Horizontal Scaling: How To Scale Facebook Ads To The Moon

0:00[Music]

0:10everyone Simmons here and welcome it to

0:12this module called horizontal scaling

0:15now in this module I'm going to show you

0:17how to scale your facebook ads rapidly

0:21to insanely high levels so you know I'm

0:26not just talking about how to scale from

0:27a hundred dollars a day to you know five

0:30hundred dollars a day but if you want to

0:32do that you know you're going to learn

0:33how to do that in this training - but

0:34I'm talking you know how you can scale

0:37to five grand a day ten grand a day all

0:39the way to forty thousand dollars a day

0:42in spend and if you're doing that in

0:44spend then you're making like a hundred

0:46and twenty grand today and that's a lot

0:47so you know in this module I'm really

0:51just going to pull back the curtain and

0:53unleash all of the best strategies and

0:56tactics that I've accumulated and

0:59learned over I think it's been about

1:02seven years of using Facebook Ads so

1:06everything I've learned spending

1:07millions of dollars on ads you know

1:09making tens of million dollars from

1:11Facebook ads and the seven years of it

1:13I'm gonna really boil it down and

1:15distill it and crystallize it and give

1:17you the best most up-to-date

1:20cutting-edge scaling strategies so

1:23here's what we're going to cover we're

1:26going to start off with what horizontal

1:28scaling is and how it works

1:31I'm gonna explain that first and

1:32foremost then we're going to cover eight

1:37scaling methods eight proven methods

1:40that you can deploy and execute to

1:43really take your ad account to the next

1:45level now the first one method one is

1:48duplicating ad sets and max budgets the

1:52second one is new audience interests the

1:56third is creating look like audience the

1:58fourth creating new ad variations the

2:01fifth creating new ad angles

2:03the six targeting additional countries

2:06the seventh scaling your retargeting

2:09campaigns and method eight is

2:12decentralized architecture so eight

2:15proven methods

2:17and in this module we're gonna dive into

2:19each one I'm going to show you what it

2:21is why we use it like the logic behind

2:24it and when we would use it in what use

2:26case in what scenario and then I'm going

2:29to show you how to set it up I'm going

2:32to show you in my Facebook Ads account

2:33where to click and what to do and

2:34everything and we're going to cover the

2:37full you know the full end-to-end what

2:40it is and how to actually execute it and

2:42manage it for each one of these eight

2:44methods so let's get to it

2:47what is horizontal scaling so scaling is

2:53when we find winning ad sets we scale

2:57them by increasing the spend and the

3:01reach and the size of the campaign and

3:05now let me give you an example let's say

3:08you're spending $100 per day on Facebook

3:10ads and you're making $500 a day back

3:12you know that's your return so you're

3:15putting $100 a day into Facebook and

3:17your sales are about 500 a day well in

3:20this situation you want to increase

3:23spend as much as possible so that you

3:26can increase your return as much as

3:27possible because if you find a machine

3:29on the side of the road and you put a

3:31dollar in and five comes out you you

3:34want to try putting another dollar in

3:35and if five keeps coming out you put

3:37another dollar in and then you want to

3:40start putting as many dollars into that

3:42thing as you possibly can because why

3:44would you not that's why we want to

3:47scale when we find something that works

3:49we want to take it to the limit we want

3:52to take it to the extreme now the

3:56general rule is that typically if

4:00something works at a small scale it

4:02should work at a large scale and we see

4:06this with our proof of concept you know

4:08if we find well that we see this you

4:12know first of all we see this with our

4:14market research if we find like a bunch

4:17of people in a niche that all have in

4:19share a similar problem then there's a

4:22high chance that you know this is

4:23widespread across the whole niche and

4:25then if we get one client and we help

4:28them and they get results and then we

4:30end

4:31you know if we get five clients we help

4:34them and they get results there's a high

4:36chance that you know we've got a proof

4:38of concept there that we've got

4:40something that can be widely applied to

4:42all participants in the niche to help

4:43them get a promising result and so

4:47there's this general rule that you know

4:49if it works at a small scale it should

4:50work at a large scale

4:52but the problem is when most people try

4:57to scale their campaigns they increase

5:00spend but they increase the ad spend but

5:03they don't increase the return and they

5:07simply pay higher prices for the same

5:09amount of traffic and customers which is

5:12what we call diminishing returns so for

5:14example they might you know be spending

5:17a hundred and making five hundred then

5:20say they start spending a thousand but

5:23they only make 1,500 so you know as they

5:29spend more the Irate of return goes down

5:33they're still you know they're still

5:35making more money but their rate of

5:37return is going right down and this is

5:41the problem and the solution is to

5:44increase sprint without increasing

5:48prices and we do this using a strategy

5:51that I've created called horizontal

5:55scaling in horizontal scaling is where

5:59we decentralize and distribute spend

6:02across multiple dimensions keeping the

6:06same ROI at a larger scale now you might

6:11not understand what the hell that means

6:12in these words so I'll make it simple

6:15for you I'll explain it to now because

6:17this is like a this is like a new way of

6:23thinking it's like a new paradigm for

6:26advertising on the internet and once you

6:29understand it like this it'll make

6:30absolute sense especially if you've run

6:32ads before it'll just click for you so

6:36here I've got something on the screen

6:37called the scale versus return continuum

6:41or the scale versus ROI continuum now

6:44over here

6:45the left side we've got you know low

6:48scale and over here on the right hand

6:50side we've got high scale now typically

6:53what we see is when we're at a small

6:56spend like if we're only spending $100 a

6:59day $50 a day we get a high ROI and for

7:05example let's say we spend a hundred

7:06bucks we get $2 cost per clicks on

7:09average that means we get 50 clicks and

7:11let's just say we get $500 in sales and

7:16that means that you know if we spend 200

7:19and we made five we've got a 500% are

7:21away right and that's what I mean by low

7:24skal low scale high are away but then as

7:27we start scaling up and we go along the

7:30side of the continuum we start getting

7:32high scale but low ROI and let's say we

7:37spend a thousand dollars but our cpc's

7:40go up to five dollars that means we

7:44don't get ten times the clerk's you know

7:47if we got ten times the clicks would be

7:48getting 500 clicks but no we only get

7:51200 clicks so we only get four times the

7:55clicks for ten times the price because

7:58the price increased right and this means

8:03that you know we don't make as much

8:04revenue we only get about a thousand

8:06dollars in revenue and then we get a

8:08zero percent our way and this is the

8:12typical thing this is the typical

8:13scenario where somebody has you know

8:17really good ROI and everything's working

8:19here at the scale and they scale up it

8:22up of it and you know the prices blow

8:25out and they make pretty much nothing

8:29and so what they do in this situation is

8:31panic and retreat back to the safe zone

8:36and then they stay there forever until

8:39maybe they one day get the balls to try

8:41it again this happens again and then

8:42they retreat back to the safe zone and

8:44then they probably stay there again

8:46until they forget why they were staying

8:49there and then they do it again and then

8:51they come back again this is what

8:53happens all right so this is the problem

8:55that everyone faces and it's big

8:58nobody has figured out how to transcend

9:01this continuum nobody has figured out

9:04how to get high scale and high ROI

9:09everybody is at the mercy of these bores

9:13here until now and what I want to do now

9:20is introduce you to the scaling strategy

9:23that we have used honed perfected and

9:25refined which is called horizontal

9:27scaling and instead of scaling up with

9:32one we scale across with many so just

9:38imagine this for a second

9:40imagine if you've got you know if you're

9:43spending $100 a day and you're making

9:46500 and return and you want to scale up

9:49to a thousand well by simply increasing

9:52the budget from a hundred to a thousand

9:55on that existing Ed said whatever we

9:56blow everything out but now imagine if

10:00we could just clone ten of these small

10:04clusters here because we know these

10:06clusters perform at $100 spend so what

10:10if we just created ten of them instead

10:14of increasing the size of the one up ten

10:17times we're achieving the same end

10:20result of spending ten times more but

10:22the way we've architectured and

10:24distributed and load-balanced that spend

10:27keeps all the individual clusters at

10:30$100 each this is what I mean by

10:34horizontal scaling instead of vertical

10:36scaling vertical scaling is when you

10:38increase the size of the one unit and

10:42horizontal scaling is when you keep the

10:44size of the one unit the same and you

10:46just get more units it's distributed and

10:50load-balanced and so this is what it

10:55looks like let's say we've got a hundred

10:57dollars a day in ad spend and we're

11:01making our own five to one our way

11:03that's why it's green because we're

11:05making good ROI but now let's say we

11:08change the budget on this and we go to a

11:12thousand

11:12and but now we are not making our away

11:17and that's why it turns red so when we

11:21change the state of this thing from one

11:24hundred to a thousand it goes from you

11:27know being in order to being in cows and

11:29it goes from being an hour away to not

11:31being an hour so we face this issue when

11:34we change the state of this thing by too

11:36much however if we just create ten

11:41individual instances of this thing we

11:45get to leave it in our away in its good

11:49zone and we just get ten times more of

11:53it that is what horizontal scaling is

11:57and this way we keep ten individual

12:01systems in order instead of having one

12:04system in chaos and here's how I can

12:08further you know illustrate how we do

12:13this so vertical scaling would be like a

12:16centralized method and this is where you

12:20find a winning ad set and then

12:21increasing the budget of that ad set

12:22resulting in entropy which is disorder

12:25right and we know this too how I talk

12:29about this in the accelerator program

12:31and I'm pretty sure I talked about not

12:32people too is we're trying to build

12:34decentralized consulting businesses

12:36where we're not the only source of

12:39information and you know there's a

12:41community this Q&A cause there's all of

12:43these different things right we're not

12:44the only source and we're also trying to

12:48set up how add accounts this way too

12:50weird we don't want one thing to be you

12:54know to be have that much load on it

12:57because if that one thing plays up then

12:59everything falls down instead we want to

13:03decentralize it and distribute it and

13:05that's what horizontal scaling is and

13:07this is where we find a winning ad set

13:10and then we duplicate it for example ten

13:12times to achieve scale without entropy

13:17now you might be thinking well this is

13:20an awesome theory Sam and it's a good

13:23con

13:24but how do I actually apply this theory

13:27and practice and there's a good question

13:30I hate theory without it being proven

13:33and practice so now let's apply it and

13:37you might be thinking well you know how

13:40do we actually distribute that spend how

13:43do we load balance it and the thing is

13:46with Facebook is well the thing about

13:49uni system right is there's so many

13:51dimensions in which you can apply the

13:55you know the load balancing so for

13:58example you know you could just let's

14:00say you had 10 ad accounts and you just

14:03spent a hundred dollars in each ad

14:04account and each ad account had one

14:07campaign and one ads it right I mean

14:09that would be that could be one way you

14:11could distribute the load over ad

14:13accounts all let's say you had one ad

14:15account and you just created 10

14:18campaigns well then you distribute the

14:20load over 10 campaigns and one at

14:22account or what say we had one ad

14:24account in one campaign but 10 ad sets

14:27and we distributed the load over ad sets

14:30you get my point here I'm talking about

14:34along what dimensions do we apply the

14:38scale because there's a lot of different

14:39things we can play with there's

14:41different campaigns there's campaign

14:43types there's a dangles there's

14:44audiences there's ads there's placements

14:48there's budgets

14:48there's optimization strategies there's

14:51there's so many different things we can

14:53play with different dimensions going all

14:55over the place

14:56now lucky for you we've played with

14:59pretty much all of them here we have

15:02tested everything even things that seem

15:04totally crazy and totally whack because

15:07we just want to tinker and experiment

15:10and play with everything to find out how

15:12this machine works and we have a pretty

15:15good idea

15:18now how we apply the horizontal scaling

15:21so horizontal scaling is when we

15:23decentralized and distribute spend

