YT YouTube2Text

YouTube2Text → Transcripts → the entire meta ads algorithm explained in 21 mins

the entire meta ads algorithm explained in 21 mins — Transcript

by Mark Builds Brands · 4,718 words · 692 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Hey guys, Mark here. In this video, I'm
0:01going to be explaining how the entire
0:03Meta Ads algorithm actually works. A lot
0:06of people think that running ads on
0:07Facebook is just all random or that if
0:09your creative is good enough, then
0:11everything else will just magically fall
0:12into place. Well, running profitable ads
0:14on Facebook is both an art and a
0:16science. The art is the creative, but
0:18the science is exactly what I'm going to
0:19be explaining in this video, how the
0:21algorithm actually works. And more
0:23specifically than a science, it's
0:24actually an equation. And this is the
0:26one equation that is actually going to
0:28determine how profitable your ads are
0:30going to be. Now, before I break down
0:32this equation, I need to explain to you
0:34the four-step process that Facebook uses
0:36to actually show your ads, the one
0:37update that broke everybody's ad
0:39account, and at the end of this video,
0:41I'm going to put in a little something
0:42special to explain what this actually
0:44all means and how the game really works.
0:46All right, so before we get too
0:47technical, this is what you have to
0:48understand. Facebook basically goes
0:50through a four-step process to determine
0:52which ad is going to show for any
0:54particular user. Every time somebody
0:56opens up their phone on Instagram and
0:58they scroll and see an ad behind the
1:00means there's literally a 200
1:01millisecond window where Facebook is
1:03doing all these incredible calculations
1:06and machine learning decisions to
1:08ultimately decide to put your ad in
1:10front of the user or somebody else's.
1:12So, in that one 200 millisecond period,
1:15there is literally tens of millions of
1:17ads that are competing for that one
1:19impression. And these are the four steps
1:21right here. First, you have retrieval.
1:23Second is light ranking. Third is heavy
1:26ranking. And the final piece is the
1:28auction. So, let me break this down for
1:29you. So, first up is retrieval.
1:31Retrieval is the most data-intensive
1:35portion of this entire process because
1:36as I said, there's tens of millions of
1:38ads, tens of millions that are competing
1:40for one individual spot. The process of
1:42retrieval is essentially Facebook going
1:45through and ultimately deciding, is the
1:47ad right now even remotely relevant to
1:49this user? Basically, like is it even in
1:52the right ballpark? Now, this is also
1:54the portion of the process where
1:56Facebook actually analyzes all the
1:58aspects of your creative to determine
2:00how to narrow down the audience. So, you
2:02may have heard this before, but this
2:04whole concept of your creative is your
2:06targeting. This is where Facebook is
2:08analyzing everything about your creative
2:09to determine who to show it to. This
2:11includes, but is not limited to, your
2:14hook, your format, the demographic of
2:18the users that are actually on your ad.
2:20So, for example, if you're running an ad
2:22that is featuring a woman that is 55
2:24years old, Facebook takes note of that
2:26and it's going to use that to power the
2:27targeting to skew much more towards
2:29women and skew much more towards older
2:31women specifically. Of course, all taken
2:32into context with what the creative is
2:34actually saying. It's also looking at
2:35your copy, it's looking at your script.
