Consulting.com
Consulting.com
 

Horizontal Scaling: How To Scale Facebook Ads To The Moon

Horizontal Scaling: How To Scale Facebook Ads To The Moon

Summary


Spending $15 mill (of your own money) on ads teaches you a few things. Your awareness expands far beyond Facebook audiences, placements, and ad copy best practices -- you become one with the machine -- you become aware of its structure.

In this module, I show you how to scale your Facebook ads without increasing costs and experiencing diminishing returns. To do this we use "horizontal scaling", where we decentralize and distribute spend across 8 different dimensions. This will blow your mind!

Most people learn something by understanding it's parts. Mastery is when you learn the parts and the interconnections between the parts. The full ontological structure and the emergent behavior that arises from it.

This video shares some of my best insights on scaling Facebook ads to levels you never imaged possible. Even for the most seasoned Facebook advertiser, this video will change everything. The video is an actual training module from "Uplevel Consulting", a 9-week online course that shows you how to scale your business to 7-figures with hyper-systemization. This module gives you a taste.

Check it out and let me know what you think in the comments?


Here's what we cover:

•  What is horizontal scaling and how does it work?

•  Method 1 — Duplicating adsets at max budgets

•  Method 2 — New audience interests

•  Method 3 — Creating lookalike audiences

•  Method 4 — Creating new ad variations

•  Method 5 — Creating new ad angles

•  Method 6 — Targeting additional countries

•  Method 7 — Scaling retargeting campaigns

•  Method 8 — Decentralized architecture


Resources mentioned in this video:

•  9 Facebook Scaling Methods — Download PDF here.

•  Uplevel Consulting — A proven and tested 9-week program that shows you how to scale your business to 7-figures using hyper-systemization. Learn more about Uplevel Consulting here.


To Your Success!

Sam Ovens & the team at Consulting.com.

