The Genius Pixel: How To Train Ad Networks To Give You A LOT More ROI

Justin Brooke
Jan 3 · 14 min read

Clicks aren’t getting cheaper…

But tactics have evolved to still get a decent return on paid ads, read on to learn how.

__________________________________________________________________TL;DR: By training your ad network pixels to find you exactly the kind of leads or customers you want, you can substantially increase ad ROI. Mainly speaking about Facebook, Instagram, Youtube, and Google ads.
__________________________________________________________________

This year marks the 10th year I’ve been running FB ads and 12th year running Google ads. In that time I’ve managed over $10,000,000 in ad spend

Things have changed, and the future is clearly about “data.”

Which leads me to “The Genius Pixel.”

If you’re in ecom, lead gen, SaaS, affiliate marketing, digital courses, or selling services, and you use Facebook or Google ads, then a “genius pixel” should be your ultimate goal.

10 years ago when we ran FB ads there was no newsfeed ads. There was just tiny right hand sidebar ads, nothing like the ones today. Also, there wasn’t Youtube ads and Instagram wasn’t even around.

Back then Facebook ads didn’t even have a conversion tracking code. You would just use your own tracking link and track sh** for yourself.

When the times change, often the tactics and strategies have to change as well.

Since times have changed, and ad networks have evolved, our tactics have to evolve too.

Back in the day, when starting an ad campaign you chose targeting that was relevant to your audience, and tried to optimize your ads for lowest CPC.

It was mostly a game of how cheaply could you “search the haystack” to find the needles.

Meaning how cheaply could you get lots and lots of clicks so you could find your converters aka customers among all those clicks.

It was a brute force approach.

And then ads got sophisticated.

Really, really sophisticated.

Custom audiences came out, then lookalikes, then website custom audiences, video ads, ad sequencing, retargeting, and tons of new campaign objectives.

Today, instead of trying to process the whole haystack it’s more about finding just the needle.

With a genius pixel things a much more obvious

Meaning having such a well trained pixel that your ads — don’t even show to the haystack — they just show to the potential needles.

Make sense?

Imagine you had x-ray glasses and could see right through the haystack right where the needle is at. You could then reach your hand right in and grab the needle, without searching through the haystack.

That’s what a well trained ad account pixel does for you today.

I call it a “Genius Pixel.”

A genius pixel is an ad account pixel that is so well trained to find your perfect customers or leads, that you don’t even need to use keyword or interest targets anymore.

This is where machine learning and AI are taking ad networks.

This is the future of digital ads.

Would you like to create a genius pixel that hunts down just your customers, lowers your cost per sale, and increases your ROI?

Then let’s get started shall we…

How To Train Your Dumb Pixel To Become a Genius Pixel

Before I go any further, none of this would have happened without the brilliant brains of my friends; Mike Rhodes, James Van Elswyk, Nick Shackleford, Ezra Firestone, Jason Hornung, and Curt Maly.

Without these guys being the pioneers that they are many of this stuff wouldn’t have been thought up.

If they speak, I take notes. Period.

— — — — — — — — — — — — — — — —

Phase 1: Starting a New Account

— — — — — — — — — — — — — — — —

It all starts with creating a new ad account for each product.

Yes, a new account for each product. I know what you’re thinking, stick with me though I have answers for all your fears and concerns.

Every account comes with their own pixel and for absolute best results you want this 1 ad account (pixel) to only learn who buys only this 1 product of yours.

…or leads, for all my “P.P.L.” lead gen homies out there.

Today is your puppy pixel days.

You’ve got to train it to do everything you want.

You’ll have to INVEST (time & money) in training your pixel for a little while, so that later it turns into a 24/7 customer hunting machine.

NOTE: If you have a lot of products, I address a solution for you in the F.A.Q. section at the end. Essentially, just start with your 1–3 best sellers.

You Can Speed Up Your Accounts Genius Pixel Training By Feeding It Data You Already Have, Here’s How…

If you already have a database of customer emails, upload that as “offline conversions” or as a custom audience.

Even better if you’ve got a “highest spenders” database — share that audience with the account.