2:38If you're running video ads, it's
2:40looking at your thumbnail. It's even
2:41going and looking at your landing page
2:43to determine generally who is this ad
2:45trying to target. This is a
2:46non-exhaustive list. I'm really
2:47simplifying this. Behind the scenes,
2:49it's a pretty substantial, literally an
2:52algorithm in order to determine which
2:54variable is going to correlate to which
2:56piece of targeting. And this is probably
2:57a good time to mention the Andromeda
3:00update. Andromeda what is was an update
3:02to the sorting process of the retrieval
3:05steps specifically. Mostly just because
3:07there was a massive influx of AI
3:08creative that was absolutely flooding
3:11Facebook to the point where their old
3:12model of sifting through all these
3:13creatives was not becoming very
3:15effective. This is not the only reason
3:16for the Andromeda update, but I believe
3:18this is one of the biggest ones simply
3:20due to the fact that AI has massively
3:22increased creative production for all
3:23advertisers. And now that's why Facebook
3:26is mostly looking at creative to
3:27determine targeting, not what you have
3:29going on within your ad sets. That's why
3:31you've probably seen a decline in
3:33audiences that are actually able to
3:35sustain high budgets profitably. Like I
3:38rarely use audiences like interest, and
3:40if I do, they're pretty capped in terms
3:42of budget or, you know, lookalikes or
3:44even custom audiences. Like I don't set
3:46any real [ __ ] targeting when it comes
3:48to my retargeting. I don't put any
3:50custom audiences in there. I literally
3:52just still run completely broad. And
3:54most high-skill advertisers that I speak
3:56to, they're doing something relatively
3:57similar. Maybe one exclusion of
3:58purchasers, but like other than that,
4:00very, very broad. So, if that was kind
4:02of the old way of doing targeting on
4:04Facebook, the new way is to do a lot of
4:06broad targeting. It's to do a lot of
4:08Advantage Plus campaigns, and it's
4:10usually within reason, within reason.
4:13Giving Facebook control. Like the reason
4:15that I say it's within reason is because
4:17if you give Facebook all the control,
4:19the algorithm is good, don't get me
4:20wrong, but holy [ __ ] the amount of
4:22suggestions that Facebook gives you that
4:24are horrible pieces of advice that are
4:27just going to drain your [ __ ] bank
4:28account is obscene. And it's always been
4:30this way, too. They're still a company
4:32at the end of the day, and you are a
4:34customer. You are paying Facebook, so
4:36they still want to get your money. I'll
4:37save that for later in the video. Now,
4:39the second step is going to be light
4:41ranking. This step barely deserves a
4:43description. It's basically taking the
4:44now millions of ads instead of tens of
4:47millions, and reducing it down to
4:49thousands. There's really not a a whole
4:51lot more to say about the light ranking
4:52process. Now, the heavy ranking is the
4:55next step, which, you guessed it, if
4:56it's going from thousands of ads, you're
4:58now trying to get to a couple hundred.
5:00Now, at this point, this is like your
5:02real, legitimate competitors. Like this
5:04is the stage where competitors are
5:06actually something that you should take
5:07into consideration, because you don't
5:09have tens of millions of competitors.
5:11You just don't. You probably only have a
5:13handful, at least in the eyes of
5:14Facebook, you only have a handful.
5:15There's only 10 million advertisers
5:18total on Facebook, which may seem like a
5:19massive number, but also think about we
5:21are in e-commerce specifically, at least
5:23I'm in e-commerce specifically, which
5:25immediately narrows it pretty
5:26substantially. Now, the really important
5:28piece about the heavy ranking step is
5:30that this is where that equation
5:32actually comes into play. So, I want to
5:33take a second to talk about that,
5:35because I remember Facebook released
5:37this. They have since withdrawn it, but
5:38they released this documentation to
5:41explain how they actually assign a
5:43specific value for every single ad when
5:46they are fighting for an impression. And
5:48the one that has a higher value,
5:49obviously, is going to win the
5:50impression and the one with the lower
5:51value is going to lose the impression
5:53and get served a lower quality
5:54impression. All right, so this is the
5:55equation I was talking about. This is
5:57actually directly from Facebook.
5:58Somebody swiped it and it is still on
6:01their website, but it's not on Facebook.