Transcript / MP3

00:00
/
00:00
Hey everyone, Sam Ovens here, and welcome to this module called Horizontal Scaling. Now in this module, I'm going to show you how to scale your Facebook ads rapidly to insanely high levels. I'm not just talking about how to scale from $100 a day to $500 a day, but if you want to do that, you're going to learn how to do that in this training too, but I'm talking how you can scale to $5,000 a day, $10,000 a day all the way to $40,000 a day in spend. If you're doing that in spend, then you're making like $120,000 a day and that's a lot. In this module, I'm really just going to pull back the curtain and unleash all of the best strategies and tactics that I've accumulated and learned over I think it's been about seven years of using Facebook ads. Everything I've learned spending millions of dollars on ads, making tens of million dollars from Facebook ads and seven years of it, I'm going to really boil it down and distill it and crystallize it and give you the best, most up to date cutting edge scaling strategies. Here's what we're going to cover. We're going to start off with what horizontal scaling is and how it works. We're going to explain that first and foremost. Then we're going to cover eight scaling methods, eight proven methods that you can deploy and execute to really take your ad account to the next level. Now the first one, method one is duplicating ad sets at max budgets. The second one is new audience interests. The third is creating lookalike audience. The fourth, creating new ad variations. The fifth, creating new ad angles. The sixth, targeting additional countries. The seventh, scaling your targeting campaigns. Method eight is decentralized architecture. Eight proven methods, and in this module we're going to dive into each one. I'm going to show you what it is, why we use it, the logic behind it and what we would use it, in what use case, in what scenario. Then I'm going to show you how to set it up. I'm going to show you in my Facebook ads account where to click and what to do and everything. We're going to cover the full end to end, what it is and how to actually execute it and manage it for each one of these eight methods. Let's get to it. What is horizontal scaling? Scaling is when we find winning ad sets, we scale them by increasing the spend and the reach and the size of the campaign. Now let me give you an example. Let's say you're spending $100 per day on Facebook ads and you're making $500 a day back. That's your return. You're putting $100 a day into Facebook and your sales are about $500 a day. In this situation, you want to increase spend as much as possible so that you can increase your return as much as possible. If you find a machine on the side of the road and you put a dollar in and five comes out, you want to try putting another dollar in. If five keeps coming out, you put another dollar in, and then you want to start putting as many dollars into that thing as you possibly can because why would you not? That's why we want to scale. When we find something that works, we want to take it to the limit. We want to take it to the extreme. Now the general rule is that typically if something works at a small scale, it should work at a large scale. We see this without proof of concept. If we find ... first of all we see this with our market research. If we find a bunch of people in a niche that all have and share a similar problem, then there's a high chance that this is widespread across the whole niche. Then if we get one client and we help them and they get results, and then we end up if we get five clients and we help them and they get results, there's a high chance that we've got a proof of concept there, that we've got something that can be widely applied to all participants in the niche to help them get a promising result. There's this general rule that if it works at a small scale, it should work at a large scale. The problem is when most people try to scale their campaigns, they increase spend. They increase the ad spend, but they don't increase their return. They simply pay higher prices for the same amount of traffic and customers, which is what we call diminishing returns. For example, they're spending $100 and making $500. Then say they start spending $1,000 but they only make $1,500. As they spend more, their rate of return goes down. They're still making more money, but their rate of return is going right down. This is the problem. The solution is to increase brand without increasing prices. We do this using a strategy that I have created called horizontal scaling. Horizontal scaling is where we decentralize and distribute spend across multiple dimensions keeping the same ROI at a larger scale. Now you might not understand what the hell that means in these words, so I'll make it simple for you. I'll explain it to you now because this is like a new way of thinking. It's like a new paradigm for advertising on the internet. Once you understand it like this, it will make absolute sense, especially if you've run ads before, it'll just click for you. Here I've got something on the screen called the Scale VS Return Continuum, or the Scale VS ROI Continuum. Now over here on the left side we've got low scale. Over here on the right hand side we've got high scale. Now typically what we see is when we're at a small spend, like if we're only spending $100 a day, $50 a day, we get a high ROI. For example, we spend $100. We get $2 costs per clicks on average. That means we get 50 clicks and let's just say we get $500 in sales, and that means that if we spend $100 and we made $5, we've got a 500% ROI. That's what I mean by low scale high ROI. Then as we start scaling up and we go along this side of the continuum, we start getting high scale but low ROI. Let's say we spend $1,000, but our CPCs go up to $5. That means we don't get ten times the clicks. Now if we got ten times the clicks, we'd be getting 500 clicks, but no, we only get 200 clicks. We only get four times the clicks for ten times the price because the price increased. This means that we don't make as much revenue. We only get about $1,000 in revenue, and then we get a zero percent ROI. This is the typical thing. This is the typical scenario where somebody has really good ROI and everything's working here at this scale, and they scale it up a bit and the prices blow out and they make pretty much nothing. What they do in this situation is panic and retreat back to the safe zone. Then they stay there forever until maybe they one day get the balls to try it again, [inaudible 00:08:42] again and then they retreat back to the safe zone, and then they probably stay there again until they forget why they were staying there, and then they do it again and then they come back again. This is what happens. This is the problem that everyone faces. It's because nobody has figured out how to transcend this continuum. Nobody has figured out how to get high scale and high ROI. Everybody is at the mercy of these fours here until now. What I want to do now is introduce you to the scaling strategy that we have used, honed, perfected and refined, which is called Horizontal Scaling. Instead of scaling up with one, we scale across with many. Just imagine this for a second. Imagine if you're spending $100 a day and you're making $500 in return, and you want to scale up to $1,000. Well by simply increasing the budget from $100 to $1,000 on that existing ad set or whatever, we blow everything out. Now imagine if we could just clone ten of these small clusters here, because we know these clusters perform at $100 spend, so what if we just created ten of them instead of increasing the size of the one up ten times? We're achieving the same end result of spending ten times more, but the way we've architectured and distributed and load balanced that spend keeps all the individual clusters at $100 each. This is what I mean by horizontal scaling instead of vertical scaling. Vertical scaling is when you increase the size of the one unit and horizontal scaling is when you keep the size of the one unit the same and you just get more units. It's distributed and load balanced. This is what it looks like. Let's say we've got $100 a day in ad spend, and we're making ROI, five to one ROI. That's why it's green, because we're making good ROI. Now let's say we change the budget on this, and we go to $1,000. Now we are not making ROI, and that's why it turns red. When we change the state of this thing from $100 to $1,000, it goes from being in order to being in chaos, and it goes from being in ROI to not being in ROI. We face this issue when we change the state of this thing by too much. However, if we just create ten individual instances of this thing, we get to leave it in ROI, in its good zone, and we just get ten times more of it. That is what horizontal scaling is. This way, we keep ten individual systems in order instead of having one system in chaos. Here's how I can further illustrate how we do this. Vertical scaling would be like a centralized method, and this is where you find a winning ad set and then increasing the budget of that ad set resulting in entropy, which is disorder. We know this too, how I talk about this in the accelerator program and I'm pretty sure I talk about in up level too, is we're trying to build decentralized consulting businesses where we're not the only source of information, and there's a community, there's Q&A calls, there's all of these different things. We're not the only source. We're also trying to set up our ad accounts this way too. We don't want one thing to have that much load on it, because if that one thing plays up, then everything falls down. Instead we want to decentralize it and distribute it. That's what horizontal scaling is. This is where we find a winning ad set, and then we duplicate it, for example, ten times to achieve scale without entropy. Now you might be thinking, "Well this is an awesome theory Sam, and it's a good concept, but how do I actually apply this theory in practice?" That's a good question. I hate theory without it being proven in practice. Now let's apply it, and you might be thinking, "Well how do we actually distribute that spend? How do we load balance it?" The thing is with Facebook, well the thing about any system, is there's so many dimensions in which you can apply the load balancing. For example, let's say you had ten ad accounts and you just spent $100 in each ad account, and each ad account had one campaign and one ad set. I mean that could be one way, you could distribute the load over ad accounts. Let's say you had one ad account and you just created ten campaigns. Well then you distribute the load over ten campaigns and one ad account. Let's say we had one ad account and one campaign but ten ad sets, and we distributed the load over ad sets. You get my point here? I'm talking about along what dimensions do we apply the scale because there's a lot of different things we can play with. There's different campaigns, there's campaign types, there's ad angles, there's audiences, there's ads, there's placements, there's budgets, there's optimization strategies. There's so many different things we can play with. Different dimensions going all over the place. Now lucky for you, we've played with pretty much all of them. We have tested everything, even things that seem totally crazy and totally whack because we just want to think it and experiment and play with everything to find out how this machine works. We have a pretty good idea. Now how we apply the horizontal scaling. Horizontal scaling is when we decentralize and distribute spend across multiple dimensions to hold ROI at scale. Now what are dimensions? Well when I say dimensions with Facebook, I'm referring to ad accounts, fan pages, campaigns, ad sets, ads, audiences, placements, campaign types, conversion objectives, budgets, bid strategies, countries, demographics. Basically every option in every layer, level and feature that Facebook gives provides another dimension of how we can use and interact with a system. Now lucky for you, I'm not just letting you go and just say, "Go load balance it somehow," because that would be very hard to figure out. You wouldn't know where to start. Instead what I'm going to give you is eight proven methods for scaling out horizontally. Our strategy is to scale horizontally across eight dimensions using the following methods. This is eight dimensions in which we can play with. Now the first one is to duplicate ad sets and max out the budgets. Another dimension is to just look for new audience interests and grow that way. Another one is to create lookalike audiences. Another one is to create new ad variations, so not completely new ad angles, but variations of the angles. Then another one is to create completely new ad angles. Then when we get new ad angles, we can create variations of those new angles. Then we've got a new angle and new variations, we can send those out to lookalikes, new audience