If you don’t have a customer audience, maybe you have a database of leads. If you have neither, no worries, there’s a solution for creating your first audience too.

By uploading your existing database you’re teaching the algorithm what a customer looks like for this account.

It’s like you’re giving the scent to a bloodhound and then it’s going to go out there and find you more of exactly what you’ve trained it to find.

Now make a (1%) lookalike audience, from the audience you just uploaded.

In Google this is called a “similar audience.”

This a great time to use a “bump, set, spike campaign.” Basically, create a video ad that is a product demo video or a product testimonial video ad.

Set the campaign objective to video views and targeting your 1% customer lookalike audience (LLA).

This campaign is the bump part, of the bump set spike campaign.

ROI and conversions is not our goal here. That will come, we aren’t going to be wasting any money. This part of the campaign is just designed to find people who engage with our video.

It’s a cog in a much larger wheel that will ultimately be very profitable for you.

Some Pro Tips:

  • Use square video (1:1 ratio) because they get better CTR and take up more space in the newsfeed. Split test your video thumbnails and video length.
  • Split test a short video (30–90 seconds) and another targeting the same audience with a longer video (3 mins+).
  • Consider targeting mobile only to save a little money, instead of desktop which is premium traffic that everyone wants.

For the set part, create another audience from anyone who watches more than 50% of your video. In Google this is called “similar to watched.”

Now we’re filtering down to real humans who were actually interested in what we are selling. Which will make the next part become your money maker.

Finally for the spike part, create an ad for your product and target that new “watched 50% audience” on desktop (because it converts better) with a conversion objective and an ad about your product.

  • [BUMP] spending pennies to create views.
  • [SET] segment the engaged viewers by watch percentage.
  • [SPIKE] Ad for your product seen only by the most engaged viewers.

Or you could try your luck pitching a cold audience who is not engaged.

Good luck with that :-)

Good News, This Works Great With Small Budgets. Even As Little As $5 Per Day.

Because it’s a new account you’ll have to start out at smaller budgets, and that’s ok.

You want the algorithm working hard to try and use your $5 — $10/day as effectively as it can. You don’t want it to have $1,000/day where it can make a lot of mistakes and bloat your CPM’s.

Save the big budget for later when it’s time to scale.

Some More Pro Tips:

  • If you have a database of refunders or chargebacks, upload those as an audience as well so you can exclude them from seeing your campaigns
  • Every time you create a campaign that targets someone you should also create an exclude rule to tell the ad network who you do not want to target. If targeting someone, you should also be excluding someone else.
  • Always exclude past leads and past customers from your new lead gen or sales campaigns. Same goes for targeting anyone beyond 30 days from seeing your stuff, if they don’t convert in 30 days don’t beat them over the head. It’s not financially worth it to keep showing your ads to them.

Remember this is about surgery.

You’re using the ad networks full functionality to cut like a scalpel through the mass market right to the buyers you want.

You are training your pixel to become… a genius pixel.

— — — — — — — — — — — — — — —

Phase 2: Growing your Account

— — — — — — — — — — — — — — —

Now your pixel ain’t a puppy anymore, it’s matured.

As your pixel gets smarter you’ll be able to raise your budget and get even more sales. However, always remember that FB works more efficiently with lots of smaller budgets than it does with 1 big budget.

Ad sets and campaigns are free, use lots of them.

Due to the way algorithms work now, it is better to have 100 ad sets spending $5 per day than to have 1 ad set/group spending $500/day.

Break them up by country or state or even city if you have too.

Break them up by mobile and desktop, male and female, young and old.

Let’s talk about scaling this up now.

Keep your ad creative fresh, my rule of thumb is no single ad should show for more than 72 hours.

I mean, unless you have conclusive data that says other wise. For example, if you’re making 300%+ ROI and killing it, by all means DO NOT STOP!

You’ll know your campaign best.

With all these micro audiences you’re going to need a lot of creative. You can try to let your ads run as long as they can, but it’s best to just keep them fresh.

We use a tool called Qwaya, but AdEspresson is great too, for quickly uploading a lot of creative (ad images/videos etc). These two tools give you the ability to click a few settings and VOILA you have 100+ ads from 1 ad.