6:03But this is how it's actually working
6:04underneath the hood, all right? So the
6:05total value, this is what you want to
6:07aim on maximizing, is equal to this
6:10piece first of maximizing advertiser
6:12value plus maximizing consumer value and
6:15consumer experience. So the piece to
6:17maximize your advertise value, number
6:19one is going to be your bid, which we'll
6:21get to in a second, times your estimated
6:23action rate. Since 99% of us and 99% of
6:26people watching this video are running
6:28conversion campaigns, that essentially
6:29means what is the estimated action rate
6:32to get a conversion. And if you look
6:33down here, you can see they actually
6:35break it down specifically for a
6:37conversion optimized ad, which is the
6:40estimated click-through rate times the
6:42estimated click-to-conversion rate. So
6:44the way to get gamify this, of course,
6:46is you want to have really high CTR.
6:48Historically, Facebook just loves ads
6:50that have really high CTR, and the
6:51reason why is because that is the the
6:53most pure indication of how relevant
6:57that one particular ad is to the user.
6:59So Facebook is going to reward you
7:01typically over time and as you spend
7:03more with lower cost to advertise, lower
7:05CPMs, and ideally higher quality
7:08traffic, too. So this is kind of like
7:09the first piece of that. The second
7:10piece is going to be you want to have a
7:12high, they call it like
7:14click-to-conversion. It's basically just
7:16your conversion rate, which is out of
7:17all the clicks that you had, how many
7:18actually purchased. And again, if you're
7:20running a e-commerce business, this is
7:23going to be like a legitimate purchase
7:24because that's typically what you're
7:25optimizing for. But if you're in, you
7:27know, the info industry, for example,
7:29you're optimizing for calls, what is the
7:31ratio between clicks to calls that
7:33you're getting or clicks to applications
7:35or insert your conversion here, it
7:36doesn't really matter. Facebook views a
7:38lot of this largely the same, at least
7:40in the context of this equation. So the
7:42way I like to think about this is the
7:43estimated click-through rate, I think
7:45it's actually more than the
7:46click-through rate. I think it's a
7:47culmination of all your soft metrics.
7:50So, that is your CTR, but that's also
7:51your CPC, your CPMs, your hook rate,
7:53your hold rate, cost per 3-second video
7:55view, like all these metrics that are
7:57associated with getting somebody engaged
7:59and getting them to click. And then the
8:01second half of this is how good are you
8:04at monetizing that click, taking that
8:06click through your funnel, and turning
8:08somebody who's a prospect into a buyer.
8:10Pretty straightforward. And this last
8:11piece is basically Facebook trying to
8:13save their own ass from advertisers that
8:15are incredibly aggressive and cause a
8:17terrible user experience for the users.
8:19Because if this piece just did not exist
8:21and we didn't care about the user
8:22experience whatsoever, advertisers would
8:24just flood the platform and completely
8:26just destroy Facebook as a platform
8:28because you just have [ __ ] loads of
8:30scammers, not like there isn't scammers
8:31on Facebook anyways, but there would be
8:33way more. Like it would be like 50/50
8:35instead of 80/20, which is roughly what
8:37it is right now. So, this is Facebook
8:38saying, "Hey, I want you to have the
8:41most relevant ads to the user, which is
8:43CTR, all the soft metrics, and I want
8:46you to actually get conversions for all
8:48the clicks that you're getting, which is
8:50the conversion rate, and I don't want
8:51you to piss off all of our [ __ ] users
8:53in the process." And that's the last
8:54piece. All right, so we took it from
8:56thousands of ads and now we're down to a
8:58few hundred. In fact, sometimes it could
9:00be even less than a few hundred. But now
9:02we have the final step in the process,
9:03which is going to be the auction. And
9:06the auction is exactly what it sounds
9:08like. It's all the advertisers that are
9:11left competing for that one impression