interests, and then we can duplicate those and max out the budgets. Each one of these can interact with the other ones. These are eight dimensions in which we can play, but every one of these can be applied in combination or with any of the others. This gives us like a Swiss army knife of tools and possibilities here. It gives us really the opportunity to ... it really means that sky's the limit. I've been able to take our ad account to about $40,000 a day, but that's by no means the limit. I think it can go much higher. Another one is to target additional countries. You can start scaling out horizontally that way. Another one is to scale your retargeting campaigns. Not many people talk about this. A lot of talk happens in cold traffic, and don't get me wrong, most of the heavy lifting in your ad account is going to happen in your cold traffic campaign, but this doesn't mean that you can't scale retargeting. We've figured out ways to scale retargeting. Then the final one, method eight, is decentralized architecture. This is some ninja level stuff, which we will get to at the end of this module. Let's cover method one, duplicating ad sets at max budgets. What is it? We identify winning ad sets, that is ad sets that are within KPI and we duplicate them and set the budget at five to 15 times cost per lead. Now why do we do it? Well changing budgets on existing ad sets is a bad idea as it throws the algorithm out of balance. Instead, we duplicate the existing ad set to create a new one, and we use this opportunity to increase the budget. We're duplicating it so we can have a fresh start with a higher budget instead of interfering with the existing one and changing its budget, which will probably make it stop working, and it won't give us a clean test. Now why five to 15 times cost per lead? Why this number? How did I arrive at this magical zone? Well ad sets perform well at a small scale, and they start underperforming at a larger scale, as I've shown you with our continuum of low scale high ROI, high scale low ROI. Now the question is what's the optimal scale point for an ad set before it starts facing entropy, which is disorder. The answer is five to 15 times cost per lead. Here's why. Here's our scale vs return continuum. Like I said, at low spend we get high ROI, high spend we get low ROI. When we're getting high ROI, the system is in order. When we get low, it's facing entropy. It's in chaos. This is the thing we're constantly balancing. The question is at what point along this continuum is it optimal? What's the most we can get to before we face entropy? We've tested it. Like I said, we've been advertising on Facebook for seven years, millions of dollars, multiple millions in spend, and we've been able to spend up to $40,000 a day, been making more than $100,000 a day. We've made more than $10 million from Facebook and we've tested a lot of stuff. This is what we've found. We've found that an ad set, what dictates an ad, it's optimal level for its budget is based not on a dollar figure, but on a multiple of the KPI. Now what I mean by this, let me show you. I'll go to my ad account, and we use this demo one here. Let's say when we created our cold traffic campaign, this cold traffic campaign, we are optimizing for conversions, which is leads. We're optimizing for cost per lead, which is for the VSL, cost per VSL opt-in, or the JIT, cost per JIT registration, or the 2KA, cost per 2KA registration. Now the ad sets, if I go here and I look at how they're being optimized, these things are optimizing for lead conversions. Then if I look here at performance, we can see that they are optimizing for performance here, which is an opt-in. This is what it's optimizing for. It's learning how to get these as efficiently as possible. That's why when you create a new ad set, it'll say it's in its learning phase and it needs to get so many before it's completed its learning phase and all of this. What it really is is it's a multiple of this conversion here, which really dictates the optimal level of efficiency for an ad set. Let me make this clear. Let's say you're getting leads for $10. You're getting VSL opt-ins for $10. If your budget was set at $10, then that's one times KPI because the KPI is the cost per result that the ad set and the campaign is optimizing for. In this instance, it is the lead, and we're getting the leads for 10 and our budget is 10. That means that our budget is one times KPI. Make sense? Now what we've found is that one times KPI is the absolute limit, the bottom level limit that an ad set can be set at in order to do anything. If you set your budgets at less than one times KPI, your ad sets are set up to fail. If you're getting leads at $15 each on average, say you're getting JIT registrations or VSL registrations for $15 on average, but you've set your ad set budget at $5, you're screwed because listen to what you're telling this algorithm to do. You're saying, "Hey Mr. Algorithm, I typically pay $15 for these things. That's what my history shows, and that's what I'm actually happy to pay for it. However, I'm tasking you to go on a mission and find these things for me for $5 each. In fact, I want multiple of them for $5 each even though I know that it's quite normal for me to pay $15." That is that you're telling the machine to do. The machine just breaks down because it can't do that, so it won't work. It's wishful thinking if you think it will. This is why if you have your ad set budget set at least the one times KPI, they will not work. Now the minimum budget you can set your ad sets at to have them perform is one times KPI. You should be getting leads for in between $10, $15, somewhere around them. That's why in the training, I told you that the minimum amount you can go for your ad sets is about $10 and that's only during the testing phase. That is, I had to make a trade-off there in the training because people are restrained by their budgets, so in an ideal world I would test with $50 per ad set per day. I can't tell people to do that when they're restrained by budget and they're also trying to test different audiences and different angles. That's why I recommended $10. Really, if you're not restrained by budget, you should always be testing with at least $50 per day per ad set. It just gives the ad set more breathing room, because if you set your budget for an ad set at one times KPI, let's say you're getting leads for $10 and you set the ad set budget at $10. That ad set can only really get one conversion a day because you've set its budget at $10, it costs at $10 to get one. It's probably only going to get one a day. Some days it might get zero. Now it can't perform very well when it only gets to fire its event once a day. It doesn't get very excited, it doesn't pick up momentum and it doesn't start racing out and finding new conversions for you because it's kind of sitting in this zone where it can't do much. Now two times KPI things get a bit better. Now to make things clear, it can function here. It can function. It cannot function here. It can here. It's actually most ... I would personally never really go below this. If you were restrained by budget, but I personally would never go below this. Now at two times KPI, that ad set is going to be able to get two conversions per day on average. That's twice as good as one, so it's going to probably learn and reach and perform way better than one. At five times KPI, so if we're getting leads at $10, we set the budget at $50, that's five times KPI, that means that ad set can get around five conversions a day. That's learning and performing 500% better than this, and so it's going to work way better. It comes into its own at this level. Then ten times KPI, you're getting leads at $10, you set your budget at $100 for your ad set. That means the ad set is going to be getting ten conversions a day. Now things start happening. The optimal level I found is around 15 times KPI. You're getting leads for $10, you set your ad set budget at $150, it gets 15 leads a day and that's really where it likes to be. Here I say ad sets experience entropy at budgets less than two times KPI, so you don't want to go really below two times KPI, and optimal efficiency exists at 10 to 15 times KPI. That's where the sweet spot is, 10 to 15 times KPI. As soon as you start going above 15, it's not really 15. I mean you can go all the way really up to 18, 19. It starts to get dicey again at about 20 times KPI, so it's a spectrum. It's bad down the bottom, it gets bad up at the top. The sweet spot is right here, 10 to 15 times KPI. This helps you immensely knowing how this works and knowing the parameters of the machine. Now when we test our ad sets initially, when we're creating our five angles with four images each and we're trying to go out to 30 audiences, we've got 600 possible combinations, we're trying to find the best ones, what wins, what loses. We're doing that with a $10 per ad set per day budget, which is one times KPI, which is not optimal, but it's okay for testing. Now once we identify the winners and we kill all the losers, that's when we want to scale our winning ad sets out of this zone and into this zone. This is where we scale our winning ad sets to the maximum possible budget, which is roughly 15 times KPI because if we're just leaving them down here, we're basically leaving them to struggle even though we know they're winners, when we could unleash them up here. That's why our first scaling method is duplicating ad sets at max budgets. How we do it is we find your best performing ad set, and this is the most within KPI ad sets. Winning ad sets are ad sets within KPI over a four day time frame. The best performing ad set is the best one. Over a four day time frame, which one has been getting the most leads at the best price? It's also taking into consideration deeper funnel conversions too. If you've got some ad sets that have got less leads at a higher price, but they've got strategies ... ... lists leads at a higher price, but they've got strategy sessions or customers. That one beats one that has more leads at a lower price but no strategy sessions or customers. You get how it works. A winner isn't always done on cost-per-lead or number of leads. It's done on the deepest funnel metric in its performance there. In the absence of that, we use leads. Now when we identify our best performing ad set, we want to duplicate it. We want to keep everything the same, but we want to change the budget to 5 to 15 times cost-per-lead. An example: if the ad set is getting leads at $10 and its budget's set $10, we just duplicate that ad set, leave everything the same, and set the budget anywhere between $50 and $150 a day. Let me show you how this works. Let's say I identify this one as the winning ad set. All I would do is go duplicate, I'd put it in the original campaign. I go to duplicate. I leave everything the same, but I just change the budget. I want to go in between 5 and 15 times KPI. Where in there you go is up to you and how much you have available to spend on your budget. Let's say you had to cull a lot of losing ad sets and you're wanting to spend more per day and you're trying to find ways to spend more. Then I would spend 10 to 15 times KPI. If you were restrained on budget a bit, you'd probably go to the lower end, 5 times KPI or something like that. That's how it works. There's a catch to this, too. This is what I mean by parameters. On cold traffic, we shouldn't exceed $10 per 10,000 people in a given audience. If you split audience interests big enough to handle 5 to 15 times cost per lead, do it. However, if your split audience interests aren't big enough, pull the winning audiences together into one new audience and use that. Here's what I mean by this. When we created our ad sets, we split them by different audience interest. This one here is targeting frankincense. This one here is targeting Tim Ferriss. This is what I mean by split. We split out all of our audience interests to test them. We're not just unlimited in how much we can increase the budget here. We have to only increase the budget so that we're not breaking the $10 per 10,000 rule. If my audience is a million people big ... Tim Ferriss' audience is big. That'll handle it. Then, I'm fine. I can spend 150 bucks easy, because 150 out of a million is not ... $1,000 is the max spend you can spend with a million. 