15:26across multiple dimensions to hold our

15:28away at scale now what our dimensions

15:31well we need to say dimensions with

15:33Facebook I'm referring to add accounts

15:36fan pages

15:38campaigns ads its ads audiences

15:40placements campaign types conversion

15:42objectives budgets bid strategies

15:44countries demographics basically every

15:47option in every every option in every

15:51layer level and feature that Facebook

15:54Gibbs provides another dimension of how

15:57we can use and interact with the system

15:59right now lucky for you I'm not just

16:03like letting you go and just say go

16:05load-balanced it somehow right because

16:07that would be very hard to figure out

16:09you wouldn't know where to start so

16:11instead what I'm going to give you is

16:12eight proven methods for scaling out

16:19horizontally and our strategy is to

16:23scale horizontally across eight

16:24dimensions using the following methods

16:27so this is eight dimensions in which we

16:30can play with now the first one is to

16:34duplicate ad sets and max out the

16:37budgets another dimension is to just

16:41look for new audience interests and grow

16:43that way another one is to create

16:46look-alike audiences another one is to

16:49create new ad variations so not

16:51completely new ad angles but variations

16:54of the angles and then another one is to

16:57actually create completely new and

16:59angles and then when we get new ad

17:01angles we can create variations of those

17:03new angles and then when we've got a new

17:06angle and new variations we can send

17:08those out to look likes new audience

17:09interests and then we can duplicate

17:12those in max out the budgets so each one

17:15of these can interact with the other

17:17ones so these are eight dimensions in

17:20which we can play but every one of these

17:23can be applied in combination or with

17:26any of the others so this gives us like

17:30a Swiss Army knife of tools and

17:35possibilities here and it gives us

17:37really the opportunity to it really

17:40means that you know sky's the limit

17:42I've been able to take our ad account to

17:44about 40 grand a day

17:45but that's no by no means the limit I

17:48think it can go much higher and another

17:51one

17:52is to target additional countries so you

17:55know you can start scaling out

17:57horizontally that way another one is to

18:00scale your retargeting campaigns so not

18:03many people talk about this you know a

18:04lot of torque happens in cold traffic

18:07and don't get me wrong most of the heavy

18:09lifting in your ad account is going to

18:12happen in your cold traffic campaign but

18:14this doesn't mean that you can't scale

18:15retargeting and you know we've figured

18:18out ways to scale retargeting and then

18:22the final one method 8 is decentralized

18:25architecture and this is some ninja

18:28level stuff which we will get to you

18:31know at the end of this module so let's

18:35cover method 1 duplicating ad sets and

18:37max budgets so what is it so we identify

18:43winning ad sets that is an sets that are

18:45within KPI and we duplicate them and set

18:49the budget at 5 to 15 times cost per

18:51lead now why do we do it well changing

18:57budgets on existing ad sets is a bad

18:59idea as it throws the algorithm out of

19:01balance and instead we duplicate the

19:04existing ad set to create a new one and

19:06we use this opportunity to increase the

19:08budget so we're to placate in it so we

19:12can have a fresh start with a higher

19:14budget instead of interfering with the

19:16existing one in changing its budget

19:18which will probably make it stop working

19:20and it won't give us a clean test now

19:24why 5 to 15 times cost per lead why this

19:28number how did I arrive at this magical

19:30zone well ad sets perform well at a

19:34small scale and they start

19:36underperforming at a larger scale as

19:39I've shown you with our continuum of low

19:43scale high ROI high scale Loara

19:46right now the question is what's the

19:51optimal scale point for an ad said

19:54before it starts facing entropy which is

19:57disorder and the answer is 5 to 15 times

20:01cost per lead and here's why

20:05so here's our scale verse return

20:08continuum like I said at low speed weak

20:11ROI high speed we get low are away when

20:14we're getting high are away the system

20:16is in order when we get low it's and

20:19it's facing entropy it's in chaos and

20:22you know with this is the thing we're

20:25constantly balancing alright but the

20:27question is at what point along this

20:29continuum is it optimal what's the most

20:33we can get to before we face intra P and

20:37we've tested it like I said over we've

20:41been advertising on Facebook for seven

20:42years millions of dollars multiple

20:45millions and spend and we've been able

20:47to spend up to 40 grand a day been

20:50making more than a hundred grand a day

20:52we've made more than ten million dollars

20:54from Facebook and we've tested a lot of

20:58stuff and this is what we've found we've

21:03found that if an ad said what took takes

21:07an F an ad like its optimal level for

21:10its budget is based not on a dollar

21:13figure but on a multiple of the KPI now

21:19what I mean by this let me show you I'll

21:24go to my ad account and we use this demo

21:30one here

21:36so when we created let's see when we

21:40created our cold traffic campaign so

21:43this cold traffic campaign we are

21:45optimizing for conversions which has

21:48leads right we're optimizing for cost

21:51per lead which is for the vs our cost

21:53per VSL optin or the JIT cost per JIT

21:57registration or the to ka cost per to ka

22:00registration now the ad sets if I go

22:04here and I look at how they're being

22:07optimized these things are optimizing

22:10for lead conversions right and then if I

22:16look here at yet performance you know we

22:21can see that these they're optimizing

22:27for performance here like which is a an

22:30optin this is what it's optimizing for

22:33it's learning how to get these as

22:36efficiently as possible

22:38all right that's why when you create a

22:40new ad see it'll say it's in its

22:42learning phase and it needs to get so

22:44many before it's completed its learning

22:45phase and all of this right so what it

22:49really is is it's a multiple of this

22:54conversion here which really dictates

22:58the optimal level of efficiency for an

23:00answer so let me make it this clear

23:04let's say you're getting leads for ten

23:06dollars you're getting obviousl opt-ins

23:09for ten bucks so if your budget was set

23:12at ten dollars then that's one times

23:15KPI because the KPI is the cost per

23:20result that the ad set in the campaign

23:22is optimizing for in this instance it is

23:25the lead and we're getting the leads for

23:28ten and our budget is ten that means

23:31that our budget is 1 times KPI right

23:34make sense now what we've found is that

23:391 times KPI is the absolute limit the

23:44bottom level limit that an ad set can be

23:47set at in order to do anything

23:49if you set your budgets at less than 1

23:54times KPI your ad sets a set up to fail

23:58so if you're getting leads at $15 each

24:02on average they say you're getting JIT

24:05registrations or veer sale registrations

24:07for $15 on average but you've set your

24:11ad set budget at $5 you're screwed

24:16because listen to what you're telling

24:19this algorithm to do you're saying hey

24:21mr. algorithm I I typically pay 15

24:25dollars for these things that's what my

24:28history shows and that's what I'm

24:30actually happy to pay for it however I'm

24:33tasking you to go on a mission and find

24:37these things for me for $5 each in fact

24:42I want multiple of them for $5 each even

24:45though I know that it's quite normal for

24:48me to pay 15 that is what you're telling

24:51the machine to do and the Machine just

24:55breaks down because it can't do that so

24:58it won't work and it's was full thinking

25:01if you think it will now the so this is

25:04why if you have your ad set budget set

25:06at less than 1 times kpi they will not

25:09work now the minimum budget you can see

25:13at your ad sets out to have them perform

25:15as 1 times KPI

25:17and you should be getting leads for in

25:20between 10 and 15 bucks somewhere around

25:23there and that's why in the training I

25:27told you that the minimum amount you can

25:29go for your ad sets is about $10 and

25:32it's only during the testing phase that

25:35is I had to you know I had to make a

25:37trade-off there in the training because

25:40you know people were restrained by their

25:43budgets so in an ideal world I would

25:45test with $50.00 per ad set per day but

25:49I can't tell people to do that when they

25:51are restrained by budget and they're

25:53also trying to test different audiences

25:55and different angles and so that's why I

25:58recommended you know 10 but really if

26:03you're not restrained by

26:03you should always be testing with at

26:05least $50.00 per day per ads it it's

26:09just it just gives the Adsit more

26:11breathing room because if you sit at

26:14your budget for an answer at one times

26:15KPI let's say you're getting leads for

26:1710 and you set the ad set budget at ten

26:21that ad set can only really get one

26:24conversion a day because you've said

26:27it's budget at 10 it costs at 10 to get

26:29one it's probably only going to get one

26:31a day some days it might get zero now it

26:35can't perform very well when it only

26:38gets to fire its event once a day it

26:42doesn't get very excited it doesn't pick

26:44up momentum and it doesn't start racing

26:46out and you know finding new new

26:51conversions for you because it's kind of

26:53sitting in this zone where it can't do

26:57much now two times KPI things get a bit

27:00better now to make things clearer it can

27:04function here it can function it cannot

27:07function here it can here it's actually

27:10most I would personally never really go

27:14below this but if you're restrained by

27:17budget you can do this but I personally

27:19would never go blood us now at two times

27:22KPI that adds it is going to be able to

27:24get two conversions per day on average

27:26that's twice as good as one so it's

27:29going to probably learn and reach and

27:31perform way better than one at five

27:37times KPI so for getting leads at ten

27:39dollars

27:40we set the budget F at 50 that's five

27:43times KPI that means that ad set can get

27:45around five conversions a day that's

27:48learning and performing five hundred

27:52percent better than this and so it's

27:55going to work way better it comes into

27:58its it comes into its its own at this

28:01level and then ten times KPI you're

28:05getting leads at ten bucks you see it

28:06your budget at a hundred for your

28:07headset that means the ad set is going

28:10to be getting ten conversions a day now

28:14things start happening and the optimal

28:17lever alive

28:17founders around 15 times kpi so you're

28:20getting leads for ten you see your ad

28:21set budget at 150 it gets 15 leads a day

28:26and that's really where it likes to be

28:28so here I say ad sits experience entropy

28:32at budgets less than two times KPI

28:34so you don't want to go really below 2

28:37times KPI and optimal efficiency exists

28:43at 10 to 15 times KPI that's where the

28:45sweet spot is 10 to 15 times give you a

28:48as soon as you start going above 15 you

28:53it's not really 15 I mean you can go all

28:56the way really up to 18 19 it starts to

29:00get dicey again at about 20 times to KPI

29:04so it's a spectrum it's bad down the

29:07bottom it gets bad up the top the sweet

29:09spot is right here at 10 to 15 times KPI

29:11and so this helps you immensely knowing

29:17how this works in knowing the parameters

29:19of the machine now when we test our ad

29:24sets initially when we are creating our

29:265 angles with 4 images each and we're

29:28trying to go out to through the

29:30audiences we've got 600 possible

29:32combinations we're trying to find the

29:34best ones what wins what losers when we

29:36we're doing that with like a $10 per

29:39head per day budget which is one times

29:42KPI which is not optimal but it's okay

29:45for testing right now once we identify

29:49the winners and we Cal all the losers

29:53that's when we want to scale our winning

29:57ad sets out of this zone and into this

30:00zone and this is where we scale our

30:02winning ad sets to the maximum possible

30:04budget which is roughly 15 times KPI

30:07because if we're just leaving them down

30:09here we're basically leaving them to

30:11struggle even though we know they're

30:12winners when we could unleash them up

30:15here so that's why our first scaling

30:21method is duplicating and sets at max

30:24budgets so how we do it is we find your

30:30best performing ad set

30:31and this is the most within KPI ads it

30:34so winning ad sets or ads it's within

30:37KPI over a four-day time frame right a

30:40winning the best performing ad set is

30:43the best one so over a four-day time

30:45frame which one has been getting the

30:48most leads at the best price and it's

30:52also taking into consideration deeper

30:55funnel conversions too so if you've got

30:57some ad sets that have got you know less

31:00leads at a higher price but they've got

31:03strategy sessions or customers that one

31:06beats one that has more leads at a lower

31:09price but no strategy sessions or

31:10customers right you get how it works

31:13so a winner isn't always done on cost

31:16per lead or number of leads it's done on

31:18the deepest funnel metric and it's

31:20performance there and in the absence of

31:22that we use leads and now when we

31:25identify our best performing ad set we