9:12and basically who can have the highest
9:15bid, of course with all the context of
9:17that equation that I said previously,
9:18that's going to be the one that wins the
9:19impression. So, majority of people
9:20watching this video right now are
9:21probably running what's called, you
9:23know, lowest cost bidding or also known
9:25as like auto bidding, which Facebook is
9:27just doing all the bidding for you,
9:29trying to get you conversions for the
9:30lowest cost possible. This is, like I
9:32said, auto bidding. You can also do
9:35manual bidding, which can come in a few
9:37different ways. Manual bidding, you can
9:39do something called cost caps, which is
9:40where you set a cap on the maximum
9:42amount that you're willing to pay to
9:44acquire a customer. You can also do bid
9:46caps, which is similar, except it's the
9:48maximum amount that you want to bid to
9:50acquire a customer. And there's some
9:51other stuff that Facebook is always
9:53releasing as well to help you optimize,
9:55you know, maximum value, target ROAS,
9:57stuff like that. But I'll tell you this,
9:58there's a lot of people that are really
10:00obsessing about cost caps or bid caps or
10:03whatever the hot, you know, manual
10:05bidding method is at the time. I can
10:07tell you for a fact, I have seen ad
10:09accounts, I have ran ad accounts that
10:10are spending multiple five figures per
10:12day. I have seen one spending six
10:14figures per day, all on lowest cost, and
10:16they're good ad accounts. They are
10:17profitable. Other people literally only
10:20run bid caps. You can see both sides of
10:21the spectrum, but the point is you
10:24should choose what is actually going to
10:26solve the constraint in your business
10:27for you. Because I talk to people all
10:29the time, they're like, "All right, in
10:30order to get from like 1K a day to 10K
10:32per day, I just got like media buy my
10:33way there." And like there's a little
10:35bit you can do, don't get me wrong. I'm
10:36not the guy that's saying like media
10:37buying is dead. I'm defending media
10:39buying. At the same time, you have to
10:41think like an entrepreneur, not like a
10:43media buyer. And just to show this
10:45entire process visually, all right, this
10:46is how Facebook actually picks ads in
10:49under 200 milliseconds. So this is the
10:51first step, tens of millions of ads
10:53competing for one impression. After
10:54that, we go through the light ranking,
10:57right? Now you can see there's
10:58substantially less, say like roughly
10:595,000. Next step, there's just a few
11:01hundred. Now we're going to go into
11:03heavy ranking, which is boom, now we're
11:05on the final step, which is actually
11:08competing who's going to have the
11:09highest bid. There's very few
11:10advertisers left at this step, until
11:12finally your ad is actually served. You
11:14have the highest bid, you win the
11:15impression. So again, retrieval, light
11:17ranking, heavy ranking, and the auction
11:20as the final step there. Okay, so now
11:22you know how this filtering process
11:23works, how does that actually affect
11:25your ability to create winning ads
11:28consistently and more importantly
11:30predictably churn out winners into your
11:31ad account. So there's a few other
11:33things that aren't part of that
11:34four-step ranking process that are still
11:36really important to understand so that
11:37you can have the most profitable ads as
11:39possible. All right, first I want to
11:40talk for a second about the learning
11:42phase, okay? So, the learning phase,
11:44according to Facebook themselves, is
11:46basically a phase that your ads go
11:48through, your ad sets go through, your
11:50campaigns go through, but especially at
11:51the ad level to acquire conversions so
11:54that it can learn about your customer
11:56and then you kind of exit the learning
11:58phase and go into like mature phase
12:01where you're getting more predictable
12:02consistent results. Here's my beef with
12:04this whole [ __ ] term. Like that's the
12:05theory. Let me tell you the [ __ ]
12:06reality. Your ads are always learning.