150 isn't near 1,000, so we're good. However, if your audience is tiny, then you can only increase your budget for that audience to the $10 per 10,000 limit. If it's only 20,000 people, your audience, then you can only really spend $20. That's what restrains your move here. Hopefully all your audiences are good enough to handle at least 50 bucks or, better yet, 100 or 150. Then you're good. You can scale these up. In the case that you've got a bunch of audiences that are proven winners but they're all small, what we want to do here is, instead of targeting them all in little clusters, we want to group them together. To do that, inside Qwaya when you're setting up your audience templates ... I showed you how to do this in the Going Live module in this week's Facebook ads training. When you create audience templates in Qwaya and what we do there is when we launch our audiences into Qwaya, we select an option that says Split. That splits them out into separate ad sets. What we want to do is we want to identify our winning audiences. We want to find all the audiences that work and have been proven to work, and we remove all the audiences that didn't work. Then, when we launch those audiences into Qwaya, we simply select Pull and we don't select Split. I think we actually, by not selecting Split, we're pulling them. I don't think there's actually a Pull button or Pull checkbox. We just don't check the Split box, and by doing that, they pull. That way, we're putting all of those audience interests into one ad set. That means if there's 10 audience interests at 20,000 each, then that's going to pull them together to make one audience of 200,000. But there's probably going to be overlap between those different audience interests, so it's probably going to be more like 150,000. When we have one audience that's 150,000, one ad set with an audience in it that has 150,000 people, we can now spend about $150 to that instead of having to do that across all of these little ad sets. An ad set, remember, is at its optimal at 10 to 15 times KPI. If an audience interest within an ad set only allows you to spend $20, then you've kind of set that ad set up for failure. It is better to pull those audience interests together so that you can get up to the optimal spend range to make that ad set really hum. This is the trick. Sometimes you pull things, sometimes you split them. If you can't spend enough to get it into the optimal zone, pull so that you can. If your audience is so massive that you really are able to spend a ton but you can't, you should be splitting them out so that you can. It's all about architecture. It's all about distributing and load balancing every single thing in your ad account so that it's sitting at the optimal zone. That's the trick to it. This is some next level stuff. They honestly do not teach you this anywhere in any Facebook ads training anywhere on the Internet, because no one spends this much money and has this level of understanding of it. That is what we do here. Now let's talk about our second method, which is new audience interests. What it is: we search for audience interests with affinity to Babel or winning audiences. We put them in and we look at audiences that have affinity to that. Then we test them to find more winning audiences and expand the audience size of our campaign, therefore increasing threshold spend. Why do we do it? In order to spend more, we need to increase the size of our audience because we can't exceed $10 spend per 10,000 people in a cold traffic campaign. It says 1,000 here, so I'm going to quickly change that, because it is 10,000. Don't want to cause any confusion here. We want to scale, but this thing can really constrain us sometimes, because we can't break this rule. If we do break this rule, then we experience entropy anyway, so there's no point. Our audience size dictates our threshold spend limit, plain and simple. In order to be able to scale a lot of the time, we have to be able to expand our audience so that it can take that additional spend. It's not just about spending more; it's getting an audience that's able to handle that spend and then spending more. That's why this is really a pre-requisite. Increasing the size of your audience alone will not scale your campaign. However, it provides you with the ability to further scale your campaign. That's why this strategy is used as a pre-requisite to scale. How do we do it? We use Facebook's audience insights tool to find new audience interests to affinity to Babel or any winning audience. When we find them, we add them to our spreadsheet for testing. Our spreadsheet is our Facebook audience angles and images spreadsheet where we put down our ideas for audiences and then what we've tested and which ones are proven and which ones don't work. We will try to look for new audience interests to test. We've probably tested our initial 30 during our initial launch where we launched Facebook's [inaudible 00:41:33] with five angles with four images each to 30 different audience interests. If you've already done that and out of those 30, you've found five that work, then we want to test another 30 to find another five. Then another 30 to find another five. We want to test enough groups of 30 until we find 30 that work. When we have 30 that work, we have a big proven audience size, which can handle a lot of spend, which gives us the ability to spend more and then we can spend more. Scaling is always a balance of trying to widen your audience and then spending more and then trying to get things to perform at their higher spend. Let me show you do this real quick. I've shown you how to do this in previous modules, so I'm not going to go really step-by-step on this one, because I've already covered it. If you can't remember how we did it, then you've got bigger issues than me showing you how to scale. You should go back and learn it. I'll go to the audience insights tool. Here, we want to put in our Tower of Babel audience. Let's say that is Tony Robbins. Then I can go to page likes and I can look at all the page likes that have affinity to Tony Robbins. You can see down here it shows all of these. The ones that are closest to him are the ones that most match him. You just want to keep going down through these and finding different ones to test. If you've tested a bunch of them from here, then test more of them from here. You're not only restrained to having to derive audiences from your Tower of Babel. Once you've found winning audiences, you can use them to derive audiences, too. Let's say I started using Tim Ferriss and I found that he was working. I know his audience was working, because it was getting leads in strategy sessions and things. Now instead of deriving from Tony, I derive from Tim. Now I can start trying these. If some of these work ... let's say now that Daily Stoic works. Daily Stoic or [inaudible 00:44:19], however the hell you say that. Then we plug this in here. No, it doesn't want that. Let's say we try Samsung. Then we find audiences from here. There's no end to how many things you can test here, because you can find tons of audiences that are just derived from your Babel. You're going to have some that work and then you can derive tons from those. From those, some are going to work and you can derive tons from those. There's no excuses here. You can identity a ton of audiences. When you find, then, the ones that you want to test, you add them to your Facebook audience angles and images spreadsheet, which I gave you in one of the previous modules, and then we're ready to test it. Then what we do is we create a new audience targeting template in Qwaya and we put in all of those new audiences. We launch them. We select Split so that it splits them into different ad sets and we push them live. The campaign we want to push them live into is our sandbox campaign, not our production cold traffic campaign. We want to push it into sandbox, because we're sandboxing different audiences here. Let's say we find 30 different audience interests, we split them, we launch them with our proven angles. We take our angles that are proven to work from our production campaign and we only launch those proven angles to those new 30 audiences through Qwaya in our sandbox. Let's say we've got two proven angles. Then we are creating two angles with four images each, going to 30 audience interests. That means that we are going to have 60 ad sets and we want to be testing those at at least $10 each. Sometimes you might not be able to spend that much. If you can't, then lower it. Only use one angle, your best one, and test as many audiences as you can. Work through it that way. When you find proven winners where you're getting leads within KPI in your sandbox campaign, then you want to duplicate them but add them to your production campaign. If I come back here and I go into my ad account and let's say I'm in my sandbox and I find a winner, I want to duplicate it. Where I select the campaign ... I'll show you what this looks like in here. Let's say I select one of these and then I go Duplicate. It's going to ask me what campaign and I want to select Existing Campaign. Then I want to select this production one. I'm going to be duplicating out of sandbox and putting it into my cold traffic campaign, because it's already proven. When I launch it into my proven production campaign, I want to set the budget at 5 to 10 times KPI, because it's already been tested, it's already been vetted. Now it's time to take it to its max threshold. That's our workflow. We find audience interests from the audience interests tool, finding things with affinity to Babel or any audience that's been proven to work. Then we add them to the spreadsheet. We launch them in Qwaya using the split method into our sandbox campaign. We wait four days. We find the winners. We cull the losers. We get the winners, and we duplicate them, but we change it from the sandbox campaign to our production campaign. Then we increase the budgets to the 5 to 15 times max threshold KPI. This is a really, really powerful workflow. Now let's talk about another one. Method three: creating lookalike audiences. This one's a biggie. What it is: when we have enough conversions, we create lookalike audiences to expand our audience size. Why we do it. Because we can't exceed $10 per 10,000 people in an audience ... damn this thing. It keeps saying 1,000. Our audience size dictates our threshold spend. By creating lookalike audiences, we rapidly expand our audience size and therefore our threshold spend. Why lookalikes? Why not just keep going with audience interests. Well, we start with one Tower of Babel audience and then we derive multiple audience interests from Babel. Then we derive multiple audience interests from our winning audiences. We're starting with one. We're deriving things from that one. When we find new winners, we're deriving things from those new winners. Then once we've been doing that for a while and we've got enough data in Facebook and we've got enough data in there for it do its magic, we then use Facebook's algorithm to derive programmatic audiences. What programmatic audiences are is that we're allowing Facebook's algorithm to find people who look like the people we want. They're not necessarily people who like Tony Robbins or people who like Tim Ferriss. They're people who might like any one of those things, but that Facebook's algorithm knows look like the people who we want. It's actually more powerful than selecting specific audiences. It's probably the most powerful tool in Facebook Ads, but you can't use it at the start. You have to start with audience interests. That's why we start with audience interests and we keep deriving, we keep creating derivatives. When we get the opportunity, we can start using lookalikes. How do we do it? Once you've had a 100 opt-ins, that's 100 people register for your VSL or your GIT or your [two KA 00:51:18]. Once you've had that, then you can create a lookalike on ad post engagements. Once you've had 300 opt-ins, you can create a lookalike on the opt-ins pixel. Once you've had 300 strategy session applications, you can create a lookalike based on strategy session applications. You start at the top of your funnel, which is ad engagements, and you create lookalikes at each stage as soon as it has sufficient conversions, which is 300 plus needed. Let me show you how to do this. Once you've had 100 opt-ins, you can create your first lookalike. Before you have 100 opt-ins, you can't create any lookalikes, so don't even try it. Once you have had 100 opt-ins, you'll know because if you go back to your campaign level and you look at your cold traffic campaign here and if you can see if you set it to lifetime, then it should say here in Results ... opt-in for VSL [inaudible 00:52:28], there should be the number 100. If this number is not 100, then you can't do it. Once it is, you can. Then, how do we create a lookalike? We want to go up here. We go to Audiences and then we want to click on Create Audience, Lookalike Audience. We want to set it at 1% and then sources, we want to go our fan page. What we want to do is we want to go to Location. You can set the locations in here. I'm pretty sure they've changed this. Hold on, let me try that again. Create Audience, Custom Audience, sorry. What you're doing is you're going Create Audience. Then you're going Custom Audience and then what you're doing is you're going Engagements. Then what you're doing is you're selecting Facebook Page and then you're selecting anyone that has interacted with your Facebook page, the