flow. And
1:15:45here's a really important nuance. Don't
1:15:47treat exclusions like a blunt
1:15:49instrument. Just because somebody spent
1:15:51a penny before doesn't mean they can't
1:15:52spend $100 later. So, don't build
1:15:54exclusions that permanently block low
1:15:57value customers from ever being
1:15:58eligible. Instead, use your optimization
1:16:01and measurement to promote the behaviors
1:16:04you want without writing off people who
1:16:06could become high value later. Just
1:16:08exclude the high-value ones. That's how
1:16:10you get more new high-value people. And
1:16:12one last reminder, if you optimize for
1:16:15new customers but measure blended CPA,
1:16:18you're going to be wrong about all of
1:16:19your data every single time. Now, a
1:16:22quick win back note because this is
1:16:24where that 730day window gets really
1:16:26interesting. You can use those purchase
1:16:28event windows to decide when somebody
1:16:30stops being hot and becomes winback
1:16:33eligible just like you do on email. So,
1:16:35you're not treating a 30-day buyer and
1:16:36an 18-month buyer like the same person.
1:16:39That's how you protect your new customer
1:16:41campaign from getting diluted while also
1:16:43having a clean and intentional way to
1:16:45bring back past buyers when it actually
1:16:48makes sense. So, let's talk about
1:16:50strategic implications. Here's the
1:16:52practical way to think about this
1:16:54because one of the hardest things to
1:16:55separate inside of the Facebook ad
1:16:57account has always been, are we actually
1:17:00growing the business, or are we just
1:17:02harvesting demand from people already in
1:17:04our ecosystem? So, this matters a lot if
1:17:07you're in a world like consumables and
1:17:09subscriptions, TTOC brands with a real
1:17:12LTV curve, or lead genen, where cheap
1:17:14leads don't mean good customers. This
1:17:16setting gives you a much cleaner way to
1:17:18tell Meta, "Go find new people, use my
1:17:21customer data, use my purchase
1:17:23exclusions, use my definition of an
1:17:25existing customer, and optimize towards
1:17:28the most valuable net new customers."
1:17:30But let's be clear, this isn't a magic
1:17:32button. This is not a substitute for
1:17:35good data. This is not a substitute for
1:17:37good creative, and it's definitely not a
1:17:40substitute for knowing your margins and
1:17:42what a customer is actually worth.
1:17:43Because the cheapest customer is often
1:17:45the person who is already going to buy
1:17:47or the one who is least valuable. And
1:17:50the most valued customer is often one
1:17:52you haven't trained the machine to go
1:17:54get. And this is what this update is
1:17:58really really about. Meta is giving
1:18:01advertisers a way to express life cycle
1:18:04intent inside the campaign itself. It's
1:18:07not just get me purchases, but get me
1:18:10the kind of purchases that expand my
1:18:13customer base to scale my cash flow. Use
1:18:16it correctly and you're going to get
1:18:17much more effective and efficient
1:18:19customer acquisition. But if you're
1:18:21lazy, it's just going to make your
1:18:22reporting messier. So, let me show you
1:18:24the exact checklist I use to decide when
1:18:28to turn this on and when to leave it
1:18:30off. And I'm also going to give you my
1:18:32practical recommendation for your
1:18:34situation based on how much you're
1:18:36spending because the right way to use
1:18:38this at $100 a day is not the same as a
1:18:41way to use it when you're spending
1:18:42$10,000 a day. Let's get into some real
1:18:45context. So, here's the checklist.
1:18:48Create a new sales campaign. Go to the
1:18:50adset level. Find the customer life
1:18:53cycle strategy section. Select acquire
1:18:56new customers. and configure the
1:18:58customer list exclusions and configure
1:19:00website purchase event exclusions.
1:19:03Choose the retention window that matches
1:19:06your business model. Make sure your
1:19:09customer list is fresh and dynamic. Set
1:19:12up new customer measurement. Watch new
1:19:16customer CPA CAC, not just blended CPA
1:19:19overall. and evaluate the campaign based
1:19:22on what you actually asked it to do.
1:19:25Because that's the theme of all of these
1:19:28new meta tools. The more control meta
1:19:30takes over the mechanics, the more
1:19:32important your inputs become. Your data,
1:19:34your exclusions, your customer
1:19:36definitions, your measurement, and your
1:19:37economics. The machine is going to get
1:19:39better and better at showing the right
1:19:40ad to the right person. But it can't fix
1:19:42a bad business. And one expectation that
1:19:45will save you a lot of stress.
1:19:47understand that this is going to
1:19:49optimize for incremental attention. So,
1:19:53it's very likely that you're going to
1:19:54see an impact ripple, a halo effect
1:19:57across search and email as well as
1:19:59social because now you're pushing meta
1:20:01to find people who weren't already on
1:20:03the about to buy path. And don't judge
1:20:07this on day one. You haven't trained it
1:20:10to do this job well yet. So, it won't be
1:20:13nearly as good at the end of the first
1:20:15week as it will be at the end of the
1:20:17first month. Give it time to stabilize
1:20:19and expect the ads that work here to
1:20:20look a little bit different than the
1:20:22ones you run elsewhere. This is a
1:20:24different objective. These are slightly
1:20:26different customers in different places
1:20:27in the journey based on your ad spend.
1:20:30Here's how I think you should be using
1:20:33this. Let's break it down into some
1:20:35tiers. So, if you're spending $100 a day
1:20:37or less, set everything up as we showed
1:20:38you here, but don't optimize for new
1:20:40customers yet. Use the new customer
1:20:43conversion event as a column in your
1:20:45dashboard so you can make better
1:20:46creative and budget decisions. You can
1:20:49optimize manually and teach the machine
1:20:53and the account how to get better at the
1:20:56job. At this spend level, you typically
1:20:58don't have enough volume to take
1:21:00advantage of this. And you'll pay a
1:21:02really high premium in learning costs
1:21:05that probably won't show up at the end
1:21:07of the day as profit in your balance
1:21:09sheet. If you're spending $100 to $500 a
1:21:11day, you can test optimizing for new
1:21:14customers with what we've talked about
1:21:15today. But right now, honestly, it's
1:21:18still a bit of a luxury item. It can
1:21:20cost you more to test and take you much
1:21:22longer to interpret unless your
1:21:24incremental tracking is absolutely rock
1:21:27solid. If it were me, I'd wait till this
1:21:29feature has matured a bit. Let bigger
1:21:32spenders train the system and use all of
1:21:34their money to help Meta get a lot
1:21:36smarter. First adopters pay a lot more.
1:21:40You don't have to. Now, if you're
1:21:42spending a,000 to 2500, I think this is
1:21:45a worthwhile test if you have robust
1:21:48incrementality measurement across all
1:21:49channels and a clearly defined most
1:21:53valuable new customer and you can
1:21:55measure it. This is where the feature
1:21:57can actually become a strategic lever,
1:21:59not just a reporting problem. Now, if
1:22:01you're spending more than that, I think
1:22:03this is a great idea to try, but I
1:22:06wouldn't rush it.

About this transcript

This page contains the full transcript of New BEST 2026 Meta Ads Course After Andromeda by Professor Charley T, generated from the public captions YouTube serves with the video. The transcript has 14,628 words across 2,127 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.