12:08They never stop learning. There's not a
12:10learning phase and then they get out of
12:11learning phase and they're just going to
12:12be good forever. Everything dies at some
12:14point. It's just the circle of ads. So,
12:17if you are obsessing over the learning
12:18phase, just stop. Of course, the ads
12:21initially are going to be a little bit
12:23more volatile, a little bit more bumpy,
12:25and sometimes you see improvements over
12:26time with more consistency, but
12:28sometimes you don't. Sometimes the
12:29solution is literally just to make
12:31better ads. So, spending too much time
12:34on the learning phase is not super
12:36smart. The main thing I want to say here
12:37is like if you're not spending a lot of
12:39money on ads, maybe you're spending, you
12:40know, 50 bucks a day, a few hundred
12:42bucks per day, even up to like a
12:43thousand dollars per day, I think that's
12:44kind of like the barrier. If you see
12:46your results being like up and down,
12:48like it literally looks like a freaking
12:49sine wave, that is normal, okay? That's
12:53100% normal. That's how the game works.
12:55[ __ ] I experience inconsistencies. I
12:57spend way beyond 1K per day. And most
13:00advertisers do. Facebook is an emotional
13:02ass platform, man. It's a bunch of
13:04emotional human beings like you and me.
13:06Do you really expect it to be the most
13:08consistent thing on the world? If you
13:09want super duper consistent results,
13:11like go advertise on Google
13:13>> [laughter]
13:13>> and uh you know, enjoy your time, all
13:15right? Google's cool. It's got its own
13:16issues, but with Facebook
13:18inconsistencies is part of the game, but
13:19especially at lower levels of spend,
13:21you'll see people just bitching about,
13:23"Oh, I was at 3X ROAS and now I'm not
13:25profitable." I'm was "Well, how much do
13:26you spend?" He's like, "A hundred
13:27bucks." I'm like, "Come on. What what
13:28are we doing here? Like, what do you
13:29What do you expect?" So, the solution
13:31there is to, when in doubt, zoom out.
13:33Look at larger time frames with larger
13:35amounts of data, and you'll see what's
13:37really happening. Instead of being like,
13:38"Oh, today was good really good. Next
13:40day was really bad." I don't care. Look
13:41at 7 days. Look at the most amount of ad
13:43spend that you can, and that's when
13:44you'll make the best decisions, and not
13:46these emotional decisions where you're
13:47just going with the [ __ ] flow of
13:48Facebook taking you for a ride. Nobody
13:50wants that. All right, next I want to
13:51talk about signal quality and your EMQ
13:54and Conversions API. So, Facebook
13:57performance is often times dependent on
13:59the data that you're sending back to the
14:01ad account and back to the Facebook
14:02servers so that it can learn about your
14:04customers and reduce your cost to
14:06advertise long-term, hopefully. And with
14:08that, that means that you have to be
14:10sure that you're sending the best
14:12quality data you possibly can to
14:14Facebook because the better quality data
14:16that you send them, often times, they're
14:18going to be able to learn more about
14:19your customers and send you better
14:21customers in the future, make your ads
14:23more efficient long-term. So, you can go
14:25into your Events Manager and go to the
14:27purchase event specifically, and you'll
14:28see something called the EMQ, which is
14:30your event match quality, and it'll be
14:32on a scale of 1 to 10. I'm not saying
14:34it's a direct correlation like 10 out of
14:3610 is super profitable ads, like five
14:38out of 10 you're going to be dog [ __ ] I
14:39actually think this whole concept is
14:41overrated, but I think generally
14:43speaking, it's a good thing to try to
14:45send good data to Facebook. And you can
14:47do that through Conversions API, which
14:49is simply sending your data back
14:51server-side and not only on the browser.
14:53For those that have been advertising for
14:55long enough, you know this became a big
14:56deal around iOS 14, which was launched
14:59in 2020, but it actually went into
15:01effect, I think, early 2021, and
15:03everybody's ads just kind of went to
15:04[ __ ] because I don't look at any
15:06conversions after the click. Like, I
15:09don't look at my CPA on Facebook at all.
15:11I always use third-party attribution
15:13tools, you know, whether that's Triple
15:14Whale, whether that's Hyros. There's so
15:16many of these out there. Like, you can
15:17build one custom now. [ __ ] you could
15:18vibe code one probably in a few hours
15:20and have your own solution that you
15:21don't have to pay for at all. The issue
15:23is, because of what the update that iOS
15:25launched, all the users can now opt out
15:27of tracking from Facebook. So,
15:28everything becomes really inaccurate.