one that you use for your ads, and anyone who has engaged with any poster ad. Here we can set just in the past 180 days. Then we can create this audience. You can create that once you've had about 100 opt-ins. Then what you can do is you can go into this audience and you can create a lookalike based off it. Here I can see ad engagements, 180d. Up here under Actions, I can click Create Lookalike. Here, I want to set the audience size at 1%. I can set the locations, whatever. Then I'd create the audience. What we're doing here is we're creating, first of all, the audience for people who engaged with ... that means they liked, commented, or shared or clicked the link in any of our ads in the past 180 days, or organic posts. We can create an audience of that and re-target them. What we can also do is we can create a lookalike off this and then use it for cold traffic. We use this ad engagements for re-targeting because we're re-targeting because they've already engaged with us. A lookalike is people who look like the people who engaged with our ads. That isn't re-targeting. In order to re-target somebody, we have to re-target them. That "re" means that they at one stage engaged. These people look like people who engaged. Therefore, they did not engage themselves. Therefore, it is not re-targeting. It is cold traffic. It has affinity to re-targeting, which makes it good. We first of all create the ad engagements audience after we've had 100 conversions. Then we can create a lookalike at 1%. Then we can run ads to that in our cold traffic campaign. When we do this, we're not creating a new campaign for lookalikes. We're just putting all of this into our production campaign. A lookalike audience can a lot of the time skip production, can skip sandboxing, because it's a strong audience. Lookalikes, I'd be happy to put those into production, because we're using a proven angle and we know the previous audience worked. We're just creating a derivative of that proven audience that worked with an angle that we know is proven to work. Its chances of working are very high. I would skip the sandbox with it and just push it straight into production. I would start off with a budget of at least two times KPI. Really, I'd want to be in the five times KPI. When it's proven work, you can dupe it and go up to 5 to 15 times KPI to bring it to its threshold budget. That's how it's done. Once you've had 300 opt-ins, you can create a lookalike on opt-ins. How we do this is we're basically creating an audience first and foremost of people who opted in. This would be visited VSL value video or registered for GIT webinar or registered for 2K webinar. That means they registered. They saw the page after the opt-in form. This audience is everybody that basically opted in or registered. When this audience size is 300 or more, it has to say here, "Size 300 or more." Then what we can do is we can click here, we can open it, and we can go Actions. We can create a lookalike. We want to go just 1% and put our countries in, launch it. Only select the countries that the original audience had countries selected for. We're just keeping all variables consistent. We're just creating a derivative of the original variables. Don't mix things up and cause chaos by just putting weird things in here. Keep it the same. You can see our method here. We create an audience here to measure everyone that does an event. We then re-target them in our warm re-targeting campaign, because these people have done something, but we don't stop there. We create a lookalike based on those people who did this. That isn't retargeting, because those people just lookalike. They didn't do it. Then we put those people that lookalike into cold. We target them there. We can start at the top of our funnel, which is just people who engage with our ads. This one's going to get the highest numbers the quickest, because more people are going to do this than anything else. This is the one to start with, 'cause it's going to get the most amount of data the fastest. Then once we've got at least 300 here on the visited this, we can create a lookalike off this. Now we can go into that one, launch it into our production campaign as per usual with a proven angle, and then bring it up to threshold spend, 5 to 15 times KPI. What's the next link down the chain? The next link is strategy session applications. Once you've had 300 of those ... you'll know when you look at your audience here. That is scheduled strategy sessions. You can set scheduled strategy sessions and then you can put 30d if you can get it there or you can change it to 180d. If you can get an audience that is scheduled strategy sessions and you create that audience by going from the people who visited the success page after completing the survey, then you can create an audience here and set it at 180d and you can wait until this size gets to 300. That means you're going to have to have had 300 people complete the survey application form within a six-month period. If you achieve that, it'll say 300 here. Then what you can do is you can just click here and then you can create a lookalike. You want to set 1% and leave the countries constant as well and launch that into production with your proven angle. Scale it up to max threshold KPI, 5 to 15 times. Then, you can keep working through. If you're doing the 2K funnel, you can start creating them on people who visited the order form, people who visited the sales page, or people who actually purchased- People who visited the sales page, or people who actually purchased. Once you've had 300 people who actually purchase, now you can create a very powerful lookalike audience, because it's people who look like buyers, and that's powerful. But 300 buyers is a lot, so you need to work your way down the chain. Start at the top and work your way through. Don't create lookalikes on things until they've got enough data on them, because you're just shooting yourself in the foot. You know it's ready to create a lookalike on it when it's had 300 people do the action and you can see that in the audience's tool and the size. Now the parameters. When creating lookalikes for the first time, use 1% audiences. We want to keep them as tight and as close as possible, so use 1%. Keep your country or countries to what's proven, and what's proven is what you were using for the original audience. If you're only targeting America, when you're creating a lookalike, only target America. But if you're targeting America, Canada, Australia, and New Zealand, when you create a lookalike, create it for America, Canda, Australia, and New Zealand, not one for each one, a lookalike for all of those together, all right? That's what you do. Just keep it the same. Keep it as it was when you were running it cold. And once you've tested and proven lookalikes at 1%, later on you can try 3%, and if that works, then you can try it even higher. If that works, you can just keep going, and you can go up to 10%. But remember we always want to try to keep things simple. So, if you're going to create another lookalike at a higher percentage point, here's what you do, because people can make a mess of this. Let's say I want to create a lookalike off ad engagements at first, and when I do that, I go create lookalike, and I set 1% here. Now let's say this one works really well, so I want to try and create a bigger one. What I don't do now is set one at 2% and create one there, because what I'm going to do is I'm going to have more people in here compared to the 1, but I'm also going to have all of the people that the 1% one has in it, and I'm still running the 1, and now I'm running the 2%, which includes the 1%, and now I have overlap. And now I'm targeting the same people and I'm making a mess, so you click show advanced options and then you want to, let's say I've already tested the one. Now let's say I want to test a 1 to 3, so I select this band. Here I'm leaving out all the people that are in the 1%, and I'm creating a 3% lookalike that excludes the 1%, so I'm creating it based on this band. This is what you want to do. This way you don't have overlap. This'll save you a lot of heartache. Now let's say the 1% works and the 3% works. Now let's say I want to test a 6. Well, I can just drag it 3 to 6. If that works, then I can create a 10 to 6, right? And all of these different lookalikes aren't going to overlap and include the same people in them, because I'm creating it based on bands. That's a little small hack that will save you a lot of heartache and trouble. Remember, always try to keep things simple, so don't get too crazy on your lookalikes. Remember, you don't need a ton of audiences, and you don't need a ton of different complexity to make things work. You just need enough people in your audience to spend the right amount of money, and you need things to perform well, and you want to keep things simple. Now let's talk about method number four, which is creating new ad variations. What it is. We create variations of our proven ads so that we can run multiple versions of them to the same audience without experiencing entropy. Ideally, we want multiple versions of an angle running at 5 to 15 times KPI or CPL. Now, why do we do it? Well, once we find a proven audience angle image combination, we want to duplicate it and set the budget at 5 to 15 times KPI so we can take it to its threshold limit where it's most optimal. Then if it works there, we want to duplicate multiple versions of that and set them at 5 to 15 times cost per lead, but we don't want it to be identical. We don't want the exact same ad running to the exact same audience, because it's going to have overlap. Instead what we do is we change something about it. We want to change a small thing, like the headline or the image or the button, so that it's unique enough to run separate from the existing ad set. Now what Facebook does here is Facebook's very complicated in how it creates auction pools. So, let's say you create an ad and then you select an audience and then you bid on it. Now if somebody else creates an ad that's similar and they are going to an audience that's similar and their bids are similar and budgets and things are similar, then they're thrown into that auction pool with you, and the person who performs or pays the most wins that auction pool. But we can enter auction pools with ourselves. If we create the exact same ad and run it to the exact same audience with the exact same everything, then we're pretty much going to just compete with ourselves a bit, and we want to try and avoid that. However, if we use different audiences with the same ad, we're not really competing with ourselves. If we create different angles with different audiences, we're not really competing with ourselves. What we can do is we can create variations so we create separate auction pools and don't compete with ourselves. I'll show you how to do this. Let's say I've got a proven ad set, and it's my winner in my cold traffic production campaign. Let's say I grab this and I go to duplicate it. Let's say I've already had this humming at 150 a day, right? Now I want to dupe it again, and I want to go for 150 again, but I want to make sure that I change the ad a bit. So I'd go duplicate and then I would go original campaign, duplicate. Now at the ad level here, I'm going to have to go down to ad. So we leave everything the same. Budget we're going to 5 to 15 times KPI, but at the ad level, I want to change something. At the ad level what I want to do is edit this. I just want to change something about the headline. I might use a different headline or I might turn on a button, like Apply Now, Book Now, whatever. Just change one thing about it. Change the headline or put a button on it. This is what you can do, or you can try a different image. Button, image, or headline. Those are the things you're playing with. You're not changing the angle itself. When you do this and then you review and publish it live, you're creating a micro-variation of an existing angle. We don't need to do this in the sandbox, because the audience has been proven with the angle, and we're just changing a slight variation of that. It's highly likely to work, so we can just do it straight into production at 5 to 15 times KPI. You might be thinking, why do these small variations? Why not just duplicate it? Well, like I've said, if we just duplicate it, then we are competing with ourselves, which we don't want, because we will create entropy, which we don't want. Now, you might be thinking, well, why don't we just change the budget a bit and also just change the age a bit? Doesn't really work anymore. Facebook's algorithms are constantly being updated, and they get better and smarter. That one doesn't really work anymore. Eventually, anyway, it would probably experience overlap, so it's not the best. It's best to actually just change, actually change. Not try to fool it change, but actually change, the ad. When we do that, we actually get put