Genius Pixel Pt. 2: How To Scale FB Ad Campaigns

by Justin Brooke

Hey how's things going my friend?

Been a while ain't it…

We've got a lofty goal of growing AdSkills to $10mil now that we've passed the $1mil mark. With that I've had to make some hires, build some systems, and still live this travel life.

If I get any interest, I'll write to you about these systems and hires. Not sure if y'all are interested in those kinds of things.

We're on the road to 8 figures and then from there we'll be looking for an investor or buyer who can take it to 9 figures.

My goal is to do all that by 2021 too so I've really gotta focus like never before.

Part of that is moving this newsletter to one bad ass email per week.

I've got a new guy on the team who is going to hold my ass too that. He's even pushing for delivery on the same day and time everyday so y'all know exactly when to expect it.

Anywho, I'm in Thousand Islands, NY now so finally back on the East Coast. I love y'all West Coasters but it just ain't for me.

Thousand Islands is where I got married and it's a real special place for me. I'll be here until October and plan to do the bulk of our $10mil growth from here.

But enough about me, what I wanted to talk to you about today is overcoming that dreaded CPA mountain climb when you raise your FB ad budget.

You know what I'm talking about…

Everything is going great at $25/day and then you push it to $100/day or higher and your cost per conversion goes from awesome to holy shit turn it off!

Let's get into how to avoid that holy shit turn it off part shall we.

You've got a new but tiny ad campaign that's working — NOW WHAT?

You try to feed it more money, but things go haywire.

Or everything was going fine and then after a week or two it just started getting worse even though you haven't changed anything.

Your cost per conversion goes up, instead of just getting more conversions for the same previous cost?

This is a common problem.

Good news is, there is a solution.

There is a way to scale your FB ad campaign from $25/day to $2,500+ per day without raising your cost per conversion.

In my last article we named the new campaign building process the genius pixel because you have to train the new pixel to become a genius at knowing your customer. This time we are naming the campaign scaling process the hungry algorithm because you have to feed it lots of food to grow a big campaign.

Also just like in my last article, this process fits for any ad network that uses pixels and algorithms. Definitely for Facebook AND Google. However, for sake of writing I'm going to mostly refer to FB as I write.

Know that I mean FB AND GOOGS.

A genius pixel needs you to feed it clean structured data.
 
A hungry algorithm needs you to feed it with more creative, more placements, and bigger audiences so it has the resources to grow a big strong campaign.


Where we were trying to keep things very narrow and specific while training the pixel, now we need to take our genius pixel off of the leash so it can work it's magic.

By magic I mean discover new customers for us at a greater volume, but not a greater cost. That sounds like the kind of magic I like!

How To Feed a Hungry Algorithm…

  • Dive Into Your Data For Success Clues
  • Create an Army of Creative Assets To Avoid Ad Fatigue
  • Consolidate Small Audiences Into Super Audiences
  • Setup Automation Rules To Prevent Losses


Diving Into Your Data

Success always leaves clues, and these clues can be found in your historical campaign data. However, if you try to dig for clues inside your Business Manager you will only be looking at the surface level.

To dive deep, you need to export your full campaign history to a spreadsheet file and use pivot tables. 

Don't freak out, pivot tables just let you move data around so you can see it in new ways. Pivot means to move and table is just a table of data. 

Here's a great Youtube video that'll turn you into a Google sheets data ninja before your lunch break is over.

For example, inside your business manager you really only see how the ad performed inside it's own ad set. But what you really want to see is how that ad performed across all ad sets by age, location, etc.

Downloading your data and using pivot tables to re-organize the data will give you insights that your competitors are too chicken shit to see. You'll be able to create new experiments, audiences, and creative based on these insights.

This is the first step to scaling, getting a deeper understanding of whats working and why. From here you can do everything else mentioned in this article, creating more creative and larger audiences.

You're going to use this new knowledge to create a giant universe for your hungry algorithm to expand it's search into. And it's going to use that genius pixel you trained to make sure it's finding you the right clickers within the new larger universe.

Making sense so far?

  1. Create new pixel and train it with clean data
  2. After you train the pixel download historical campaign results
  3. Use historical results to create new larger pixel hunting ground


Creating an Assets Army

In the early days of your campaign you wanted to keep everything tight and narrow. Well because of physics you just can't grow something large while also keeping it tight and narrow.

This is why your pants don't fit after Thanksgiving dinner.

Up until now you've probably only created the typical newsfeed style ad. Except there is only so many ads they can fit in the newsfeed before the users riot.

To scale you need to leverage all of the formats the ad network gives you so you can show your ads in more places. Don't be that guy who tries to force fit one ad format into all of the different placements.

We've all seen those ugly square ads in our Instagram or FB stories.

You're just wasting your ad spend if you do this.

Now that you have a winning ad from your small early campaign, leverage and re-purpose that ad into multiple formats. You should create a video version, long (90 seconds) and short (30 seconds). You should create a summarized version for right hand side bar and messenger, plus one specifically for Instagram versus FB newsfeed.

The more ads you create right now the more it is going to pay off in the future. We are building a potential million dollar earner here, so don't become Mr. Skimpy Pants.

The goal is to create plenty of assets that the algorithm can insert at just the right time so it's not trying to use only one ad. Also avoiding the dreaded ad fatigue that comes from your audience seeing the same ad over and over again.

“The Most Important Skill Today's Digital Advertisers Can Learn Is The Ability To Feed Clean Structured Data Back To The Algorithms Inside Ad Networks”

The Most Important Skill Today's Digital Advertisers Can Learn Is The Ability To Feed Clean Structured Data Back To The Algorithms Inside Ad Networks

Click to Tweet

Building Super Sized Audiences

When you start an ad campaign, you should start with lots of very narrow small ad sets. Even as small as one interest target or one keyword per ad set/group.

This is because in the beginning we don't know yet what is going to work.

After a couple weeks and a few hundred dollars spent, we have campaign data that tells us what is working and what is not. Especially if you have done the data deep dive I spoke of earlier.

To scale up you need bigger audiences.

Instead of trying to think like a sniper take one high precision shot, you need to think like an NSA agent using a facial recognition program from a satellite in space.

You've already trained the pixel on what types of faces (customer data) it needs to target, and now you just need to zoom way out so it can find you hundreds if not thousands more.

The way you do this is by consolidating your small singular audiences into one super sized audience. You take all that data you deep dived into and use the good targets to create large audiences of a few million to tens of millions of people.

If a smaller target was working and had an audience size of 500,000 people, you're looking to combine 10 of those to have an audience of 5,000,000 people.

This combined with having more creative assets in all the popular formats will allow your campaigns to expand to the full universe of the ad network – WITHOUT increasing your costs per conversion.

The reason ad networks increase your cost per conversion is because you're trying to show the same ads to the same people.

When you use different ad types and different ad placements and larger audiences, the algorithm doesn't have to work as hard. When you use just a few ads to a small audience the algorithm has to work 4x as hard to get those same people to convert with that same ad.

Using the same ad to the same audience is like trying to find more gold by digging deeper in the same spot in the same direction. When instead you should be digging a wider hole in all directions.

Hence why gold mines are acres and acres wide instead of vertical tunnels.

Combining your audiences into one larger audience and then adding in lots of different creative types is the advertisers version of digging a wide gold mine.

 

Using Automation Rules As a Stop/Loss

Everything we've discussed until now has been about growth.

This last section is about managing risk. It's about creating a type of built in insurance to your ad campaigns so that they can never create massive losses. But at the same time still giving them that large universe to expand into without obstruction.

I'm talking about using automation rules to control your ad campaigns remotely and automagically. In the stock market trading world they call these stop/loss rules.

For example if I buy a stock at $70 per share but I want to make sure I never lose more than $10 per share, I can set a stop/loss rule to sell as soon as it hits $60/share. Ad networks have this same type of functionality these days.

There are many tools that do this and even some built right into your business manager account, however I like RevealBot. You can use whichever tool you like best or the one I recommend, it doesn't really move the needle much between the choices.

What does move the needle is just making sure you are using one of them.

The first automation rule I setup is one to cut my losses before they get too much to bear. You simply set the tool up with these if/then rules…

 

AUTOMATION RULE #1

If my ad has 10,000 or more impressions and my cost per conversion is greater than 3x our desired cost, then turn off this ad.
 
And this is usable in all ad networks that allow automation rules.

The reason we first look for 10,000+ impressions is we want to make sure the ad has had enough of a chance to perform. Otherwise you're cutting adds prematurely.

I also like to use 3x my desired cost as my stopping point because sometimes 2 sales could happen back to back and fix your 2x cost. Very rarely would 3 sales happen back to back, so this is why we cut after 3x cost.

Adjust for your own level of risk tolerance but I would worn you against going too tight. If you do something like 1x or 0.5x you could be cutting a lot of volume. We're not trying to force the rules to make us profitable, we are just trying to reduce our chance of taking a blood bath in profit loss.

 

AUTOMATION RULE #2

If my ad has 10,000 or more impressions and my cost per conversion is equal to or less than our desired cost, then increase ad budget by 30%.

This automation is the exact opposite of the first automation. Instead of managing losses this one is managing profits by growing your profits and increasing your conversion volume.

Just make sure this rule isn't running more than 1x per day or it could get pretty crazy pretty quickly. You don't want it to jump you up 90% in one day and then tomorrow the results are totally different.

Again, we aren't trying to make the automations replace a human campaign manager, we are just using them to assist the campaign manager. So don't try to make the campaign grow to fast or turn off too suddenly, a little automation goes a long way.

 

AUTOMATION RULE #3

If this ad has been running for 72 hours, then pause it and start another ad.

This only goes for ads of the same ad unit type like a newsfeed ad pausing to start another newsfeed ad. Or pausing a story ad to start another story ad.

I like to create 10 variations of the same ad type when I'm at the scaling phase.

With this rule in place and 10 ad variations it means my audience is never seeing the same ad twice in one month. Which REALLY helps with avoiding ad fatigue.

When you combine these rules with the larger audiences and more ad unit types with 10 variations you create the perfect storm scenario for scaling your ad campaigns. Including fail safe measures to insure you don't incur any large losses because you happened to not check your campaign one morning.

If you have any questions and would like myself or one of our pro ad buyers to look over your campaigns, consider joining us inside our paid membership.

We are all very active every day helping each other with ideas, feedback, and recommendations. Just ask yourself how much it'd be worth to you to have 1 ad campaign reach full scale?

 

Are you ready to step up and become an AdSkills Upperclassmen?

INSTANT DOWNLOAD: 50 ONLINE ADS THAT MADE MILLIONS FOR THEIR FOUNDERS

We searched 50 companies that made millions online with digital ads. Then packaged all their landing pages and ads in a neat file for you. Download it now and have an instant swipe file for your own inspiration.