15:30Now, if you're able to use like
15:31conversions API, any server-side
15:33tracking, you can improve your EMQ and
15:36get better data into the Facebook
15:37servers to help you optimize with your
15:39ad account your ads a lot better than
15:40you are right now. So, to summarize this
15:42point, try to send Facebook good data.
15:45And also, take all the conversion data
15:46with a grain of salt. Okay, let's go.
15:48Now, the third thing here is going to be
15:49the whole concept of creative diversity.
15:51So, because of Andromeda, there was this
15:53really weird narrative where people are
15:55basically saying, "Oh, there's kind of
15:56like two types of ads that you can
15:57test." You know, you can test something
15:58completely new, we'll just call that
15:59like a new concept and that new concept,
16:01or you can test, you know, a variation.
16:03You can only do one of the two, which
16:04it's really not that simple, but let's
16:06just say for these purposes, this is
16:07what it is. And people like, "All right,
16:09all you need to do to survive Andromeda
16:11is just like pump out more new concepts
16:13and do less variations." No, variations
16:15are still highly highly effective. What
16:18Andromeda did is let's say like this is
16:20within an ad set and we're looking at
16:22the ad level, and each one of these is a
16:23new ad, right? You have ad one, two,
16:25three, and four. So, every time I launch
16:28creatives, I'm actually launching
16:29variations. So, this entire ad set, per
16:32se, will be one net new concept, but ad
16:35number two, ad number three, and ad
16:37number four will all be variations of
16:40that original net new concept. So, it's
16:42not necessarily variations of the
16:43winner. Like it's difficult to get a
16:44variation of like your top ad in an ad
16:47account to outperform the top ad. It'll
16:50do decent. It's typically not going to
16:51outperform, at least in my experience.
16:53But, variations are still highly
16:54effective because you're basically
16:56getting more at-bats at the concept
16:58level. Technically, you're testing a new
16:59concept, but you can test one concept
17:02with three different hooks, and it turns
17:04out that this variation actually ends up
17:06being the winner, and this original like
17:08net new concept is actually the loser.
17:10But, if you said variations are stupid
17:12and they don't make any sense, then you
17:13never would have figured that out, and
17:14you would have just concluded that this
17:16net new concept was a loser. When in
17:18reality, you're a loser
17:20>> [laughter]
17:20>> for believing that.
17:22I swear I could rant on all this stuff
17:23for so much longer. I'm trying to stay
17:25focused here. Let me distill all this
17:26technical crap into something that you
17:29can actually take away. Because
17:31basically what we just did is we went
17:33through all the super autistic
17:36scientific mathematical explanations of
17:39how the Facebook ads algorithm actually
17:41works, but let's take our [ __ ] hat off
17:43for a second and think like a Chad,
17:45okay? Three things you should take away.
17:48All right, number one, your creative is
17:50the targeting. Stop trying to [ __ ]
17:51around so much with the ad sets
17:53targeting, custom audiences, lookalikes.
17:55That can be a little experimentation
17:57sandbox, but for the most part, run
17:59broad. The creative is the targeting.
18:00Number two, make good creatives that
18:02thought people from their doom scroll.
18:03Of course, easier said than done, and
18:04that's not even what this video is
18:06about, but if you just have damn good
18:08creatives, like that's literally half of
18:09the [ __ ] formula. It's just having
18:11that high CTR, high soft metrics, of
18:14course, excluding the conversions. Like
18:15that's half the [ __ ] battle right
18:16there. And then third is going to be
18:18don't [ __ ] up the user experience. Now,
18:20I have a portion of this video that I
18:22intentionally left for the end of this
18:23video for people that are genuinely
18:25serious about figuring out how Facebook
18:27works. And all this stuff that I've said
18:28so far is what we'll call, you know, by
18:30the books. A lot of it is in, you know,
18:32Facebook's own documentation or
18:34documentation that they had years ago
18:35and they have since removed, but let me
18:37tell you how the game really works.
18:39Facebook is a for-profit [snorts]
18:41company based in America. Facebook is a
18:44massive, massive company. We can assume
18:46they probably don't do everything by the
18:47books. So, what I'm really leading to
18:49here is Facebook doesn't play by the
18:51rules. I could go down the list of what
18:54this actually means, but at the end of
18:56the day, as I said, Facebook is a
18:57for-profit company. And like most
18:59companies, they are trying to get the
19:01most amount of money from the largest
19:03advertisers, and then all the smaller
19:05ones that cause like 99% of their
19:07headaches, they're not going to treat
19:08them super duper well. So, a lot of you
19:10have probably launched ad accounts and
19:12got them disabled, have ads rejected,
19:14have ad accounts banned, have business
19:16managers banned. Maybe you've had
19:17profiles banned, but you have to
19:18understand that Facebook is a company,
19:20and they're a company that is public.
19:23And what happens when you run a public
19:24company is you have to answer to your
19:27shareholders, which means depending on
19:30how the earnings reports come out
19:32regarding Facebook, Facebook is either
19:34going to constrict their guidelines,
19:37their restrictions, and this is where
19:38you see ban waves come through, which is
19:40where everybody's just getting their ad
19:42account banned, everybody's dealing with
19:43business manager bans, page bans, and
19:45then a month later it's like totally
19:47fine. That's all due mostly to earnings
19:51reports that are coming out directly
19:52from Facebook. So, depending on how
19:54those look, they will either be way more
19:57laid back, way more chill, or super
19:59duper strict. That's why saying, "Oh,
20:01you can't run like this type of format
20:03of ad on Facebook." One person runs it,
20:05gets approved. Another person runs an
20:06identical creative and gets it rejected.
20:08Why is that? Well, Facebook don't play
20:09about their rules, all right? So, what
20:11do you do about this? I mean, I'm not
20:12going to get too much into specifics
20:13because I don't know if I can. I think
20:15the main thing anyone can take away from
20:16this is just question everything. Like,
20:19when I was talking about the CAPI and
20:21the signal quality and the EMQ score,
20:23technically speaking, like it makes
20:24sense logically, theoretically, if
20:26you're able to get higher quality data
20:28back to Facebook, they should be able to
20:29optimize better. However, I haven't seen
20:32a direct correlation between that. I
20:34have had really, really good server-side
20:36setups to where my EMQ was like 9.8,
20:399.9. The little thing on my freaking ad
20:41account was like, matter of fact, I'll
20:43show you. I'm literally at a zero
20:45>> [laughter]
20:46>> at a zero opportunity score, and this ad
20:48account is incredibly profitable. So,
20:50all the [ __ ] from Facebook, they're
20:52still wanting to get money from you. So,
20:54just because they say something or it's
20:55in their documentation or like their
20:57CEO, CMO, head of advertising is like,
20:59"This is what advertisers should do."
21:00[ __ ] it. Think for yourself, question
21:02everything. And I want to take a moment
21:04of silence for everyone that listened to
21:06all the Facebook suggestions at the ad
21:08level that said they would lower your
21:10cost per result by 11% and then their
21:12ads went to [ __ ] RIP. Don't listen to
21:14Facebook. Other than that, guys, I hope
21:16you enjoyed. Thank you so much for
21:17watching this video. See you on the next
21:18one. Peace.

About this transcript

This page contains the full transcript of the entire meta ads algorithm explained in 21 mins by Mark Builds Brands, generated from the public captions YouTube serves with the video. The transcript has 4,718 words across 692 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 choose a caption track available on the video.

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, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.