Then we use a tool called RevealBot to pause and unpause our ads with automation rules. These rules rotate out our ads and pause any that are too expensive.

Keeps our ads fresh and profitable, without having to be tied to the desk.

Why is “fresh ads” so important?

The more times someone see’s your ad the more likely they are to click “hide ad” we call that x’ing out the ad.

This is a negative feedback signal to FB.

Your accounts could even get banned for too much negative feedback.

IMPORTANT NOTE: As you’re filling this new account with more ad sets, ads, and campaigns it’s critical that you use a good naming convention from the start. It’ll also help you use the built in search, filter, and sort feature so you can only see what you need at the moment instead of digging through 100 ad sets. Also, always submit 1 ad first, get it approved, then submit all the others. Too many disapproval's can cause an ad account termination also.

At this point you scale either by adding another account and repeat everything you did up until now with another product. Or create multiple accounts for this same product doing a very similar setup.

Won’t that compete?

Yes, and no.

Every account is different. The dates, behaviors, financials, ads, everything is unique. And FB’s algo optimizes better with lower budgets. It’s better to have two ad accounts with half the budget on each, than one ad account with all the budget.

I know, it sounds asinine, frankly I think it is.

Why would they incentivize use to spend less, instead of more???

My thinking is it’s so they can hook people early, get them a taste of victory at small budgets. It’s a long tail strategy, millions of successful small accounts versus a few monster accounts.

Our job is not to wish the system was different, it is to figure out the system and exploit it to our benefit (within the law and TOS of the platform).

Here’s another way to exploit FB’s spend less incentive.

As you are scaling up, let’s say an ad set gets to $500/day and becomes unprofitable. Instead of shutting it off right away, back it back down to $250 per day and see if it becomes profitable again.

And then just duplicate it with similar ads, but not exact copies.

Lastly, it is ok and mandatory that your ads start and stop from time to time for creative freshness.

However, the campaigns and ad sets should not stop. Try lowering the budget instead of doing a hard stop. Choose a budget that is comfortable to keep spending while you make whatever fixes or changes.

I prefer to use lifetime budgets over daily budgets and rather than using day parting to shut off an ad set use it to modify your bid.

Daily budgets start and stop your ad sets when the budget is spent. Lifetime budget spreads the budget out and lets it run continuously.

You’re not spending any more or less money, you’re just avoiding your ad set jumping out of line in the ad auction.

You want your ad sets and campaigns always in line at the auction, otherwise you could re-enter only to find now you can’t get the same results you were.

Same thing with the day parting.

If weekends or nights or holidays are not profitable for you at 100% of your bid price, then try using day parting to drop it down to 50% or 25% of your bid.

You might even trick FB’s low budget optimization algorithm to your benefit.

You want your campaigns always running, not starting and stopping. Ads are ok to start and stop, they are designed too. Campaigns are designed to run constantly or until their expiration date.

— — — — — — — — — — — — — — — — — —

Frequently Asked Questions

— — — — — — — — — — — — — — — — — —

How many FB Ad Accounts can I have for my business?

I have 12 currently, but plan to expand that to 50. Some of the top guys I’ve talked with have as many as 800 (that’s rare) but many have said they have between 50–200 ad accounts.

How many Business Manager accounts can I have?

You can create two business manager accounts, but you can be part of many. I am a part of 6 different business manager accounts currently (mostly because of consulting).

Is it safe to link them all? (All my Ad Accounts in all my BM’s)?

Yes, if you’re doing things within compliance and not getting too many negative signals like negative comments, hide ad, etc. If you’re not doing things in compliance, well then try that first :-)

I sell 3 different but related services. Is that 3 separate products which need 3 ad accounts / 3 pixels?

Are they different people? Different leads? If it’s the same person, but 3 different services, then really you just need one pixel for the one lead gen campaign. With an email follow up campaign that sells 3 diff services.

If it’s 3 separate lead gen campaigns targeting different people, then yes I would use 3 pixels.

What if we have 500 SKU’s, do we really need to create 500 accounts???