31:28want to duplicate it and we want to keep

31:31everything the same but we want to

31:34change the budget to five to fifteen

31:37times cost per lead now an example if

31:40the ad set is getting leads at ten

31:42dollars and it's budget set ten dollars

31:44well we just duplicate that ads it leave

31:47everything the same and set the budget

31:48anywhere between five $50 $150 a day now

31:53let me show you how this works

31:55they'd say identify this one is the

31:57winning ad said all I would do is go to

32:00placate I put it in the original

32:03campaign I go to placate I leave

32:12everything the same but I just changed

32:14the budget and I want to go in between

32:17five and 15 times KPI where in there you

32:23go is up to you and how much you have

32:27available for budget to spend on your

32:31budget you well if you've got let's say

32:34you had to kill a lot of losing and sets

32:36and you're wanting to spend more per day

32:39and you're trying to find ways to spend

32:41more then oh it spends like 10 to 15

32:44time's keep you but if you're restrained

32:46on budget a bit I mean you'd probably go

32:48to the lower end of that five times keep

32:50your something like that so that's how

32:53it works now there's a catch to this too

32:58and this is what I mean by perimeters so

33:03on cold traffic

33:04we shouldn't exceed $10 per 10,000

33:07people in a given audience and if you

33:09split audience interests if you're split

33:13audience interests are big enough to

33:15handle 5 to 15 times cost per lead do it

33:18however if your split audience interests

33:21aren't big enough

33:22pull the winning audiences together into

33:24one new audience and use that so here's

33:27what I mean by this so here when we were

33:31sitting our when we created our headsets

33:36right we had we split them by different

33:41audience interest so this one here is

33:44targeting Frank in sense right and then

33:54this one here is targeting Tim Ferriss

34:04right so this is what I mean by split we

34:08split out all of our audience interests

34:10to test them now we can only increase

34:15the budget we you know we're not just

34:17unlimited in how much we can increase

34:19the budget here we have to only increase

34:23the budget so that it's we're not

34:27breaking the ten dollars per 10,000 rule

34:31so if I have if my audience is you know

34:36a million people bug by Tim Ferriss

34:38audiences BIC that'll handle it then you

34:41know I'm fine I can spend 150 bucks easy

34:44because 150 out of a million is not you

34:48know a thousand dollars is is the max

34:52Bend you can spend with a million 150

34:54isn't near a thousand so we're good

34:57however if you're

34:58tiny then you know you can only increase

35:03your budget for that audience

35:06to the $10.00 per 10,000 limit so if

35:11it's only 20,000 people your audience

35:13then you can only really spend $20 all

35:17right so that's what restrains your move

35:21here but hopefully all your audiences

35:23are good enough to handle you know at

35:26least 50 bucks

35:27or better yet 100 or 150 then you're

35:30good you can scale these up

35:31now in the case that you've got a bunch

35:34of audiences that are proven winners but

35:36they're all small what we want to do

35:39here is instead of targeting them all in

35:41little clusters we want to group them

35:44together now to do that inside choir

35:49when you're setting up your audience

35:52templates and I showed you how to do

35:54this in the going live module in the in

35:59this week's Facebook ad training and you

36:02know when you create audience templates

36:05in choir and what we do there is when we

36:08launch our our audiences into choir we

36:12select an option that says split and

36:14that splits them out into separate ad

36:16sets so what we want to do is we want to

36:18identify our winning ad sets our winning

36:21audiences sorry so we want to find all

36:24the audience's that work and have been

36:26proven to work and we remove all the

36:29audiences that didn't work and then when

36:32we launch those audiences into choir we

36:35simply select Paul and we don't select

36:40split so I think we actually by not

36:42selecting split we're pulling them I

36:45don't think there's an act there's

36:46actually a pull button or Paul checkbox

36:49we just don't check the split box and by

36:52doing that they pull and that way we're

36:55putting all of those audience interests

36:57into one ad set and that means if

36:59there's 10 audiences 10 audience

37:02interests that are two at 20,000 each

37:05then that's going to pull them together

37:07to make one audience of 200,000 but

37:09there's probably going to be overlap

37:10between those different audio

37:12centrists so it's probably going to be

37:13more like 150,000 now when we have one

37:18audience that's 150 thousand one ad set

37:21with an audience in it that has 150

37:23thousand people we can now spend about

37:26150 dollars to that instead of having to

37:31do that across all of these little ad

37:33sets because an ad set remember it said

37:36it's optimal at 10 to 15 times KPI but

37:40if an audience interest within an ad set

37:43only allows you to spend twenty dollars

37:46then you've kind of set that ad set up

37:48for failure it is better to pull those

37:51audience interests together so that you

37:54can get up to the optimal spend range to

37:58make that ad set really Hummer this is

38:01the trick sometimes you pull things

38:03sometimes you split them if you can't

38:07spend enough to get into the optimal

38:08zone pool so that you can if you can if

38:13you if your audience is so massive that

38:16you have that you really are able to

38:17spend a ton but you can't you should be

38:20splitting amount so that you can it's

38:23all about architecture and it's all

38:25about distributing and load-balancing

38:28every single thing in your ad account so

38:31that it's sitting at the optimal zone

38:33that's the trick to it this is some

38:37next-level stuff they honestly do not

38:40teach you this anywhere in any Facebook

38:43Ads training anywhere on the internet

38:44because no one spends this much money

38:46and has this level of understanding of

38:49it and so that is what we do here now

38:59let's talk about our second method which

39:03is new audience interests so what it is

39:08so we search for audience interests with

39:11affinity to bevel or winning audiences

39:15we put them in and we look at audiences

39:17that have affinity to that and then we

39:21test them to find more winning audiences

39:24and expand

39:25the audience size of our campaign

39:27therefore increasing thresholds pinned

39:31now why do we do it so in order to spend

39:34more we need to increase the size of our

39:36audience because we can't exceed $10

39:39spend per 10,000 people in a cold

39:44traffic campaign

39:45now it says 1,000 here so I'm going to

39:47quickly change that because it is 10,000

39:51don't want to cause any confusion here

39:56so you know we want to scale that this

40:00thing can we can really constrain us

40:02sometimes because we can't break this

40:04rule if we do break this rule then we

40:06experience entropy anyway so there's no

40:09point now

40:11our audience size dictates our threshold

40:14spins limit plain and simple so in order

40:18to be able to scale a lot of the time we

40:20have to be able to expand our audience

40:22so that it can take that additional

40:24spend it's not just about spending more

40:27it's a lot it's getting an audience

40:30that's able to handle that spend and

40:32then spending more and that's why this

40:35is really a prerequisite so increasing

40:38the size of your audience alone will not

40:40scale your campaign however it provides

40:44you with the ability to further scale

40:46your campaign and that's why this

40:48strategy is used as a prerequisite to

40:50scale now how do we do it so we use

40:55Facebook audience insights tool to find

40:58new audience interests with affinity to

41:00bevel or any winning audience and when

41:03we find them we add them to our

41:05spreadsheet for testing and our

41:08spreadsheet is our Facebook audience

41:09angles in a major spreadsheet where we

41:11put down our ideas for audiences and

41:14then what we've tested and which ones

41:18are proven and which ones don't work and

41:22so we were trying to look for new

41:24audience

41:25interests to test we've probably tested

41:27our initial 30 doing our doing our

41:31initial launch where we launched

41:32Facebook's golden mean with five angles

41:35with four images each to 30 different

41:38audience

41:39interests now if you've already done

41:41that and out of those 30 you found five

41:43that work then we want to test another

41:4530 to find another five and then another

41:4830 to find another five and we want to

41:50test enough groups of 30 until we find

41:5430 that work and then when we have 30

41:57that work we have a big proven audience

42:00size which can handle a lot of spend

42:03which gives us the ability to spend more

42:05and then we can spend more so scaling is

42:10always a balance of trying to widen your

42:12audience and then spending more and then

42:15trying to get things to perform at their

42:16highest pinned so let me show you how

42:23you do this real quick

42:24but I've shown you how to do this in

42:26previous modules so you know I'm not

42:28going to go really like step by step on

42:31this one because I've already covered it

42:34and if you can't remember how we did it

42:36then you know you've got bigger issues

42:39than me showing you how to scale you

42:41should go back and learn it so I'll go

42:44to the audience insights tool and here

42:55we want to put in our tariff level

42:58audience let's say that is Tony Robbins

43:05right you know you can go to page likes

43:07and I can look at all the page likes

43:11that have affinity to Tony Robbins and

43:15you can see down here it shows all of

43:17these now the ones that are closest to

43:19him are the ones that most match him and

43:23you just want to keep going down through

43:25these and finding different ones to test

43:30now if you've tested a bunch of want

43:32them from here then test more of them

43:34from here but you're not only restrained

43:37to having to write to derive audiences

43:40from your tariff or once you've found

43:43winning audiences you can use them to

43:45derive audiences too so let's say I

43:49started using Tim Ferriss

43:52and I found that he was working I know

43:54his audience was working because it was

43:56getting leads and strategy sessions and

43:59things so now instead of deriving from

44:01Tony I derived from Tim and now I can

44:08start trying these all right and then if

44:13some of these work like let's see now

44:15that day least week works daily stoic or

44:18stoic however they how are you saying

44:19that and then we plug this in here no it

44:27doesn't want them so let's try it let's

44:30say we try Samsung then we find

44:37audiences from here like there's no

44:39limit there's no end to how many things

44:42you can test here because you can find

44:45tons of audiences that are just a lot

44:48just derived from your Babel and then

44:51you're going to have some that work and

44:53then you can drive tons from those and

44:55then from those some are going to work

44:57and you can derive tons from those so

44:59there's no excuses here you can identify

45:00a ton of audiences and when you find

45:03them the ones that you want to test you

45:06add them to your Facebook audience

45:07angles and images spreadsheet which I

45:10gave you in one of the previous modules

45:11and then we're ready to test it and then

45:15what we do is we create a new audience

45:19targeting template inquire and we put in

45:23all of those new audiences we launch

45:26them we select split so that it splits

45:28them into different ad sets and we push

45:30them live and the campaign we want to

45:34push them live into is our sandbox

45:37campaign not how production cold traffic

45:41campaign we want to push it into sandbox

45:43because we're sandbox and different

45:46audiences here now let's say we push in

45:49this so we find 30 different audience

45:51interests we split them we launch them

45:54with our all proven angles so we take

45:57our angles that are proven to work from

45:59our production campaign and we only

46:02launch those proven angles to

46:05those new xxx audiences through quiet

46:09and our sandbox so now we've got they

46:11said we've got two proven angles then we

46:16are creating two angles with four images

46:21each going to 30 audience interests that

46:26means that we're going to have 60 ad

46:27sets and we want to be testing those

46:30that at least $10 each and sometimes you

46:33might not be able to spend that much if

46:34you can't then lower it you only use one

46:37audience Oh only use one angle your best

46:42one and test as many audience as as you

46:47can and work through it that way and

46:51then when you find winners when you find

46:54proven winners where you're getting

46:57leads within KPI and your sandbox

47:00campaign then you want to duplicate them

47:03but add them to your production campaign

47:06so if I come back here and I go into my

47:10ad account and let's am in my sandbox

47:20and I find a winner

47:23I want to duplicate it but where I

47:25select the campaign so I'll show you

47:28what this looks like in here I gotta

47:31come in from this it so I select one of

47:34these and then I go triplicate it's

47:37going to ask me what campaign and I want

47:40to select to select urgent no existing

47:45campaign then I want to select this

47:48production one so I'm going to be

47:50duplicating out of sandbox and putting

47:52it into my cold traffic campaign because

47:57it's already proven and then when I

47:59launch it into my proven production

48:02campaign I want to set the budget at

48:05five to ten times KPI because it's

48:08already been tested it's already been

48:09vetted now it's time to take it to its

48:11max threshold so that's our workflow we

48:16find audience interests

48:18from the audience interest still finding

48:21things with affinity to babble or any

48:23audience that's been proven to work then

48:26we add them to the spreadsheet then we

48:28launch them in choir using the split

48:30method into our sandbox campaign and

48:32then we wait four days we find the

48:35winners we call the losers we get the

48:38winners and we duplicate them but we

48:41change it from the sandbox campaign to

48:43our production campaign and then we

48:45increase the budgets to the five to

48:48fifteen times max threshold KPI this is

48:51a really really powerful workflow now

48:56let's talk about another one method 3 3

48:59creating look-alike audiences and this

49:01one's a biggie so what it is so when we

49:05have enough conversions we create look

49:07like audiences to expand our audience

49:10size now why we do it because we can't

49:14exceed $10 per 10,000 people in an

49:18audience

49:19damn this thing it's keeps saying 1,000

49:26our audience size dictates our threshold

49:29spend and by creating look-alike

49:33audiences we rapidly expand our audience

49:35size and therefore our threshold spend

49:38so why look-alikes why not just keep

49:41going with audience interests well we

49:46start with one Tower of Babel audience

49:48and then we derive multiple audience

49:50interests from bevel and then we derive

49:52multiple audience interests from our

49:54winning audiences and you know we're

49:58starting with one we're deriving things

50:00from that one when we find new winners

50:03we're deriving things from those new

50:04winners and then once we've been doing

50:07that for a while and we've got enough

50:09data and Facebook and we've got enough

50:11like data and there for it to do its

50:13magic we then use Facebook's algorithm

50:17to derive programmatic audiences and

50:22what programmatic audiences are is it's

50:26we're allowing Facebook's algorithm to

50:29find people who look like

50:31the people we want so they're not

50:33necessarily people who like Tony Robbins

50:35or people who like Tim Ferriss they're

50:39people who might like any one of those

50:41things but that Facebook's algorithm

50:43knows look like the people who we want

50:47it's actually more powerful than

50:49selecting specific audiences it's

50:51probably the most powerful tool in

50:53Facebook heads but you can't use it at

50:57the start you have to start with

50:58audience interests and so that's why we

51:01start with audience interests and we

51:03keep driving we keep creating

51:04derivatives and then when we get the

51:06opportunity we can start using

51:08look-alikes now how do we do it

51:11so once you've had a hundred opt-ins

51:13that's a hundred people register for

51:16your via cell you're jit or you took a a

51:18once you've had that then you can create

51:22a look-alike on add post engagements and

51:27once you've had 300 opt-ins you can

51:31create a look like on the opt-ins pixel

51:34and once you've had 300 strategy session

51:38applications you can create a look-alike

51:40based on strategy session applications

51:42and you start at the top of your funnel

51:45which has aired engagements and you

51:47create look-alikes at each stage as soon

51:50as it has sufficient conversions which

51:52is 300 plus needed now let me show you

51:56how to do this so once you've had a

52:00hundred opt-ins you can create your

52:01first look like before you have a

52:04hundred opt-ins you can't create any

52:05look-alikes so don't even try it

52:07once you have had a hundred opt-ins

52:09you'll know because if you go back to

52:11your campaign level and you look at your

52:14cold traffic campaign here and if you

52:19can see if you set it to you know two

52:22lifetime then it should say here in

52:24results opt-in for via cell to kogo T

52:28there should be the number 100 if this

52:32number is not 100 then you can't do it

52:34once it is you can then how do we create

52:37look like so we want to go up here and

52:41we go to audiences

52:45and then we want to go to we want to

52:51click on create audience look-alike

52:55audience we want to set it at 1% and

53:00then source is we want to go our fanpage

53:06and then what we want to do is we want

53:13to go to location you can set the

53:22locations in here and I'm pretty sure

53:30they've changed this or let me try that

53:32again

53:33cRIO audience custom audience sorry so

53:41yeah what you're doing is you're going

53:43create audience then you're going custom

53:46audience and then what you're doing is

53:48you're going engagements and then what

53:51you're doing is you're selecting

53:53Facebook page and then you're selecting

53:57anyone that has interacted with your

54:01Facebook page the one that you use for

54:03your ads and anyone that who is engaged

54:06with any poster ad and then here we can

54:09see it just in the past like 180 days

54:13all right and we can create this

54:16audience now

54:19once you've had about you know once

54:24you've created that you can create that

54:27once you've had about 100 opt-ins then

54:29what you can do is you can go in to this

54:34audience and you can create a look-alike

54:36based off it so here I can see and

54:39engagements 1 ad D and up here under

54:42actions I can click create look-alike

54:44and here I want to set the audience size

54:48at 1% and then you know my location I

54:53can set the locations whatever and then

54:56I would create the audience

54:59so what we're doing here is we're

55:03creating first of all the audience for

55:06people who engaged with that means they

55:09liked commented or shared or clicked the

55:12link in any of our ads in the past 180

55:15days or organic posts we can create an

55:20audience of that and retarget them but

55:23what we can also do is we can create a

55:26look-alike off this and then use it for

55:29cold traffic because why we use this at

55:33engagements for retargeting because they

55:35retired we were targeting them because

55:36they've already engaged with us a

55:38look-alike is people who look like the

55:42people who engage with our ads so that

55:45isn't retargeting because in order to

55:47retarget somebody we have to retarget

55:50them that that Rimi pne's that they were

55:53there at one stage engaged but these

55:56people look like people who engaged

55:57therefore they did not engage themselves

56:00therefore it is not retargeting it is

56:03cold traffic but it has affinity to

56:06retargeting which makes it good alright

56:08so we first of all create the ad

56:10engagements audience after we've had a

56:13hundred conversions then we can create a

56:16look-alike at one percent and then we

56:19can run ads to that and our cold traffic

56:22campaign and when we do this we're not

56:26creating a new campaign for we're not

56:30creating a new campaign for look likes

56:33we're just putting all of this in to our

56:36production campaign a look-alike

56:39audience can turn a lot of the time like

56:43skip production can skip sandboxing

56:47because it's you know it's a it's a

56:49strong audience so look-alikes I'd be

56:53happy to put those into production

56:55because we're using a proven angle and

56:57we know the proof the previous audience

56:59worked and we're just creating a

57:00derivative of that proven audience that

57:02worked with an angle that we know is

57:04proven to work so it's it's chances of

57:07working are very high so I would skip

57:10the sandbox with it and just

57:12push it straight into production and you

57:15know I would start off with a budget of

57:17at least you know two times KPI but

57:22really I'd want to be in the five times

57:24KPI and then when it's proven to work

57:27you can dupe it and go up to five to 15

57:29times KPI to bring it to its threshold

57:31budget that's how it's done

57:36now once you've had 300 opt-ins you can

57:40create a look like on opt-ins so how we

57:44do this is we're basically creating an

57:50audience first and foremost of people

57:52who opted in so this would be by going

57:57to this would be visited via cell value

58:05video or attend or registered for jit

58:08webinar or interested for 2k webinar

58:10that means they registered they saw the

58:12page after the opt-in form now this here

58:16this audiences everybody that that

58:19basically opted in or interested now

58:23when this whips when this audience size

58:25is 300 or more it has to say here size

58:29300 or more then what we can do is we

58:32can click here we can open it and we can

58:35go actions and we can create a look like

58:37and we want to go just one percent and

58:40put our countries in launch it only

58:44select the countries that the original

58:46audience head countries selected for

58:49we're just keeping all variables

58:52consistent we're just creating a

58:54derivative of the original variables

58:57don't mix things up and cause chaos but

59:00just putting weird things in here keep

59:04it the same and so you can see our

59:08method here we create an audience here

59:11to measure every one that takes an event

59:13that doesn't event we then retarget them

59:16in our warm retargeting campaign because

59:18these people have done something but we

59:20don't stop there we create a decorator

59:23look like based on those people

59:25who did this but that isn't retargeting

59:28because those people just look like they

59:30didn't do it and then we put those

59:33people that look like into cold and we

59:36target them there and we can start at

59:40the top of our funnel which is just

59:41people who engage with our ads because

59:43this one's going to get the highest

59:45numbers the quickest because more people

59:48are going to do this than anything else

59:50so this is the one to start with because

59:52it's going to get the most amount of

59:53data the fastest then once we've got at

59:56least 300 here on the visited this we

1:00:01can create a look like off this and now

1:00:03we can go into that one launch it into

1:00:06our production campaign as per usual

1:00:08with the proven angle and then bring it

1:00:12up to threshold spend 5 to 15 times keep

1:00:16you out now let's say that we've had

1:00:21what's the next link down the chain so

1:00:24the next link is strategy session

1:00:29applications right so once you've had

1:00:32300 of those and you'll know when you

1:00:35look at your audience here that is you

1:00:38know scheduled strategy session then you

1:00:44can you can set scheduled strategy

1:00:47sessions and then you can put like 30 D

1:00:49if you can get it there or you can

1:00:50change it to 180 D and if you can if you

1:00:53can get an audience that has scheduled

1:00:55strategy sessions and you create that

1:00:57audience by going from the people who

1:01:02visited the success page after

1:01:05completing the survey then you can

1:01:08create an audience here it's set it at

1:01:10180 day and you can wait until the size

1:01:14gets to 300 that means you're going to

1:01:16have to have had 300 people complete the

1:01:18survey application form within a six

1:01:21month period and if you achieve that

1:01:23it'll say that 300 here and then what

1:01:27you can do is you can just click here

1:01:29and then you can create a look-alike and

1:01:31you want to see it 1% and leave the

1:01:33countries constant as well and launch

1:01:36that into production with your proven

1:01:38angle

1:01:39and scale it up to max threshold KPI 515

1:01:44times and then you can you can keep

1:01:53working through so if you're doing the

1:01:552k funnel you can start creating them on

1:01:57people who visited the order form people

1:02:00who visited the sales page or people who

1:02:02actually purchased once you've had 300

1:02:05people who actually purchase now you can

1:02:09create a very powerful look like

1:02:10audience because it's people who look

1:02:12like buyers and that's powerful but you

1:02:16know 300 buyers is a lot so you need to

1:02:19work your way down the chain start at

1:02:22the top and work your way through

1:02:24don't create look-alikes on things until

1:02:27they've got enough data on them because

1:02:29you're just shooting yourself in the

1:02:30foot and each you know it's ready to

1:02:35create a look-alike on it when it's had

1:02:37300 people do the action and you can see

1:02:40that in the audience is tool and the

1:02:43size now the parameters when creating

1:02:47look-alikes for the first time use 1%

1:02:49audiences we want to keep them as tight

1:02:51and as close as possible so use 1% and

1:02:56keep your country your countries -

1:02:59what's proven and what's proven is what

1:03:02you were using for the original audience

1:03:04so if you're only targeting America when

1:03:07you're creating a look-alike only target

1:03:09America but if you were targeting

1:03:11America Canada Australia New Zealand

1:03:12when you create look-alike create it for

1:03:16America Canada Australia New Zealand not

1:03:18one for each one look like for all of

1:03:21those together alright that's what you

1:03:25do just keep it the same keep it as it

1:03:27was when you were running it cold and

1:03:32once you've tested and proven

1:03:34look-alikes at 1% later on you can try

1:03:373% and if that works then you can try it

1:03:41even higher and if that works you can

1:03:43just keep going and you can go up to 10%

1:03:46but remember we always want to try to

1:03:49keep things some

1:03:51so if you're going to create another

1:03:54look-alike at a higher percentage point

1:03:57here's what you do because people can

1:04:01make a mess of this so let's say I want

1:04:03to create a look-alike off at

1:04:04engagements at first and when I do that

1:04:08I go create look like and I set 1% here

1:04:13right now let's say this one works

1:04:15really well so I want to try and create

1:04:18a bigger one well what I don't do now is

1:04:20set one at 2% and create one there

1:04:23because what I'm going to do is I'm

1:04:26going to have more people in here

1:04:28compared to the one but I'm also going

1:04:30to have all of the people that the 1%

1:04:32one hasn't it and I'm still running the

1:04:341 and now I'm running the 2% which

1:04:36includes the 1% and now I have overlap

1:04:39and now I'm targeting the same people

1:04:41and I'm making a mess so you click show

1:04:43Advanced Options and then you want to

1:04:46let's say I've already tested the 1 now

1:04:51let's say I want to test a 1 2 3 so I

1:05:00select this band and so here I'm

1:05:04creating I'm leaving out all the people

1:05:06that are in the 1% and I'm creating a 3%

1:05:10look-alike that excludes the 1% so I'm

1:05:12creating it based on this band this is

1:05:15what you want to do this way you don't

1:05:16have overlap this will save you a lot of

1:05:19heartache

1:05:19now let's say this the let's say the 1%

1:05:23works and the 3% works now let's say I

1:05:26want to test a sex well I can just drag

1:05:30it 3 to 6 if that works then I can

1:05:32create a 10 a 10 to 6 right and all of

1:05:38these different look likes aren't going

1:05:40to overlap and include the same people

1:05:41in them because I'm creating it based on

1:05:43bands that's a little small hack that

1:05:48will save you a lot of heartache and

1:05:50trouble and remember always try to keep

1:05:55things simple so don't get too crazy on

1:05:57your look-alikes remember you you don't

1:06:00need a ton of audiences and you don't

1:06:01need a ton of different complexity to

1:06:03make things work

1:06:04you just need enough people in your

1:06:06audience to spend the right amount of

1:06:08money and you need things to perform

1:06:10well and you want to keep things simple

1:06:14now let's talk about method number four

1:06:16which is creating a new add variations

1:06:19so what it is so we create variations of

1:06:23our proven ads so that we can run

1:06:25multiple versions of them to the same

1:06:27audience without experiencing entropy

1:06:30and ideally we want multiple versions of

1:06:33an angle running at five to 15 times KPI

1:06:36recife l now why do we do it

1:06:39well once we find a proven audience

1:06:42angle image combination we want to

1:06:45duplicate it and set the budget at five

1:06:48to 15 times KPI so we can take it to its

1:06:51threshold limit where it's most optimal

1:06:53and then if it works there we want to

1:06:56duplicate multiple versions of that and

1:06:59set them at five to 15 times cost per

1:07:03lead but we don't want it to be

1:07:05identical we don't want the exact same

1:07:08ad running to the exact same audience

1:07:12because it's going to have overlap and

1:07:15so instead what we do is we change

1:07:18something about it and we want to change

1:07:21a small thing like the headline or the

1:07:24image or the button so that it's unique

1:07:27enough to run separate from the existing

1:07:30ad set now what Facebook does here is

1:07:35Facebook's very complicated in how it

1:07:38creates auction pools so let's say you

1:07:43create an ad and then you select an

1:07:45audience and then you bid on it now if

1:07:49somebody else creates an ad that's

1:07:50similar and they going to an audience

1:07:55that similar and their birds are similar

1:07:58in budgets and things are similar then

1:08:00they're thrown into that auction pool

1:08:03with you and the person who performs or

1:08:06pays the most wins that auction pool but

1:08:09we can enter option pools with ourselves

1:08:11so if we create you know the the exact

1:08:17same ad and run

1:08:18to the exact same audience with the

1:08:20exact same everything then we're pretty

1:08:22much going to just compete with

1:08:24ourselves but and we we want to try and

1:08:26avoid that however if we use different

1:08:29audiences with the same ad we're not

1:08:31really competing with ourselves and if

1:08:34we create different angles with

1:08:37different audiences we're not really

1:08:39competing with ourselves and what we can

1:08:42do is we can create variations so that

1:08:45we don't we create set production pools

1:08:48and don't compete with ourselves and

1:08:51I'll show you how to do this so let's

1:08:54say I've got a proven and set and it's

1:09:00my winner

1:09:04in my cold traffic production campaign

1:09:08and let's say I grab you know grab this

1:09:11and I go to to placate it well it's a

1:09:15I've already had this humming at hundred

1:09:17and fifty a day right

1:09:19now I want to droop it again and I want

1:09:22to go for a hundred and fifty again but

1:09:25I want to make sure that you know I

1:09:29changed the add of it so I'd go

1:09:30duplicate and then I'd go you know

1:09:34original campaign duplicate now at the

1:09:39ad level here I'm gonna have to go down

1:09:43to add so we leave everything the same

1:09:46budget we're gonna go five to ten times

1:09:48five to fifteen times KPI but at the ad

1:09:51level I want to change something and at

1:09:56the aired level what I want to do is

1:10:01edit this and so I just want to change

1:10:04you know something about the headline I

1:10:06might use a different headline or I

1:10:09might turn on a button like apply now

1:10:12book now whatever just change one thing

1:10:15about it but change the headline or put

1:10:18a button on it and so this is what you

1:10:22can do or you could try a different

1:10:23image button image your headline those

1:10:26are the things you're playing with

1:10:27you're not changing the angle itself and

1:10:29when you do this and then

1:10:31you review and publish it live you're

1:10:34creating a micro variation of an

1:10:37existing angle and we don't need to do

1:10:41this in the sandbox because the audience

1:10:45has been proven with the angle and we're

1:10:49just changing a slight variation of that

1:10:51so it's highly likely to work so we can

1:10:54just do it straight into production at

1:10:55five to 15 times KPI and you might be

1:11:04thinking why do these small variations

1:11:07why not just duplicate it well like I've

1:11:09said if we just duplicate it then we are

1:11:13competing with ourselves which we don't

1:11:15want because we will create entropy

1:11:17which we don't want now you might be

1:11:20thinking well why don't we just change

1:11:22the budget a bit and also just change

1:11:25the age of it it doesn't really work

1:11:28anymore the you know Facebook's

1:11:30algorithms constantly being updated and

1:11:33they you know they get better and

1:11:37smarter and that one doesn't really work

1:11:39anymore

1:11:40and eventually anyway it would probably

1:11:42experience overlap so it's not the best

1:11:45it's best to actually just change

1:11:46actually change not try to fool it

1:11:50change but actually change the ad and

1:11:54when we do that we actually get put into

1:11:57a different option pool and we actually

1:11:58have a better advantage it works better

1:12:01do it that way now you might be thinking

1:12:04well why not new angles why don't we

1:12:06just create totally new angles instead

1:12:07of doing these variations and you should

1:12:11always be sandboxing for new angles and

1:12:13when you find winners you should be

1:12:16scaling them into production and new

1:12:19angles are an extremely effective

1:12:21scaling method they're actually one of

1:12:24the best in fact it's the best if you

1:12:27create a new ad that is just a killer ad

1:12:30it's the best scaling method you can

1:12:32have but it's also very hard and scaling

1:12:36is not necessarily you know the scaling

1:12:38strategies I'm sharing with you aren't

1:12:40necessarily about how to create if my

1:12:42scaling method was just create a really

1:12:44good ad

1:12:45it's not really people be like oh you

1:12:47serious we knew that would scale so of

1:12:50course that is the best strategy but

1:12:53scaling when I talk about that and when

1:12:55most people talk about it they want

1:12:57methods to increase or to get more juice

1:13:01out of existing angles and so that's why

1:13:05variations are hacks to get more juice

1:13:07out of an existing angle so if you've

1:13:10got an existing angle that works and

1:13:12you've taken it to threshold now you can

1:13:14get more juice out of it by drooping it

1:13:16and making a variation and if then that

1:13:18one works you can droop it again create

1:13:20another variation and get five to 15

1:13:22times KP out of it and you want to keep

1:13:24doing that until you hit the threshold

1:13:27speed limit of that audience so let's

1:13:31say the audience is you know Tim Ferriss

1:13:34and it's got a million people in it and

1:13:36I keep drooping these things at 150 and

1:13:39creating variations $150 the threshold

1:13:43amount that I'm going to be able to

1:13:44spend on that across these different ad

1:13:46sets is still going to obey the $10 per

1:13:5010,000 person rule which if Tim Ferriss

1:13:54audience is a million then I'm only

1:13:55going to be able to spend a thousand

1:13:57divided by 150 we're going to end up

1:14:00with about seven so I'm going to have

1:14:01seven variations of an angle at 150 each

1:14:06going to the same audience before I max

1:14:11that out get it now how we do it so once

1:14:16you've got a winning ad set running at 5

1:14:1815 times KPI

1:14:19or cost per lead triplicated again and

1:14:22change the headline like I showed you

1:14:23image or the button you can do any one

1:14:26of these I suggest just doing one not

1:14:29all because when you do all you're

1:14:31almost creating a new angle and you

1:14:33might screw it up

1:14:34just a micro variation and you want to

1:14:39keep everything else the same including

1:14:41the budget at five to 15 times cost per

1:14:43lead and try not to have more than one

1:14:46identical aired running to the same

1:14:47audience you know if you've got one

1:14:49proven angle running to an audience then

1:14:52only have that running to their audience

1:14:54you can have that identical angle

1:14:57running to other audiences

1:14:59it's fine but if you're going to keep

1:15:01scaling that that or that our angle to a

1:15:06particular audience first of all you

1:15:08stretch it up to five 15 times

1:15:10KPI and then once you've done that then

1:15:13you have to create variations to get

1:15:15more shots at it and scale it up and

1:15:18when you create these variants you get

1:15:21and get them to 15 five to 15 times KPI

1:15:24- you can keep creating variants and

1:15:26taking them to this five to 15 times KPI

1:15:30till the point you reach the audience

1:15:32threshold now let's talk about method 5

1:15:37which is creating new ad angles so first

1:15:44of all sorry

1:15:45it just says number 4 here but this is

1:15:48method 5 so what it is so we experiment

1:15:51for new ad angles in our sandbox

1:15:53campaign and when we find winners we

1:15:56scale them into our production campaign

1:15:58at a 5 to 15 times cost per lead budget

1:16:03and then once they work at that then we

1:16:06create ad variations using the previous

1:16:10scaling method that we covered where we

1:16:12changed the headline the image or the

1:16:14button and so you can see here this is a

1:16:19insane process rights it's so powerful

1:16:22because all of these things we can play

1:16:24with we can combine them and switch them

1:16:26with each other so when we're sandboxing

1:16:29for new ad angles the moment we find an

1:16:32air dangle that works in the sandbox

1:16:34we droop it into production and when we

1:16:37put it into production we take it to 5

1:16:39to 15 times KPI boom once that's going

1:16:42there then we create variations of that

1:16:46by changing with one droop we might

1:16:49change the headline and set it at five

1:16:51to 15 times KPI if that one works then I

1:16:54might troupe the original and change it

1:16:56and put a button on it and then boom if

1:16:58that one works at five to 15 times KPI

1:17:00now I might droop the original and

1:17:01change the image set it at 5 15 times

1:17:04KPI inform networks and now I've been

1:17:06able to expand you know my spend for

1:17:10that specific audience

1:17:11up until the threshold limit which is

1:17:14$10.00 per 10,000 at the limit if that's

1:17:17a million in the audience I'm spending

1:17:19up to a thousand over those different ad

1:17:21sets I'm achieving optimal efficiency

1:17:23I'm reaching the max I can in that

1:17:27without hitting entropy zone now why do

1:17:36we create new ones

1:17:38well just because we have angles that

1:17:40work and that we can create new and very

1:17:46end variations that work just because we

1:17:49can do that doesn't mean we can't make

1:17:51one that works better right never lit

1:17:54good stop great even when you're great

1:17:58don't need that stop you from being

1:18:00great - you're never done you're never

1:18:02finished you can always do ten times

1:18:05better than you're doing no matter what

1:18:07level you're at in the moment in the day

1:18:09you think otherwise someone will come

1:18:11along and take you out so it never stops

1:18:15and you should never stop and so we're

1:18:20always looking for something better and

1:18:22our methods are designed to extract more

1:18:25juice out of an existing angle but

1:18:30creating a better one improves

1:18:32everything so creating a killer ad can

1:18:38make you millions of dollars right like

1:18:40I told you about advice for consultants

1:18:42or 28 you're all 26 year-old punk and

1:18:45those ones have made me Millions

1:18:47so those are like you know killer ads

1:18:50and what you're really aiming for is a 9

1:18:53to 10 quality score and if you simple as

1:18:58it as it is if you scale something

1:19:01you're just gonna get more so if

1:19:03you're spending $100 and you're not

1:19:05making anything if you scale it well

1:19:08then you're going to be spending a

1:19:09thousand dollars are not making anything

1:19:10all right

1:19:11if you scale something you just get more

1:19:14of what you had if you didn't have

1:19:16anything you're not you're gonna have

1:19:17more of nothing if you had a loss you're

1:19:20going to have a bigger loss but if you

1:19:22have something that works you're going

1:19:24to have something

1:19:25that works more and if you have

1:19:27something that's phenomenal you're going

1:19:30to have something that is more

1:19:32phenomenal and so really we want to make

1:19:37sure that you know to scale effectively

1:19:39we want to already have something

1:19:41awesome that's one of the real

1:19:43requirements of big scale you need to

1:19:46have something awesome and you can't

1:19:49scale something that doesn't work and

1:19:51you can't sell you you can't scale

1:19:53something that's average either you know

1:19:55when we scale things we we're making

1:19:58we're testing things so if something's

1:20:01barely profitable if we scale it it's

1:20:03definitely not going to be profitable

1:20:05right you need to have wide margins on

1:20:09things to scale them and if things are

1:20:12barely working then it's not going to be

1:20:15fun so you need to get really good

1:20:17results at a small scale so that they

1:20:19can handle they have the safety

1:20:21tolerance to perform at a high scale and

1:20:26you need a six plus quality score to do

1:20:29anything like if any of your cold

1:20:33traffic ads don't have it but if you're

1:20:35not getting six plus average quality

1:20:37score on your cold traffic ads you need

1:20:40to keep creating new ad angles until you

1:20:44get higher than six you ain't going

1:20:46anywhere unless you have six if you're

1:20:49below six even if you're at five do

1:20:52something about it

1:20:53it's not the audience it's not your

1:20:57niche it's not Facebook it's just your

1:21:01ad your ad sucks if you're getting less

1:21:05than six your ad sucks and it sucks

1:21:08because you wrote it and you didn't do a

1:21:11good job of it so you need to keep

1:21:13working on it you need to practice you

1:21:15need to test new things and you need to

1:21:17make it better because that ain't gonna

1:21:20perform well at all

1:21:21forget scaling it's just not even going

1:21:23to perform before we scale if you're

1:21:26below six and if you really want to

1:21:29scale properly you need at least a nine

1:21:32to ten quality score every single one of

1:21:35our main ads that we have like our

1:21:38killer ads that we

1:21:39scale-out they've got tens tens across

1:21:42the board it's rare if we have a nine

1:21:45now that doesn't mean that we're so

1:21:47smart and everything that we just create

1:21:4910 out of 10 quality score ads most of

1:21:52the ads we create don't get that score

1:21:56most of the ads we create don't work and

1:22:00the way we've got an you know a

1:22:04collection of ten out of ten quality

1:22:05score ads is by creating lots of them

1:22:08and testing and doing the hard work so

1:22:10that's what you have to do so if you

1:22:12don't have a 9 to 10 quality score on

1:22:14cold traffic keep sandboxing new angles

1:22:16keep trying new images until you get one

1:22:20because you ain't going to be able to

1:22:22scale to the moon and this threat 9 or

1:22:2310 quality score on cold traffic now how

1:22:27we do it so once we've got a winning ad

1:22:31set running at 5 to 15 times cost per

1:22:33lead triplicated again and change the

1:22:36headline so this is actually not how we

1:22:40do it

1:22:41sorry it's actually just that so you

1:22:48know creating you and engels it's as

1:22:52simple as what I showed you in it the

1:22:56the daily management daily workflow our

1:23:00module we're just coming up with a new

1:23:02angle which is new body copy new

1:23:04headline new images and then we're

1:23:08testing it with for image variations to

1:23:11our audiences to see if it works and

1:23:13we're doing this in the sandbox campaign

1:23:15all the time and we're just creating new

1:23:18angles and you know how to do that

1:23:19because you created the original five

1:23:21you just are doing that process again

1:23:22and so that's how you do it and when you

1:23:26find one that works in sandbox duplicate

1:23:30it into production set the budget at

1:23:32five to 15 times KPI let it run if it

1:23:35works droop it and create variations on

1:23:38that and get them to five 15 times keep

1:23:41your and then create new variations and

1:23:42then you can go back to sandbox and then

1:23:46you can look for new angles you see how

1:23:50this works

1:23:51now let's talk about method six

1:23:54targeting additional countries so where

1:23:58things get interesting

1:24:00then these numbers so what it is so we

1:24:08identify countries similar to the

1:24:10countries we're succeeding in and test

1:24:13them to widen our audience now why do we

1:24:17do it because we can't exceed this damn

1:24:21rule that I've put wrong and almost all

1:24:25the slides apologies for that because we

1:24:29can't exceed $10 per 10,000 people in an

1:24:31audience our audience size dictates our

1:24:34threshold spend now by identifying

1:24:36additional countries we were able to

1:24:39expand our audience and thresholds meant

1:24:42now why countries so once we've deployed

1:24:46multiple scaling methods within a given

1:24:48country like let's say we've tried you

1:24:51know let's say we've found lots of

1:24:53audience interests and then let's say

1:24:56we've got three angles working and we've

1:24:58got a bunch of audience interests

1:24:59working and then we've scaled to five to

1:25:0215 times KPI with those angles in those

1:25:05audience interests and then let's say

1:25:08we've even created some ad variations to

1:25:11get more reach into those audience

1:25:13interests and then let's say we've also

1:25:15tried some look-alikes well now we've

1:25:19we've really tried a lot and we've

1:25:22really got now tentacles deep into this

1:25:25country we've we've really gotten and

1:25:28we've gotten into the bloodstream

1:25:30alright now once we've gotten to this

1:25:34point it gets harder to extract more

1:25:36scale from that country and at this

1:25:40point we should take our initial country

1:25:42to the absolute limits so you should

1:25:47never touch another country until you

1:25:49have taken the existing country to the

1:25:50limit just like you shouldn't bother you

1:25:53know you shouldn't bother going to

1:25:55look-alikes until you've taken audience

1:25:57interests to the limits you shouldn't

1:25:59worry about creating at different angles

1:26:01until you have taken those angles to the

1:26:03limits you shouldn't worry

1:26:05about duplicating and sets until you've

1:26:06taken those to the limits you want to

1:26:08take the initial thing to the absolute

1:26:10limit before you add in another another

1:26:12thing because you don't want lots of

1:26:14stuff you want performance and you want

1:26:18to make sure that the only time you add

1:26:19additional stuff is when you can't get

1:26:21more performance out of the stuff you've

1:26:23got so the only option is to add

1:26:25something else on and so once you've

1:26:29taken their initial country to the

1:26:31limits then seek new territory how do we

1:26:35do it

1:26:36so once you've scaled to the limit in

1:26:39your initial country start testing

1:26:41countries most similar to it so you know

1:26:45this is just like everything you can see

1:26:46that we do here is just derivatives and

1:26:48infinity in in all this so once we find

1:26:51an audience that you know we start with

1:26:54our Tower of Babel and then we look at

1:26:55audiences that have affinity to Babel

1:26:57and then we go and test them and then if

1:26:59those work then we find audiences with

1:27:01affinity to those and then if those work

1:27:03then we find audiences with affinity to

1:27:05those and then if an angle works then we

1:27:07might create another angle that's

1:27:08similar to that and then if an emit a

1:27:11certain type of image works and we might

1:27:12get another image that's similar to that

1:27:14and so we're constantly looking at what

1:27:16works and we're finding something that's

1:27:17similar to that and we're trying that -

1:27:18this is what we do and you know

1:27:21look-alike is letting the algorithm find

1:27:25something that's similar to what we've

1:27:27got and letting that go and then what

1:27:31we're trying to do - is find countries

1:27:35that are similar because this is a way

1:27:38to scale out in countries share affinity

1:27:41with each other by the way it's

1:27:42fascinating they share affinity just

1:27:44like everything else in the world the

1:27:46universe and so how we do it is once

1:27:51you've scaled to the limit we start

1:27:53testing countries most similar to it and

1:27:55you triplicate your best performing add

1:27:58sets and you simply change the country

1:28:00to the new country for testing and we

1:28:03use the same campaign so if you've got

1:28:07let's say you've got ads working let's

1:28:10say you've got an ad set working and

1:28:12you've got an angle they say I've got an

1:28:15angle and a lookalike working

1:28:18really well it's a it's a 1% look-alike

1:28:21working really well with your best

1:28:23performing angle and it's in the United

1:28:25States then what you'd want to do is

1:28:29duplicate that ad said leave the angle

1:28:33the same leave except you're gonna want

1:28:37to create another look-alike based on

1:28:40the same thing but in another country in

1:28:45a country with affinities are the one

1:28:47you've got that you've started with and

1:28:50we do this in the same campaign we'd

1:28:52probably just pull this off in

1:28:53production campaign it doesn't really

1:28:56need to go to sandbox because it's got

1:28:58closer Finity to something that's

1:28:59working it has high propensity to work

1:29:03now to to think about what countries you

1:29:07want to target you want to look for

1:29:08clues and closer everywhere if you know

1:29:11where to look and if you're targeting

1:29:13the US only the United States only but

1:29:16you seem to be getting the odd customer

1:29:18from Canada in the odd one from

1:29:19Australia maybe from people just finding

1:29:22you organically or social or email

1:29:24broadcasts or as a friend of a friend

1:29:27told someone else right if you're

1:29:28getting customers from other countries

1:29:31that you're not targeting with Facebook

1:29:33that is a clue and if you have existing

1:29:37clients that are not in the US that is a

1:29:40clue and this isn't just for the US if

1:29:42you're in France and you're advertising

1:29:44in France and you're getting the old

1:29:45customer in the US even though you're

1:29:47not running ads into the US that's a

1:29:49clue these are what Clues look like and

1:29:53that is where you can get ideas for

1:29:55audiences I mean ideas for countries so

1:29:59when I was targeting you know New

1:30:01Zealand then I noticed back in the day I

1:30:04noticed that's on the edge the edge

1:30:06cases would get like maybe Australia or

1:30:08something I started doing that then I

1:30:10noticed some edge cases we'd get you

1:30:12know the United States then I targeted

1:30:14that then I ended up crushing it in the

1:30:16u.s. so then I moved to the US and then

1:30:18I started going out into Europe and all

1:30:20these other things and now you know we

1:30:22we're everywhere and so this is how you

1:30:26do it and I just was paying attention to

1:30:27the clues

1:30:29another powerful tool you can use that

1:30:32we've you know had a lot of work we've

1:30:35had a lot of success with is something

1:30:38called cultural clusters and just as

1:30:41audiences have affinity countries have

1:30:44affinity to and we use cultural clusters

1:30:47to see this and this is what cultural

1:30:50clusters look like and their ways to

1:30:53cluster together different countries by

1:30:55their culture and their heuristics

1:30:59biases tendencies and just the way they

1:31:02think and the way they believe and

1:31:05perceived the world and so over here

1:31:10we've got the egalitarian group which is

1:31:14like the Western world and the the

1:31:17Western world likes empowerment and

1:31:20decentralization at a very high level

1:31:22but then we can break it down even

1:31:25further the Western world we can go into

1:31:26the group that likes competition and

1:31:29contests a lot

1:31:31now that's United States you know the

1:31:33Kingdom Ireland New Zealand Australia in

1:31:38Canada right so I'm from New Zealand it

1:31:43was very easy for me to work into

1:31:45Australia because it was just like the

1:31:47same and I was shocked at how easy it

1:31:49was to just work in America - and I

1:31:52always wondered why but you know when I

1:31:54when I found this I really understood

1:31:55why because although New Zealand's a

1:31:58tiny little country ages away from

1:32:00America it has a lot in common with it

1:32:02because our culture is similar and it's

1:32:05because you know our culture's sheer

1:32:08competition we like competition we like

1:32:10autonomy we like decentralization we

1:32:13like risk-taking results ambition and

1:32:15innovation that's like the American

1:32:17dream the Andrew Carnegie that

1:32:19Rockefeller story the Elon Musk the you

1:32:22know the you know all of those stories

1:32:24that's like the American Way and that's

1:32:26also how these other countries believe -

1:32:29and so chances are if you've got ads

1:32:35that work in any one of these countries

1:32:37they should work in all of these

1:32:39countries so if you're got ads that work

1:32:43in Australia

1:32:43you should try targeting one of the

1:32:45other ones if you've got ads that work

1:32:47on candidate you should try targeting

1:32:48all of these other ones they are

1:32:49basically the same damn thing they all

1:32:51speak English they all think the same

1:32:54they all see the world the same way and

1:32:56it all pretty much is the same then

1:33:00we've got network which is Sweden

1:33:05Netherlands Norway Finland this is like

1:33:07getting into Europe and these people

1:33:10these are more westernized European

1:33:13countries so Sweden Norway and in

1:33:17Denmark and Germany and Switzerland and

1:33:22these are more westernized European

1:33:25countries compared to the other ones and

1:33:27what I mean by that is they're just

1:33:29they're more like Americans than the

1:33:33other European countries and they like

1:33:37decentralization and empowerment they

1:33:39also like the main things is

1:33:42decentralization risk-taking empowerment

1:33:45and and all of this that's what these

1:33:49ones like then we get into the

1:33:51hierarchical group which they like

1:33:53centralization in hierarchy and rules

1:33:56and so that's when we get into France

1:33:58and Belgium and Italy and Poland in

1:34:01Spain so these guys like hierarchy rules

1:34:04centralization formalism and all the

1:34:06stuff and then we also get into these

1:34:08other countries over here too and so

1:34:11wherever you're running ads that are

1:34:14successful chances are you can run the

1:34:17same ads and make them successful in

1:34:18everything else within this cultural

1:34:20cluster it's easy to scale to all the

1:34:25countries within a cultural cluster it

1:34:27gets hard to scale to countries not end

1:34:31with not within the cultural cluster

1:34:33your audience because the same ad angle

1:34:36won't necessarily work because their

1:34:39belief systems are different try running

1:34:42a capitalist you know look at me like in

1:34:47how much money I make and how successful

1:34:50I am add into France

1:34:52watch what happens you

1:34:56just it's not all running into run it

1:34:58into Thailand or something just watch

1:35:01what happens you know the the cultures

1:35:04change between these different clusters

1:35:06but within these clusters they pretty

1:35:08much remain the same so it's easy to

1:35:10scale within it it's harder to scale

1:35:12into a new one quite often to scale into

1:35:16a new one requires new angles and in a

1:35:20bit of work so I recommend scaling

1:35:23within the one you're already in first

1:35:25take over all of this once you've taken

1:35:28all of that over if you're hungry for

1:35:30more have a crack at going into these

1:35:32other ones go to the one most like the

1:35:34one you're in so if you're in this one

1:35:36then you then you want to take over

1:35:37everything in here once you've done that

1:35:39then you want to go to this one you

1:35:41shouldn't start here take over all of us

1:35:43and then decide to come over here and

1:35:44try that because this is far away from

1:35:47that start here go through that then go

1:35:50to here work through that then got here

1:35:52work through that then go to here work

1:35:53through that then go to here work

1:35:54through that didn't go to here work

1:35:56through that got it good method seven

1:36:01scaling retargeting campaigns so what it

1:36:06is we use different campaign objectives

1:36:08to increase reach and performance of our

1:36:11retargeting campaigns why do we do it

1:36:13well most of our scaling efforts go

1:36:15towards cold traffic campaigns because

1:36:17this is the entry point in the most

1:36:19important thing however as we scale cold

1:36:23traffic naturally our targeting audience

1:36:27size increases and when it does this it

1:36:30gives us more options now how do we do

1:36:33it how do we scale returning campaigns

1:36:35so once you've taken your warm

1:36:37retargeting campaign to the absolute

1:36:39limit duplicate the best performing ad

1:36:42set into a new warm retargeting campaign

1:36:45with the page post engagement subjective

1:36:48now first and foremost never create

1:36:52something new until you've taken the

1:36:53existing to the limit

1:36:54so what we want to do first of all is we

1:36:59should have our warm retargeting

1:37:01campaign

1:37:06here it is and we should be bidding for

1:37:10we should be optimizing for conversions

1:37:11using the auto bid method and in here we

1:37:14should be targeting you know page post

1:37:18engagements 180 days all website

1:37:21visitors 180 days we should be targeting

1:37:23all of these people so people who have

1:37:25engaged with us within the last 180 days

1:37:27and we're optimizing for conversions and

1:37:30we're using auto bird and the angles

1:37:32we're using in here are the best

1:37:34performing angles from our cold traffic

1:37:36campaign now what we want to do is we

1:37:40want to increase the budget of these as

1:37:46much as possible

1:37:50like we want to increase the budget of

1:37:52these as much as possible and take them

1:37:54to their absolute limits before we go

1:37:57and create something new

1:37:58so if our retargeting audience size is

1:38:02like this say it's ten thousand well we

1:38:07don't need to use the same ten dollars

1:38:09per ten thousand dollar rule I put out

1:38:13for ten thousand people rule for

1:38:15retargeting we can go about five times

1:38:17that so we could if we've gotten

1:38:19retargeting audience warm of ten

1:38:22thousand people we can spend probably

1:38:23fifty bucks so you know we we take that

1:38:27up to its limit and it's set 50 bucks if

1:38:29it's still working there good that means

1:38:32it's pretty much at its limit we're

1:38:33spending at the threshold of the

1:38:36audience size once that is true once

1:38:41you've done that and you have to do that

1:38:42first before you do this other thing

1:38:44otherwise you're just doing stupid stuff

1:38:46then we want to scale further by

1:38:49creating a new campaign with a different

1:38:53objective so what we do here is we start

1:38:57by going to campaigns and then we want

1:39:02to go create and then we want to call

1:39:09this one warm retargeting PE PE which is

1:39:14like page boost engagements OTO

1:39:18we want to go auction and our campaign

1:39:22objective is going to be post engagement

1:39:29and then that we're going to create an

1:39:33ad set we're not going to create an ad

1:39:35set we're just going to go skip skip

1:39:37save to draft and then we're going to

1:39:43publish that up well so now we've got to

1:39:48warm our targeting audience o campaigns

1:39:50one is one retargeting conversions Auto

1:39:52the other is warm retargeting petrus

1:39:53engagements we'll do then we come in

1:39:55here we find our winning ad sit and then

1:39:59we go to placate and then we go to an

1:40:02existing campaign and we select our warm

1:40:06retargeting P P P P P e campaign this

1:40:08time and then we troupe it into that and

1:40:12then what we do is we go down and we

1:40:22want to make sure our audiences see it

1:40:24correctly which will be you know people

1:40:26who have it should just be the same

1:40:28which is people who have engaged with

1:40:32your fan page or people who have engaged

1:40:36with one of your ads in the past 180

1:40:38days or visited your website in the past

1:40:39180 days or visited your your landing

1:40:41page in the past 180 days you're

1:40:43excluding customers your location is

1:40:46going to be the same scene with age and

1:40:47things this isn't going to have an

1:40:49audience interest because it's a

1:40:51retargeting campaign and then placements

1:40:54you basically want to go like all

1:40:59placements you want to just try all

1:41:03placements for this and then what you

1:41:07want to do is all mobile devices and

1:41:10then optimization for air delivery you

1:41:13just want to go post engagement and then

1:41:16that's it then you can just publish it

1:41:18up boom but what about budget how do we

1:41:21set the budget so here what we do is we

1:41:28set our budget

1:41:32at 40% of what the conversion campaign

1:41:37budget is at so if I have 10,000 people

1:41:40in my warm retargeting audience and my

1:41:45warm retargeting campaign which is

1:41:47optimized for conversions that's going

1:41:51to be spending $50 per 10,000 because

1:41:54there's the threshold limit the rules

1:41:56change a bit when we get into warmer

1:41:58traffic so if I'm spending $50 per day

1:42:02into this campaign then I want to find

1:42:0740 percent of 50 and that's going to be

1:42:13like was that 20 bucks or something so

1:42:16I'm gonna set this at like 20 bucks so

1:42:19that's what you want to do you want to

1:42:22take two trajectories through the warm

1:42:26retargeting audience the first and it

1:42:29always comes first is the warm

1:42:32retargeting conversions although using

1:42:36your two best angles from your cold

1:42:39traffic campaign to the 180-day audience

1:42:43using a $50 per 10,000 people in the

1:42:47audience budget to hit that maximum

1:42:51point of efficiency before entropy sits

1:42:53in once we achieved that we can't go

1:42:57higher on those budgets because we'll

1:42:59get entropy so there's another way to

1:43:01get in and it's a new campaign that's

1:43:05for page post engagements using the same

1:43:08angle to the same audience at 40% budget

1:43:11of the conversions don't worry about how

1:43:16hard all that was to figure out you just

1:43:17get it told to you and easy as that

1:43:21that's how you do it now

1:43:27hot retargeting

1:43:32so the above strategy and the strategy

1:43:33we've covered here is only used for warm

1:43:36retargeting campaigns not hot

1:43:38retargeting for hot retargeting we stick

1:43:41to our conversions objective at the

1:43:44campaign level and daily unique reach at

1:43:47the ad set level with a max bid of a

1:43:50hundred and fifty dollars per 1000 CPM

1:43:54and to scale hot retargeting keep

1:43:57increasing the spend and always taking

1:44:00it to its limits and as the audience

1:44:02size increases increase spend and you

1:44:05can try testing new images and new

1:44:07angles but it's quite hard to beat the

1:44:11ink a simple angle here so images can

1:44:13help sometimes but really that - I mean

1:44:16this isn't one where you have to test a

1:44:18bunch of angles and images really what

1:44:20you're doing is you've just got to keep

1:44:22increasing that that's Bend until you

1:44:25find its limit where it hits entropy

1:44:27that's where you keep the spend but as

1:44:30you scale your cold traffic in your

1:44:32warmer targeting you're naturally

1:44:33increase in the audience size of your

1:44:35hot retargeting and as that increases

1:44:37gradually you can keep increasing the

1:44:39spend here and that's really all you

1:44:42need to do to increase in scale hot

1:44:44retargeting is as cold traffic increases

1:44:48warm traffic increases when warm traffic

1:44:51increases increase the bid there when

1:44:54all of this happens hot increases when

1:44:56hot increases increase the bid debt and

1:44:58you just keep working through these

1:44:59three three layers and the one that

1:45:02starts the whole domino effect is cold

1:45:04so cold where most of your focus goes

1:45:06once you've had a bump there you go to

1:45:08warm bump it there go to hot bump it

1:45:11there come back to cold but look but it

1:45:13just land and you can also try and

1:45:19creasing max bid - right you can try

1:45:21going 300 bucks per thousand seat views

1:45:24you're just trying to you're just trying

1:45:26to go as hard as you possibly can with

1:45:28hot retargeting now you're not going to

1:45:30go harder and hot retargeting by

1:45:32creating more hot retargeting campaigns

1:45:34bad idea you're not going to go harder

1:45:36and hot retargeting by creating lots of

1:45:38aired angles

1:45:39idea lots of duplications bad idea and

1:45:42you know trying to create granular

1:45:45audiences bad idea trying to create a

1:45:48page post engagement for talking bad

1:45:50idea trying to create a conversion

1:45:53objective campaign and Adsit for talking

1:45:57a bad idea just stick to this thing it

1:46:01is the best you've just got to keep

1:46:02increasing spend now let's talk about

1:46:05method 8 decentralized architecture so

1:46:11what it is

1:46:12so we structural at accounts campaigns

1:46:14ads hits and ads in a way that

1:46:17decentralizes spend and load balances it

1:46:20evenly across the infrastructure and

1:46:23this allows us to scale our spend by

1:46:26adding additional clusters not

1:46:29increasing the size of existing clusters

1:46:33why we do it so the biggest enemy of

1:46:37scale is entropy and at a certain load

1:46:40disorder and diminishing returns set in

1:46:43and to avoid this we architect our ad

1:46:47accounts so that the spend is evenly

1:46:50distributed at safe levels without

1:46:53entropy not at little levels we

1:46:57want to take them to the levels right on

1:46:59the knife's edge of entropy because

1:47:03that's where things work their best not

1:47:06in really safe zones but in almost

1:47:09danger zones and so we're trying to get

1:47:12everything on that knife's edge not over

1:47:15it not too far before it on it

1:47:19everywhere that's where the magic

1:47:23happens and initially we scale using one

1:47:28ad account and one cold traffic campaign

1:47:31with all your audience and angle

1:47:34variation happening at the ad set level

1:47:37within that one campaign with a net one

1:47:40ad account and what we're trying to do

1:47:43first initially as we're trying to scale

1:47:46all of our winning angles to five to

1:47:49fifteen times cost per lead that's our

1:47:51initial thing

1:47:52we're testing five angles with four

1:47:55images each and 30 audience interests

1:47:57we're finding the winners and when we

1:48:00find those winners we droop them and set

1:48:01the budgets to five fifteen times KPI

1:48:04once we get them there

1:48:05Carlos's keep the winners that's our

1:48:09first round of optimization and scale

1:48:12from there we find those winners and we

1:48:16create aired variations on those angles

1:48:18and then we scale those to five fifteen

1:48:20times KPI and then we do another

1:48:24variation maybe the first variation we

1:48:27do we change the button the next one the

1:48:29headline the next one an image and then

1:48:33we keep going until we reach the limits

1:48:36of our audiences which is not exceeding

1:48:39ten dollars per 10,000 people that must

1:48:45have just copied pasted that same damn

1:48:47thing in there that's why it's identical

1:48:49in every slide so not exceeding ten

1:48:52dollars per 10,000 people and so we're

1:48:56trying to get right up to that knife's

1:48:58edge but entropy exists everywhere not

1:49:03just in the spend so you know we see it

1:49:08occur at the ad set level when we spend

1:49:12less than two times KPI or if we spend

1:49:17more than 15 times KPI we see entropy

1:49:20happen at the ad sea level but we also

1:49:23it doesn't just occur at the ad sea

1:49:26level we see entropy occur at the

1:49:29campaign level at about a thousand

1:49:32dollars a day so once you have let's say

1:49:34like you know seven ad sets that are

1:49:39running at 15 times KPI at ten dollars

1:49:41each you know that you're getting let's

1:49:44say or a bit over a thousand a day in

1:49:46spend then you're going to start to see

1:49:49entropy occur at that campaign level and

1:49:52just like an ad set likes to be in

1:49:55between 5 and 15 times KPI a campaign

1:49:58likes to be around a thousand bucks a

1:50:00day

1:50:02and that's where it likes to be and at

1:50:06the Add Account level it kind of likes

1:50:08to be at about five thousand dollars a

1:50:11day and over there it starts to see it

1:50:14but it entropy and so you know it

1:50:18entropy opens at multiple dimensions now

1:50:21a lot of you aren't going to be anywhere

1:50:22near $1,000 a day for a while so you're

1:50:27good with one cold traffic campaign and

1:50:29one ant account all of your all of your

1:50:32variation is going to happen at the ad

1:50:34set level within one called traffic

1:50:36campaign within one ad account simple

1:50:40once you get to a thousand dollars a day

1:50:43in total spend then you want to create

1:50:46another campaign and then when we create

1:50:53that other campaign how do we how do we

1:50:55architect it so once you've hit a

1:50:59thousand dollars a day in total spend

1:51:02for a particular ad for a particular

1:51:07campaign or your cold traffic campaign

1:51:08then you want to create different cold

1:51:11traffic campaigns based on the angle so

1:51:15if you've got one angle that's 26 year

1:51:17old punk and you've got another angle

1:51:19which is advice for consultants and

1:51:21they're both running within one cold

1:51:24traffic campaign the moment we start

1:51:26hitting a thousand dollars total spend

1:51:28per day with that one cold traffic

1:51:30campaign split across those two angles I

1:51:35now want to create a separate campaign

1:51:40for in separate both at angles so I'm

1:51:43gonna have a cold traffic 26-year old

1:51:45punk campaign and I'm gonna have a cold

1:51:48traffic advice for consultants campaign

1:51:51and within those two cold traffic

1:51:53campaigns you are only going to find

1:51:55that angle or variations of or add

1:52:00variations of that angle going to all

1:52:03different audiences and budgets of 5 to

1:52:0615 times KPI right but then the moment

1:52:09we start let's say we then have five

1:52:12aired angles

1:52:15but five different ad angles all of

1:52:18$1,000 a day in spend for cold traffic

1:52:21and within those campaigns five

1:52:25campaigns they consist of seven ad sets

1:52:29at 15 times KPI at $10 cost per lead so

1:52:32like you know they're all at about a

1:52:33thousand so the ad set level is that

1:52:36it's efficient it's at its knife's edge

1:52:38efficiency we can't exceed that the

1:52:41campaign level is at its knife's edge

1:52:44efficiency and now we've run out of room

1:52:47in their ad account that set its knife's

1:52:49edge efficiency so only at this stage

1:52:52that we have five angles with within

1:52:59there we have you know seven ad sets at

1:53:0215 times KPI in the we're spending

1:53:05$1,000 a day and cold traffic across

1:53:07five angles only at this point now do we

1:53:12consider creating a new Facebook ad

1:53:14account to start load balancing across

1:53:18another dimension and this is what I

1:53:21mean by decentralized architecture we

1:53:24build based on the parameters in the

1:53:28thresholds of the system and lucky for

1:53:32you I already know what all of those are

1:53:34and so and then I just tell you how to

1:53:36architect it right it saves you maybe

1:53:38five years so congratulations and 75

1:53:41years and millions of dollars in two

1:53:44minutes so you know not to take ads hits

1:53:47pass 15 times keep you own you know not

1:53:50to take spend past ten dollars per

1:53:5410,000 on cold you know not to take

1:53:56spend past $50.00 per 10,000 our warm

1:53:58you know you can take spend on hot

1:54:01retargeting as high as you want until

1:54:02you start seeing entropy yourself and

1:54:04then you know to not have a campaign

1:54:09exceed about $1,000 a day before you

1:54:12have to start splitting it out and then

1:54:14you know not to have an ad account

1:54:15really exceeding five grand a day before

1:54:17you have to start splitting it out this

1:54:19is the architecture and so what you end

1:54:22up happening is when you get up to

1:54:24massive scale you end up with multiple

1:54:26ad accounts and how you separate ad

1:54:29account

1:54:29is by the traffic type so I can show you

1:54:36Al's like we have we've got all sorts of

1:54:43different ad accounts and you can see we

1:54:46just call them concerning comm line all

1:54:47these different things

1:54:48so you know if I find the consoling comm

1:54:52number one account where is this little

1:54:55thing so here it looks like we use this

1:55:04for warm and some hot so this is like a

1:55:08warm retargeting campaign right and then

1:55:12let's say number two is we use this one

1:55:16for this looks like it's for something

1:55:22else we're doing that is slightly more

1:55:25advanced don't worry about that one now

1:55:29it's fine another one we're using that

1:55:31number three so we've got a lot going on

1:55:35we've got like teen ad accounts going

1:55:37because you know we and we justify that

1:55:40because we've we're pretty much spinning

1:55:415k and all of them so we load balancing

1:55:44it all out with the ideal architecture

1:55:46and then we've got one for running ads

1:55:49for uplevel specifically so we're

1:55:51starting to you know to load balance it

1:55:54out that way too and so basically what

1:55:58we're doing is splitting things out and

1:56:02what we want to do on the ad set level

1:56:05is we want to have one ad account which

1:56:09might be used for coal traffic one ad

1:56:12account which will be used for warmer

1:56:15targeting in hot retargeting together at

1:56:19first when we go from one ad account to

1:56:21to add accounts we want to have one for

1:56:22cold traffic one for retargeting the one

1:56:25for retargeting includes both warm and

1:56:27hot the one for cold traffic includes

1:56:29the production cold traffic and the

1:56:32sandbox and cold traffic and then once

1:56:36you're spending more than $1,000 a day

1:56:38in a campaign create new cold traffic

1:56:41campaigns

1:56:42angle within those cold traffic

1:56:46campaigns per angle you have ad sets

1:56:49which have different audiences in them

1:56:51and the same angle or variations of that

1:56:55angle and those ad sets should not

1:56:58exceed 15 times KPI or the rule of ten

1:57:02dollars per 10,000 people in that

1:57:04audience and if you do that you're good

1:57:07and then once you're spending five grand

1:57:10a day on that cold traffic campaign then

1:57:13you probably want to create another

1:57:14campaign called cold traffic number two

1:57:17and then you can have you know more

1:57:19stuff in there and you keep scaling out

1:57:22that way

1:57:22this is what I mean by horizontal

1:57:24instead of vertical if we were doing

1:57:26vertical scaling vertical centralized

1:57:28scaling would just keep one ad account

1:57:31and we just keep increasing the spend

1:57:34and we would just not care about the

1:57:36entropy and just start losing money

1:57:38right this is centralized hores our

1:57:43vertical scaling decentralized

1:57:45horizontal scaling is when we go out not

1:57:48up we achieved the same thing but we go

1:57:51out instead of up so we go we spread

1:57:54across the dimensions we go into

1:57:56different countries we go into different

1:57:59ad sets different variations different

1:58:03bidding strategies we go into different

1:58:05audience types look-alikes and things

1:58:07and then you know we spread across the

1:58:10different countries and then we spread

1:58:11across different campaign structures and

1:58:14different architecture types of the

1:58:16system and then we spread into different

1:58:18ad accounts and we just start building a

1:58:22serious machine a seriously intricate

1:58:25and beautifully designed machine for

1:58:30serious firestorm of ads and yeah they

1:58:37should keep you occupied until you're

1:58:38making like 30 million dollars here now

1:58:43I know we've covered a lot of ground

1:58:45some of your brains are probably smoking

1:58:47right now that's alright because I've

1:58:50included a cheat sheet available for

1:58:53download beneath this video in the

1:58:54resources section and it's called nain

1:58:55scaling

1:58:56methods and in this cheat sheet you can

1:58:58see all nine scaling methods their pros

1:59:00cons and how to execute them using this

1:59:04cheat sheet so you know it makes life

1:59:07easier for you so this is what it looks

1:59:10like and it's just a summary of all of

1:59:12these methods so the scaling method

1:59:15number one what it is while we do it how

1:59:18to do it the parameters for all of our

1:59:21different scaling methods you know I

1:59:23tell you what it is how we do it and the

1:59:27parameters and everything so I recommend

1:59:29you download this it's in the resources

1:59:31section beneath this video download it

1:59:34keep this handy and you can use it as

1:59:36your cheat sheet for deploying these

1:59:38scaling methods because there's a lot of

1:59:42stuff in here this will keep you

1:59:44occupied for years and these scaling

1:59:47methods are enough to take you to forty

1:59:50thousand dollars a day in ad spend and

1:59:53they're enough to grow your business to

1:59:56thirty million dollars per year and this

1:59:59isn't an opinion of mine nor is it what

2:00:02I think I know because I actually did it

2:00:08so this is true I make this claim

2:00:12because I grew my business this level I

2:00:15know I can and I we spend that much on

2:00:19ads so this is actually the level that

2:00:23these scaling methods and these

2:00:26strategies in this architecture can take

2:00:28you to I believe it can actually take

2:00:31you further than this but I'm not

2:00:33comfortable promising that because I

2:00:34just haven't done it myself the moment I

2:00:36do I will change my promise there is no

2:00:40limits to it people used to tell me oh

2:00:42you can't make like a million dollars a

2:00:44month and info you can't make a hundred

2:00:47grand a month and consulting oh you

2:00:49can't spend more than a thousand dollars

2:00:51a day on ads or you can't run a webinar

2:00:53all the time

2:00:54or you can't run an ad that works for

2:00:56longer than a week or you can't you know

2:00:59do this or ads don't work and Europe or

2:01:02ads don't work in South America there's

2:01:05no way you'll get ads to work in Nigeria

2:01:07I just hear all sorts crap and

2:01:10all of it I've found to be wrong through

2:01:13experimentation so you know in the space

2:01:16what training you probably heard and

2:01:17seen a lot of things that go against a

2:01:20lot of things that people have been

2:01:21telling you that's fine the only reason

2:01:24I tell you anything in here is because

2:01:26it is grounded in experimentation in

2:01:29practice

2:01:30nothing is theory here all is grounded

2:01:34in experienced and documented improving

2:01:37over time in practice

2:01:40and armed with these methods you will

2:01:43decimate the competition on Facebook and

2:01:47I know that one too because I've been

2:01:48doing it for a long time and so with

2:01:51these methods you can seriously take

2:01:54your business to the moon and unleash a

2:01:57firestorm of advertising all over the

2:02:00planet so congratulations on going

2:02:04through this Facebook Ads training it's

2:02:06been pretty intense we've covered a lot

2:02:08of stuff in this scaling video really

2:02:12just adds the you know the cherry on top

2:02:15with these methods you can really scale

2:02:17anything to massive levels so thanks for

2:02:23watching this training and I look

2:02:24forward to seeing you in the next one

2:02:26soon

2:02:28[Music]

2:02:37you