into a different auction pool and we actually have a better advantage. It works better. Do it that way. Now, you might be thinking, well, why not new angles? Why don't we just create totally new angles instead of doing these as variations? And you should always be sandboxing for new angles, and when you find winners, you should be scaling them into production. New angles are an extremely effective scaling method. They're actually one of the best. In fact, it's the best. If you create a new ad that is just a killer ad, it's the best scaling method you can have, but it's also very hard. The scaling strategies I'm sharing with you aren't necessarily about how to create ... If my scaling method was, just create a really good ad, it's not really ... People would be like, "Oh, are you serious? We knew that would scale." So, of course, that is the best strategy, but scaling, when I talk about that and when most people talk about it, they want methods to increase or to get more juice out of existing angles. That's why variations are hacks to get more juice out of an existing angle. If you've got an existing angle that works and you've taken it to threshold, now you can get more juice out of it by duping it and making a variation, and if that one works, you can dupe it again, create another variation, and get 5 to 15 times KPI out of it. You want to keep doing that until you hit the threshold spend limit of that audience. Let's say the audience is Tim Ferris and it's got a million people in it, and I keep duping these things at 150 and creating variations, $150. The threshold amount that I'm going to be able to spend on that across these different ad sets is still going to obey the $10 per 10,000 person rule, which if Tim Ferris's audience is a million, then I'm only going to be able to spend $1,000 divided by 150. We're going to end up with about seven, so I'm going to have seven variations of an angle at 150 each going to the same audience before I max that out. Get it? Now, how we do it? Once you've got a winning ad set running at 5 to 15 times KPI, or cost per lead, duplicate it again and change the headline like I showed you, image, or the button. You can do any one of these. I suggest just doing one, not all, because when you do all, you're almost creating a new angle and you might screw it up. Just a micro-variation. You want to keep everything else the same, including the budget at 5 to 15 times cost per lead. Try not to have one identical ad running to the same audience. If you've got one proven angle running to an audience, then only have that running to that audience. You can have that identical angle running to other audiences, that's fine. But if you're going to keep scaling that angle to a particular audience, first of all, you stretch it up to 5 to 15 times KPI, and then once you've done that, then you have to create variations to get more shots at it and scale it up. When you create these variants and get them to 5 to 15 times KPI too, you can keep creating variants and taking them to this 5 to 15 times KPI to the point you reach the audience threshold. Now let's talk about method five, which is creating new ad angles. First of all, sorry, it just says number four here, but this is method five. What it is. Experiment for new ad angles in our sandbox campaign, and when we find winners, we scale them into our production campaign at a 5 to 15 times cost per lead budget. And then once they work at that, then we create ad variations using the previous scaling method that we covered where we change the headline, the image, or the button. You can see here, this is an insane process, right? It's so powerful, because all of these things we can play with, we can combine them and switch them with each other. When we're sandboxing for new ad angles, the moment we find an ad angle that works in the sandbox, we dupe it into production. When we put it into production, we take it to 5 to 15 times KPI. Boom, once that's going there, then we create variations of that by changing ... With one dupe, we might change the headline and set it at 5 to 15 times KPI. If that one works, then I might dupe the original and change it and put a button on it, and then boom, if that one works at 5 to 15 times KPI, now I might dupe the original and change the image. Set it at 5 to 15 times KPI and boom, that works, and now I've been able to expand my spend for that specific audience up until the threshold limit, which is $10 per 10,000 at the limit. If there's a million in the audience, I'm spending up to $1,000 over those different ad sets. I'm achieving optimal efficiency. I'm reaching the max I can in that without hitting entropy zone. Now, why do we create new ones? Just because we have angles that work and that we can create new ad variations that work, just because we can do that doesn't mean we can't make one that works better. Never let good stop great. Even when you're great, don't let that stop you from being greater. You're never done, you're never finished. You can always do 10 times better than you're doing, no matter what level you're at. The moment in the day you think otherwise, someone will come along and take you out, so it never stops. And you should never stop. We're always looking for something better. Our methods are designed to extract more juice out of an existing angle, but creating a better one improves everything. Creating a killer ad can make you millions of dollars, right? Like I told you about advice for consultants, or 26-Year-Old Punk, and those ones have made me millions. Those are killer ads. What you're really aiming for is a 9 to 10 quality score. This is as simple as it gets. If you scale something shit, you're just going to get more shit, so if you're spending $100 and you're not making anything, if you scale it, well, then you're going to be spending $1,000 and not making anything, right? If you scale something, you just get more of what you had. If you didn't have anything, you're going to have more of nothing. If you had a loss, you're going to have a bigger loss. But if you have something that works, you're going to have something that works more. If you have something that's phenomenal, you're going to have something that is more phenomenal. Really, we want to make sure that ... To scale effectively, we want to already have something awesome. That's one of the real requirements of big scale. You need to have something awesome. You can't scale something that doesn't work, and you can't scale something that's average, either. When we scale things, we're testing things. If something's barely profitable, if we scale it, it's definitely not going to be profitable. You need to have wide margins on things to scale them. If things are barely working, then it's not going to be fun, so you need to get really good results at a small scale, so they have the safety tolerance to perform at a high scale. You need a 6 plus quality score to do anything. If you're not getting 6 plus average quality score on your cold traffic ads, you need to keep creating new ad angles until you get it higher than 6. You ain't going anywhere unless you have 6. If you're below 6, even if you're at 5, do something about it. It's not the audience, it's not your niche, it's not Facebook. It's just your ad. Your ad sucks. If you're getting less than 6, your ad sucks, and it sucks because you wrote it and you didn't do a good job of it, so you need to keep working on it. You need to practice. You need to test new things and you need to make it better, because it ain't going to perform well at all. Forget scaling, it's just not even going to perform before we scale if you're below 6. If you really want to scale properly, you need at least a 9 to 10 quality score. Every single one of our main ads that we have, like our killer ads that we scale out, they've got 10s. 10s across the board. It's rare if we have a 9. Now, that doesn't mean that we're so smart and everything, that we just create 10 out of 10 quality score ads. Most of the ads we create don't get that score. Most of the ads we create don't work. The way we've gotten a collection of 10 out of 10 quality score ads is by creating lots of them and testing and doing the hard work, so that's what you have to do. If you don't have a 9 to 10 quality score on cold traffic, keep sandboxing new angles. Keep trying new images until you get one, because you ain't going to be able to scale to the moon unless you're at 9 or 10 quality score on cold traffic. Now, how we do it? Once we've got a winning ad set running at 5 to 15 times cost per lead, duplicate it again and change the headline. This is actually not how we do it, sorry. It's actually just that. Creating new ad angles, it's as simple as what I showed you in the daily management, daily workflow module. We're just coming up with a new angle, which is new body copy, new headline, new images, and then we're testing it with four image variations to our audiences to see if it works, and we're doing this in the sandbox campaign all the time. We're just creating new angles. You know how to do that because you created the original five. You just are doing that process again. That's how you do it. When you find one that works in sandbox, duplicate it into production. Set the budget at 5 to 15 times KPI. Let it run. If it works, dupe it and create variations on that and get them to 5 to 15 times KPI, and then create new variations and then you can go back to sandbox, and then you can look for new angles. You see how this works? Now let's talk about method six, targeting additional countries. This is where things get interesting. Damn these numbers. So, what it is. We identify countries similar to the countries we're succeeding in and test them to widen our audience. Now, why do we do it? Because we can't exceed this damn rule that I've put wrong in almost all the slides. Apologies for that. Because we can't exceed $10 per 10,000 in an audience, our audience size dictates our threshold spend. Now by identifying additional countries, we're able to expand our audience and threshold spend. Now why countries? Once we've deployed multiple scaling methods within a given country, like let's say we've found lots of audience interests, and then let's say we've got three angles working and we've got a bunch of audience interests working, and then we've scaled to 5 to 15 times KPI with those angles in those audience interests, and then let's say we've even created some ad variations to get more reach into those audience interests. Then let's say we've also tried some lookalikes. Well, now we've really tried a lot, and we've really gotten our tentacles deep into this country. We've gotten into the bloodstream. Now, once we've gotten to this point, it gets harder to extract more scale from that country, and at this point, we should take our initial country to the absolute limits. You should never touch another country until you've taken the existing country to the limit, just like you shouldn't bother going to lookalikes until you've taken audience interests to the limits. You shouldn't worry about creating different angles until you've taken those angles to the limits. You shouldn't worry about duplicating ad sets until you've taken those to the limits. You want to take the initial thing to the absolute limit because you add in another thing, because you don't want lots of stuff. You want performance, and you want to make sure that the only time you add additional stuff is when you can't get more performance out of the stuff you've got, so the only option is to add something else on. Once you've taken that initial country to the limits, then seek new territory. How do we do it? Once you've scaled to the limit in your initial country, start testing countries most similar to it. This is just like everything you can see that we do here. It's just derivatives and affinity and all this. Once we find an audience that ... We start with our tower of Babel and then we look at audiences that have affinity to Babel, and then we go and test them. Then if those work, then we find audiences with affinity to those. Then if those work, then we find audiences with affinity to those. Then if an angle works, then we might create another angle that's similar to that, and then if a certain type of image works, then we might get another image that's similar to that. We're constantly looking at what works and we're finding something that's similar to that and we're trying that, too. This is what we do. A lookalike is letting the algorithm find something that's similar to what we've got and letting that go. Then what we're trying to do, too, is find countries that are similar, because this is a way to scale out. Countries share affinity with each other, by the way. It's fascinating. They share affinity just like everything else in the world, the universe. How we do it is once we've scaled to the limit, we start testing countries most similar to it, and you duplicate your best performing ad sets and you simply change the country to the new country for testing. We use the same campaign. So, if you've got, let's say you've got ads working ... Let's say you've got an ad set working and you've got an angle ... Let's say you've got an angle and a lookalike working really well. Let's say it's a 1% lookalike working really well with your best performing angle and it's in the United States. Then what you'd want to do is duplicate that ad set. Leave the angle the same, leave ... Except you're going to want to create another lookalike based on the same thing, but in another country. In a country with affinity to the one you've got that you started with. We do this in the same campaign. We'd probably just pull this off in production campaign. It doesn't really need to go to sandbox because it's got close affinity to something that's working. It has high propensity to work. Now, to think about what countries you want to target, you want to look for clues. Clues are everywhere if you know where to look. If you're targeting the US only, the United States only, but you seem to be getting the odd customer from Canada and the odd one from Australia, maybe from people just finding you organically or social or email broadcasts or a friend of a friend told someone else, if you're getting customers from other countries that you're not targeting with Facebook, that is a clue. If you have existing clients that are not in the US, that is a clue. This isn't just for the US. If you're in France and you're advertising in France and you're getting the odd customer in the US, even though you're not running ads into the US, that's a clue. These are what clues look like. That is where you can get ideas for countries. When I was targeting New Zealand back in the day, I noticed that the edge cases we'd get maybe Australia or something and I started doing that. Then I noticed some case edges we'd get the United States. Then I targeted that. Then I ended up crushing it in the US, so then I moved to the US, and then I started going out to Europe and all these other things, and now we're everywhere. This is how you do it, and I just was paying attention to the clues. Another powerful tool you can use that we've had a lot of success with is something called cultural clusters. Just as audiences have affinity, countries have affinity too. We use cultural clusters to see this. This is what cultural clusters look like. They're ways to cluster together different countries by their culture and their heuristics, biases, tendencies, and just the way they think and the way the believe and perceive the world. Over here, we've got the egalitarian group, which is like the Western world. The Western world likes empowerment and decentralization at a very high level. But then we can break it down even further, the Western world. We can go into the group that likes competition and contests a lot. Now that's United States, United Kingdom, Ireland, New Zealand, Australia, and Canada. I'm from New Zealand. It was very easy for me to work into Australia because it was just like the same, and I was shocked at how easy it was to just work in America, too. I always wondered why, but when I found this, I really understood why. Because although New Zealand's a tiny little country ages away from America, it has a lot in common with it because our culture is similar. It's because our cultures share competition. We like competition, we like autonomy, we like decentralization, we like risk taking, results, ambition, and innovation. That's like the American dream, the Andrew Carnegie, the Rockefeller story, the Elon Musk. All of those stories. That's like the American way, and that's also how these other countries believe, too. Chances are, if you've got ads that work in any one of these countries, they should work in all of these countries. If you've got ads that work in Australia, you should try targeting all of the other ones. If you've got ads that work in Canada, you should try targeting all of these other ones. They are basically the same damn thing. They all speak English, they all think the same, they all see the world the same way, and it all pretty much is the same. Then we've got our network- Then we've got our network, which is Sweden, Netherlands, Norway, Finland, this is getting into Europe. These are more westernized European countries. So, Sweden, Norway, and Denmark, and Germany and Switzerland, these are more westernized European countries compared to the other ones. What I mean by that is they're more like Americans than the other European countries. They like decentralization and empowerment. The main things is decentralization, risk taking, empowerment, and all of this. That's what these ones like. Then, we get into the hierarchical group, which they like centralization and hierarchy and rules. That's when we get into France and Belgium and Italy and Poland and Spain. These guys like hierarchy, rules, centralization, formalism and all of the stuff, and then we also get into all of these countries over here, too. Wherever you're running ads that are successful, chances are, you can run the same ads and make them successful in everything else within this cultural cluster. It's easy to scale to all the countries within a cultural cluster. It gets hard to scale to countries not within the cultural cluster you're already in, because the same ad angle won't necessarily work, because their belief systems are different. Try running a capitalist, look at me and how much money I make and how successful I am ad into France. Watch what happens. Or run it into Thailand or something, just watch what happens. The cultures change between these different clusters, but within these clusters, they pretty much remain the same. It's easy to scale within in, it's harder to scale into a new one. Quite often to scale into a new one requires new angles and a bit of work. I recommend scaling within the one you're already in first, take over all of this. Once you've taken all of that over, if you're hungry for more, have a crack at going into these other ones. Go to the one most like the one you're in. So if you're in this one, then you want to take over everything in here. Once you've done that, then you want to go to this one. You shouldn't start here, take over all of this and then decide to come over here and try that. This is far away from that. Start here, go through that, then go to here, work through that, then go to here, work through here, then go to here, work through that, then go to here, work through that, then go to here, work through that. Got it? Good. Method seven, scaling retargeting campaigns. What it is. We use different campaign objectives to increase reach and performance of our retargeting campaigns. Why do we do it? Most of our scaling efforts go towards cold traffic campaigns, because this is the entry point and the most important thing. However, as we scale cold traffic, naturally, our retargeting audience size increases, and when it does this, it gives us more options. Now, how do we do it? How do we scale retargeting campaigns? Once you've taken your warm campaign retargeting campaign to the absolute limit, duplicate the best performing ad set into a new warm retargeting campaign with the "page post engagements" objective. Now, first and foremost, never create something new until you've taken the existing to the limit. What we want to do, first of all, is we should have our warm retargeting campaign. Here it is. And we should be bidding. We should be optimizing for conversions using the auto-bid method, and in here, we should be targeting page post engagements, 180 days, all website visitors, 180 days, we should be targeting all of these people. People who have engaged with us within the last 180 days, and we're optimizing for conversions, and we're using auto-bid, and the angles we're using in here are the best performing angles from our cold traffic campaign. Now, what we want to do is we want to increase the budget of these as much as possible. We want to increase the budget of these as much as possible and take them to their absolute limits before we go and create something new. If our retargeting audience size is, let's say it's 10,000. Well, we don't need to use the same $10 per 10,000 people rule for retargeting. We can go about five times that. If we've got a retargeting audience warm of 10,000 people, we can spend probably $50. We take that up to its limit and it's at $50. If it's still working there, good. That means it's pretty much at its limit. We're spending at the threshold of the audience size. Once that is true, once you've done that, and you have to do that first before you do this other thing, otherwise you're just doing stupid stuff. Then we want to scale further by creating a new campaign with a different objective. What we do here is we start by going to campaigns, and then we want to go, create, and then we want to call this one warm retargeting PPE, which is like page post engagements, auto. We want to go auction, and our campaign objective is going to be post engagement and then we're going to create an ad set. We're not going to create an ad set, we're just going to go skip, skip, save to draft. Then we're going to publish that up, boom. Now we've got two warm retargeting campaigns. One is warm retargeting conversions auto, the other is warm retargeting page post engagements auto. Then we come in here, we find our winning ad set, and then we go duplicate, and then we go to an existing campaign and we select our warm retargeting PPE campaign this time, and then we dupe it into there. Then, what we do is we go down and we want to make sure our audience is set correctly. It should just be the same, which is people who have engaged with your fan page or people who have engaged with one of your ads in the past 180 days or visited your website in the past 180 days or visited your landing page in the past 180 days. You're excluding customers. Your location's going to be the same, same with age and things. This isn't going to have an audience interest because it's a retargeting campaign. And then placements. You basically want to go all placements. You want to just try all placements for this, and then what you want to do is all mobile devices, and then optimization for air delivery. You just want to go post engagement, and then that's it. Then you can just publish it up. Boom. But what about budget? How do we set the budget? So, here what we do is we set our budget at 40% of what the conversion campaign budget is at. So if I have 10,000 people in my warm retargeting audience, in my warm retargeting campaign, which is optimized for conversions, that's going to be spending $50 per 10,000, because that's the threshold limit. The rules change a bit when we get into warmer traffic. If I'm spending $50 per day into this campaign, then I want to find 40% of 50 and that's going to be like, what's that? $20 or something, so I'm going to set this at like $20. That's what you want to do. You want to take two trajectories through the warm retargeting audience. The first, and it always comes first, is the warm retargeting conversions auto using your two best angles from your cold traffic campaign to the 180 day audience using a $50 per 10,000 people in the audience budget to hit that maximum point of efficiency before entropy sets in. Once we achieve that, we can't go higher on those budgets because we'll get entropy, so there's another way, and it's a new campaign that's for page post engagements using the same angle to the same audience at 40% budget of the conversions. Don't worry about how hard all of that was to figure out. You just get it told to you. Easy as that. That's how you do it. Now, hot retargeting. The above strategy, the strategy we've covered here, is only used for warm retargeting campaigns, not hot retargeting. For hot retargeting, we stick to our conversions objective at the campaign level and daily unique reach at the ad set level with max bid of $150 per 1,000 CPMs. To scale hot retargeting, keep increasing the spend and always taking it to its limits. As the audience size increases, increase spend and you can try testing new images and new angles, but it's quite hard to beat a simple angle here. Images can help sometimes, but really that, too, this isn't one where you have to test a bunch of angles and images. Really what you're doing is you've just got to keep increasing that spend until you find its limit where it hits entropy. That's where you keep the spend, but as you scale your cold traffic and your warm retargeting, you're naturally increasing the audience size of your hot retargeting. As that increases, gradually you can keep increasing the spend here. That's really all you need to do to increase and scale hot retargeting, is as cold traffic increases, warm traffic increases. When warm traffic increases, increase the bid there. When all of this happens, hot increases, when hot increases, increase the bid there. And you just keep working through these three layers. The one that starts the whole domino effect is cold, so cold's where most of your focus goes. Once you've had a bump there, you go to warm, bump it there, go to hot, bump it there, come back to cold. Just like that. You can also try increasing max bid, too. You can try going $300 per 1,000 CPMs. You're just trying to go as hard as you possibly can with hot retargeting. Now, you're not going to go harder in hot retargeting by creating more hot retargeting campaigns. Bad idea. You're not going to go harder in hot retargeting by creating lots of ad angles. Bad idea. Lots of duplications, bad idea. And trying to create granular audiences, bad idea. Trying to create a page post engagement for hot retargeting, bad idea. Trying to create a conversion objective campaign and ad set for hot retargeting, bad idea. Just stick to this thing. It is the best. You've just got to keep increasing spend. Now, let's talk about method eight, decentralized architecture. What it is. We structure our ad accounts, campaigns, ad sets, and ads in a way that decentralizes spend and load balances it evenly across the infrastructure. This allows us to scale our spend by adding additional clusters, not increasing the size of existing clusters. Why we do it. The biggest enemy of scale is entropy. At a certain load, disorder and diminishing returns set in. To avoid this, we architect our ad accounts so that the spend is evenly distributed at safe levels without entropy. Not at little sissy levels. We want to take them to the levels right on the knife's edge of entropy, because that's where things work their best. Not in really safe zones, but in almost danger zones. We're trying to get everything on that knife's edge. Not over it, not too far before it, on it, everywhere. That's where the magic happens. Initially, we scale using one ad account and one cold traffic campaign with all your audience and angle variation happening at the ad set level within that one campaign within that one ad account. What we're trying to do first, initially, is we're trying to scale all of our winning angles to 5 to 15 times cost per lead. That's our initial thing. We're testing five angles with four images each and 30 audience interests. We're finding the winners, and when we find those winners, we dupe them and set the budgets to 5 to 15 times KPI. Once we get them there, cull the losers, keep the winners. That's our first round of optimization and scale. From there, we find those winners and we create ad variations on those angles, and then we scale those to 5 to 15 times KPI. Then we do another variation. Maybe the first variation we do, we change the button. The next one, the headline. The next one, an image. And then we keep going until we reach the limits of our audiences, which is not exceeding $10 per 10,000 people. I must have just copy pasted that same damn thing in there. That's why it's identical in every slide. So, not exceeding $10 per 10,000 people. We're trying to get right up to that knife's edge. But, entropy exists everywhere, not just in the spend. We see it occur at the ad set level when we spend less than two times KPI, or if we spend more than 15 times KPI, we see entropy happen at the ad set level. But it doesn't just occur at the ad set level. We see entropy occur at the campaign level at about $1,000 day. So, once you have, let's say, like seven ad sets that are running at 15 times KPI at $10 each, let's say you're a bit over $1,000 a day in spend, then you're going to start to see entropy occur at that campaign level. Just like an ad set likes to be between 5 and 15 times KPI, a campaign likes to be around $1,000 a day. That's where it likes to be. At the ad account level, it kind of likes to be at about $5,000 a day, and over that it starts to see a bit of entropy. Entropy happens at multiple dimensions. A lot of you aren't going to be anywhere near $1,000 a day for a while, so you're good with one cold traffic campaign and one ad account. All of your variation is going to happen at the ad set level within one cold traffic campaign within one ad account. Simple. Once you get to $1,000 a day in total spend, then you want to create another campaign. Then when we create that other campaign, how do we architect it? Once you've hit $1,000 a day in total spend for a particular campaign or your cold traffic campaign, then you want to create different cold traffic campaigns based on the angle. If you've got one angle that's 26-Year-Old Punk and you've got another angle which is Advice for Consultants, and they're both running within one cold traffic campaign, the moment we start hitting $1,000 total spend per day with that one cold traffic campaign split across those two angles, I now want to create a separate campaign and separate both ad angles. So, I'm going to have a cold traffic 26-Year-Old Punk campaign and I'm going to have a cold traffic Advice for Consultants campaign. Within those two cold traffic campaigns, you are only going to find that angle or variations or ad variations of that angle going to all different audiences at budgets of 5 to 15 times KPI. But then the moment we start, let's say we then have five ad angles, like five different ad angles, all at $1,000 a day in spend for cold traffic, and within those five campaigns, they consist of seven ad sets at 15 times KPI at $10 cost per lead, so they're all at about $1,000. So the ad set level is at its knife's edge efficiency. We can't exceed that. The campaign level is at its knife's edge efficiency. And now, we've run out of room in that ad account. It's at its knife's edge efficiency. Only at this stage that we have five angles, within there we have seven ad sets at 15 times KPI, and we're spending $1,000 a day in cold traffic across five angles. Only at this point now do we consider creating a new Facebook ad account to start load balancing across another dimension. This is what I mean by decentralized architecture. We build based on the parameters and the thresholds of the system. Lucky for you, I already know what all of those are, and then I just tell you how to architect it, right? Saves you maybe five years, so congratulations on saving five years and millions of dollars in two minutes. You know not to take ad sets past 15 times KPI, you know not to take spend past $10 per 10,000 on cold, you know not to take spend past $50 per 10,000 on warm, you know you can take spend on hot retargeting as high as you want until you start seeing entropy yourself, and then you know to not have a campaign exceed about $1,000 a day before you have to start splitting it out, and then you know not to have an ad account really exceeding $5,000 before you have to start splitting it out. This is the architecture, and so what you end up happening is when you get up to massive scale, you end up with multiple ad accounts. How you separate ad accounts is by the traffic type. So, I can show you ours. We've got all sorts of different ad accounts, and you can see we just call them Consulting.com 1, all these different things. If I find the Consulting.com number one account, where is this little thing? So here, it looks like we use this for warm and some hot. This is like a warm retargeting campaign. Then let's say number two is we use this one for ... This looks like it's for something else we're doing that is slightly more advanced. Don't worry about that one. Now let's find another one we're using, number three. We've got a lot going on. We've got like 10 ad accounts going, and we justify that because we're pretty much spending $5,000 in all of them, so we're load balancing it all out with the ideal architecture. Then we've got one for running ads for Uplevel specifically, so we're starting to load balance it out that way, too. Basically what we're doing is splitting things out and what we want to do on the ad set level is we want to have one ad account which might be used for cold traffic, one ad account which will be used for warm retargeting and hot retargeting together at first. When we go from one ad account to two ad accounts, we want to have one for cold traffic, one for retargeting. The one for retargeting includes both warm and hot. The one for cold traffic includes the production cold traffic and the sandboxing cold traffic. Then once you're spending more than $1,000 a day in a campaign, create new cold traffic campaigns per angle. Within those cold traffic campaigns per angle, you have ad sets which have different audiences in them, and the same angle or variations of that angle. Those ad sets should not exceed 15 times KPI or the rule of $10 per 10,000 people in that audience, and if you do that, you're good. Once you're spending $5,000 a day on that cold traffic campaign, then you probably want to create another campaign called cold traffic number two, and then you can have more stuff in there. You keep scaling out that way. This is what I mean by horizontal instead of vertical. If we were doing vertical scaling, vertical, centralized scaling, we would just keep one ad account, and we'd just keep increasing the spend, and we would just not care about the entropy and just start losing money. This is centralized, vertical scaling. Decentralized, horizontal scaling is when we go out, not up. We achieve the same thing, but we go out instead of up. We spread across the dimensions. We go into different countries. We go into different ad sets, different variations, different bidding strategies. We go into different audience types, lookalikes and things. And then we spread across the different countries, and then we spread across different campaign structures and different architecture types of the system, and then we spread into different ad accounts, and we just start building a serious machine. A seriously intricate and beautifully designed machine for serious firestorm of ads. Yeah, that should keep you occupied until you're making like $30 million a year. Now, I know we've covered a lot of ground. Some of your brains are probably smoking right now. That's all right, because I've included a cheat sheet available for download beneath this video in the resources section, and it's called Nine Scaling Methods. In this cheat sheet, you can see all nine scaling methods, their pros, cons, and how to execute them using this cheat sheet. It makes life easier for you. This is what it looks like. It's just a summary of all of these methods, so scaling method number one, what it is, why we do it, how to do it, the parameters. For all of our different scaling methods, I tell you what it is, how we do it, and the parameters and everything. I recommend you download this. It's in the resources section beneath this video. Download it, keep this handy, and you can use it as your cheat sheet for deploying these scaling methods, because there's a lot of stuff in here. This will keep you occupied for years. These scaling methods are enough to take you to $40,000 a day in ad spend, and they're enough to grow your business to $30 million per year. This isn't an opinion of mine nor is it what I think. I know, because I actually did it. This is true. I make this claim because I grew my business to this level, and we spend that much on ads, so this is actually the level that these scaling methods and these strategies and this architecture can take you to. I believe it can actually take you further than this, but I'm not comfortable promising that, because I just haven't done it myself. The moment I do, I will change my promise. There is no limits to it. People used to tell me, oh, you can't make $1 million a month [inaudible 02:00:45]. You can't make $100,000 a month in consulting. Oh, you can't spend more than $1,000 a day on ads. Oh, you can't run a webinar all the time. Oh, you can't run an ad that works for longer than a week. You can't do this, or ads don't work in Europe, or ads don't work in South America. There's no way you'll get ads to work in Nigeria. I just hear all sorts of crap and pretty much all of it I've found to be wrong through experimentation. In this Facebook training, you've probably heard and seen a lot of things that go against a lot of the things that people have been telling you. That's fine. The only reason I tell you anything in here is because it is grounded in experimentation and practice. Nothing is theory here. All is grounded and experienced and documented and proven over time and practice. Armed with these methods, you will decimate the competition on Facebook. I know that one too, because I've been doing it for a long time. With these methods, you can seriously take your business to the moon and unleash a firestorm of advertising all over the planet. Congratulations on going through this Facebook ads training. It's been pretty intense. We've covered a lot of stuff. This scaling video really just adds the cherry on top. With these methods you can really scale anything to massive levels. Thanks for watching this training, and I look forward to seeing you in the next one soon.

Comments