Where should we draw the line?
What about Ecommerce stores?

You don’t have to do any of this, all of this is just if you want better results. And yes if you want the best results you have to do a little more work.

The short answer is you need as many as you need.

But you also don’t need all of them right now. You don’t start with 50 accounts, you build up to that as you go, a little bit at a time. Personally, I have 16 products, but I’ll only advertise the 3 best sellers.

Once I have the customer I can use email follow up to sell all our other products.

If we’re supposed to use one pixel per product, how many pixels can we really use per site?

Think per page, not per site. You’ll have to get strategic and plan things out on paper (whiteboard) first.

Does this really mean we should start using mini-sites with separate products / pixels for each mini-site?

I have been an advocate of mini-sites for a long time.

You don’t HAVE TO do anything. You could ignore all this. It all depends on how clean and how targeted you want to be. Maybe you just use a lot of separate pages on one site, or maybe you develop a lot of mini-sites.

Start small, add a little more each month. You’ll be surprised what you can accomplish in a year and the competitive advantage you’ll have because of it.

Can a pixel develop some “bad behaviors”, based on past and recent activity, that could hurt a specific campaign and/or product?

Let’s say you’ve been training your pixel on an outdoor gardening kit, and then you come out with a hydroponic gardening kit.

Your pixel believes the right customer for you is Billy Bob, the 42yr old home owner. It doesn’t know that your new customer is actually Slick Rick, the 28yr old millennial trying to grow a marijuana garden in his closet.

Can it be untrained, yes. Can it be trained for both, maybe. This is why the best method is to just have a single hyper focused pixel for each product.

And yes, pixels can “run away” if left totally unsupervised. We are not yet capable of real artificial intelligence. These machines are still dumb, they are just very very fast.

Keep a watch over your results, and if you see things you don’t like reduce the budget. Then increase the budget on things you do like, this will re-train the pixel to do as you wish.

If an audience is big enough having more than one ad set or campaign targeting the exact same audience — would this be an issue, with ad placement competition and whatever else?

In Google, yes. In Facebook, no, because is constantly showing different ads to different people and in different areas of the feed/ecosystem.

On a web page with a banner ad (Google) there is only one ad position. So If I create 5 campaigns targeting that ad position they are going to compete. However, Facebook has many ad positions within their newsfeed so it can spread the competition around better.

There is still a limit, so don’t go crazy with it. Some overlap is ok. 50 campaigns with overlap is probably not.

You mentioned the importance of a naming convention — would you mind sharing your ads naming convention?

Campaign: product — objective — persona
Ad Set: location — demo — targeting — placement
Ad: angle — image — color — version #

For example, BP FB Ads — Conversion — M 25–45

— — — — — — — —

Conclusion

— — — — — — — —

I know I’ve just unloaded a lot on you.

My advice right now…

Print this article.

Re-read it one phase at a time, highlight parts you want to remember. You don’t do all of this at once.

This whole process I just explained is designed to take place over 3–6 months, but it starts today with 1 new ad account and just $5 per day in ad spend.

Lastly, I’ve tried to be pretty thorough (obviously) in this article.

However, there are lots of nuances that I just couldn’t fit in (some things require video tutorials), if you want a deeper dive try our Bulletproof FB Ads course.

This is not a ploy to hide stuff behind a paywall, I just can’t cover nearly as much in an article as I can in the 50+ videos inside our course.

We also have a Slack group, if you want direct help from myself and others who have already done this process. Myself and our course instructors are inside the Slack group daily answering questions and giving feedback on campaigns.

Thanks for reading
✌❤️

About Justin Brooke

I’ve been running online ads for over a decade. Managed $10mm+ and now I train paid traffic beginners to become media buying masters at AdSkills.com.

FREE CASE STUDY…
How I Scaled My Google Ad Campaign From $500/day — $65,000/day

POPULAR ARTICLES…
I Spent $30,000 Testing Different Blog Designs — Here’s What I Found
10 Online Ads That Made Millions For Their Founders

Justin Brooke

Written by

Spent $10MM+ on ads, acquired 50K+ customers for my clients, now founder of http://www.AdSkills.com

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade