How I’m building cold audience Facebook adverts in 2021

In this blog post I’m going to share my plan for building cold audience adverts in Facebook, using Facebook adverts.


Facebook ads, frankly, drives me bananas. Cuckoo bananas in fact. I’ve had ad accounts shut down, I hate the interface and there’s always some new Tai Ovens Becker bellend telling me that they have the latest secret to cracking Facebook ads.

Creating adverts on Facebook is very very different to advertising in traditional media. It’s more surgical, specific and precise. In some respects, this is good because it means we can change our tactics quickly. In other respects it means that the method of making money from adverts also changes.

Miles Beckler is my Yoda

The exception to the rule is the recent course and content from Miles Beckler. I’ve been a member of his private forum for years and what makes Miles’ content on Facebook ads different, is that he has NO templates – he teaches why Facebook works the way it does and he’s constantly updating his tactics.

So with that, the method I’m going to share with you today in this blog post, is partly from his content and partly from what I’ve learned over the years.

My aim is to generate an audience on Facebook of engaged viewers

The first thing to note is that I’m creating two separate advertising campaigns.

  1. Build an audience of people who are engaged with me, my message and appear to be funnel builders
  2. Try to convert that audience of engaged users into leads and sales

Qualify out “non-funnel builders”

What I’ll be doing at a high level is building an audience of people who seem to be funnel builders. I’m trying to qualify out non funnel builders.

Cold advertising is a lot like cold calling. The goal is not to make a sale or even generate a lead, it’s to find out “am I talking to the right person?”

Target engaged audience members later with conversion ads

If I have an audience of 100,000 people, I want to put a video/message in front of them that asks “are you a funnel builder? Because if you are I’m going to teach you something and target you later.”

I want funnel builders to show me who they are. I’ll be doing that by showing them a video and anyone who isn’t interested won’t watch. Anyone who does watch is presumably a funnel builder, so I’ll tag them and show them more content later.

My 2021 Facebook advertising strategy

Research all my audiences and targeting

Arguably the most important part of the process, audience research is what most courses and gurus miss out. Because it’s not sexy or fun, lots of funnel builders skip this step or don’t know how important it is.

Audience research comes first and I’ll spend an hour discovering and creating lists of interests that I can target. I follow a few principles, but what I want to end up with is a list of interests inside Facebook that I can target.

What I’m looking for are interests which I can target inside Facebook’s users. From golf to headphones to baking, I can target specific interests that would be relevant to my audience.

Weather you’ve liked a page or not, Facebook has grouped all of us into various interest categories. If you want to see yours, you can head to your ad preferences and see that categories and interests that Facebook believes are pertinent to you.

Inside the Facebook Ads Manager, you can head to Audience Insights (which is totally free) and start to explore the interests that you’re going to target.


Demographics like age, location and gender are certainly important, but it’s more than just trying to be “accurate”. What you’re looking to find is an audience that no one else is targeting.

I’ll target most English speaking nations in one ad campaign and European nations in another, EMEA, APAC etc. But I’ll also make sure that I pay attention to demographic data from YouTube and Google telling me who my audience is. Typically, it’s men and women aged 35 – 55.

That’s a good place to start with demographics, but what I’m really looking for are targeting interests.

Because Facebook ads work like an auction, lots of advertisings are vying for the screen space and attention of lots of different audiences. In short, the larger the audience, the higher the price to advertise to them.

Think about the price difference between advertising at the Super Bowl on TV at halftime (in view of billions of people), compared to advertising in a local niche interest magazine.

In truth, we can never truly know what the best interests to target are, until we target them. Some people might do really well targeting one interest and others might not. It’s a case of having lots of interests ready to test and killing/keeping those that do and don’t work.


For example if I run a golf store, I might think about targeting people who are interested in Tiger Woods. In the US alone, according to Facebook, 150 million to 200 million people have shown interest in Tiger. That’s probably too big an audience.

This is where “true fan knowledge” comes into play. Because there are literally millions of people who have heard of Tiger and maybe even love him and what he does. But who have absolutely zero interest in golf and buying golf gear. From his older EA sports PGA games, to Nike sponsorship deals, he has an audience larger than the game of golf.

If we get more niche and target someone like Bubba Watson, it’s likely that only golfers will know who he is.


As you can see, the audience size is “only” 200,000. That’s 0.1% the size of Tiger’s audience. Not 1%, 0.1%. Yet 200,000 – 250,000 is still a great audience size to get in front of.

The closer we get to “true fans” the more casual fans we’ll eliminate.

To be fair, Bubba Watson is a killer interest to target and I’d be keen to test. However there’s no harm in trying to get more specific. In my tests, I’ve had metrics like cost per view, cost per lead and cost per click go from $5, $10 and $25 plummet to $0.50 or even to $0.05.

If we look at a golfer like Vijay Singh (my boy) in the US there are around 5000 people “interested” in him. However, if we expand that to India and Fiji (Vijay is from Fiji) it jumps to 350,000.

It’s worth noting that the numbers displayed on Facebooks audience size aren’t 100% accurate. But they’re a good starting point.

There’s an important lesson here in regional targeting and causal fan knowledge. While bigger players might seem more obvious, there are literally millions of potential people that you could reach with a little digging.

I want to create a list of 30+ interests that I’ll target and test per campaign. While I won’t test all of them at once, I want to make sure I have a list of interests that I can target before I get going. This makes the second half of my campaigns easier to manage.

There are no sweet-spot numbers that I’ve seen, which guarantee results or lower costs. I’ve heard some people say target fewer than 100,000. I’ve heard others say 50K- 75K is the sweet spot. In truth, you’ll never know until you test and experiment for yourself.

A good starting point is to look for audiences lower than 1m, but throw a few big players in there just to see what happens. I’ve had some videos and adverts do really really well with an interest, and then become 100x more expensive for another advert but the same audience.

How to 10X this research

I use a program called that does all this research lighting fast. What does is take one interest or niche or topic and spit out LISTS of interests that you can target. Within 10 minutes you can have DOZENS of lists of audiences and interests to target, rather than do it all manually on Facebook.

Check out here (and get a sweet discount).

Set up my lookalike audiences

“Lookalikes” is the horrifying and bleak world of A.I. and machine learning come to fruition. For all of your unique nuances and quirks, Facebook is able to take your activity, quantify it and find millions of other people similar to you. If you thought it was bad “just being a number”, now you’re not even a unique number.

Lookalike audiences take real data such as website visitors, email subscribers and page followers and asks “are there any other people out there who are similar to these ones?” It’s basically finding people who have already taken an action of some kind and finding people who would also most likely take that action.

To do this, it does require that you’ve got data to upload and use. The more data and the more recent, the better. I’ll create a series of lookalike audiences based on customers, leads and conversions.

  • Lookalike 1% of customers from my customer list
  • Lookalike 1% of website traffic from my website
  • Lookalike 1% of leads from my leads list
  • Lookalike 1% of newsletter subscribers from my email list
  • Lookalike 1% of conversions from my landing pages/squeeze pages

I see lookalikes as hidden interest categories. It’s reasonable to think that people who visit my site or sign up to my email list are similar in nature and interests. Therefor, if we can find others with the same interests and behaviours, they’ll likely become customers or leads too.

With these 5 audiences and my list of 30+ interests, I’m now ready to build some content for my campaigns.

Create a handful of videos ranging in length and topics

My basic strategy is to create 5 videos and display them to cold audiences i.e. lookalike (Lal) and interests. The idea is that if someone who has never heard of me watches some or all of my videos, then they’re probably a funnel builder.

If I put a video in front of 100 people and use language like “funnel building, squeeze pages and conversion rates”, people who are likely to be funnel builders will pay attention. It might only be 20 or 5 or even 1, but that person is likely a funnel builder.

Later on, because I can track and target people who have watched these videos, I’ll show anyone who has watched them another series of ads or videos (maybe even the same ones). The fact that they watched part or all of a video, qualifies them as someone who potentially could be a lead.

Cold audience outreach is NOT about making the sale. It’s the exact same for cold email, cold calling and door-to-door sales. The only question we’re looking to answer with any kind of cold campaign is “are you the right person to talk to?”

With that, I’ve created dozens of videos ranging in topics, length and style to start testing with my audiences. It starting to get a little complex, with multiple videos and multiple audiences. How can I possibly know where to start and what to test first?

We’ll come to that later, but the two key points are:

  1. Have a methodical repeatable approach to testing
  2. Start testing something and make changes later

As a rule, longer videos are going to cost most to get people to watch. That’s a very broad statement but in my experience it tends to be true and it’s a fair assumption to make. Getting someone to watch a 15 minute video is more costly than a 5 minute video. And that’s more costly than a 30 second video.

However, they are also potentially more qualified. If you’ve stopped someone from mindlessly scrolling through memes and passive aggressive statuses, to watch a video for 15 minutes and they watch even 25% of that video, it’s clear you’ve found someone who is probably interested in what you’ve got to say. You’ve just bought 3-4 minutes of someone’s time and and in Facebook time, that’s eons.

So we want to balance length of video time = qualification vs. cost of their view. In truth, again, you’ll find a sweet spot if you test and experiment. I’ve found that 5 minute videos do well for my audience and cost way way less than longer 15 minute videos.

Opposingly, creating super short videos might get dirt cheap views and engagement, but are they worth the cost? If all you end up with a load of low quality traffic and audiences, then you’ll only pay for it later on down the road.

I created a list of 15 different topics that I would talk about. I did create a list of a lot more, but narrowed it down to 15 to start.

  1. how much to charge for a marketing funnel
  2. how to build funnels for customers
  3. how to sell marketing funnels to customers
  4. how to define a niche for your funnel business
  5. the best tools to use for a funnel
  6. how to write a funnel proposal
  7. how to answer “you’re too expensive”
  8. what a $25,000 funnel looks like
  9. do you need to build the whole funnel?
  10. do you need a portfolio
  11. where to find new leads
  12. why you should stop building marketing funnels
  13. how to raise your prices without annoying people
  14. how to create a webinar funnel
  15. top marketing strategies for 2021

I’ll only be using 5 of these videos at a time, and I’ve created mostly 5 minute long videos outlining ONE key point. I have got a couple of longer videos in there, but I would imagine they’ll be too expensive to keep running.

Typically I’ll put these videos on my Facebook page and publish them like videos on YouTube. I’ll actually create the ad copy and creative on the Facebook page because I’ll be using that post later to promote and share.

Lastly, I’ll make sure to create an audience that targets anyone who has watched at least 25% of any of the videos in the last 28 days. That is what builds my “engaged” audience for our warm traffic and conversion ads.

Create campaigns based on video views, and one ad set to start

In the ad manager, I’ll create a new ad campaign focusing on video views and I’ll create one ad set to start.

The lexicon of Facebook used to confuse me and I never really knew what I was doing, until someone made this very simple table for me.

The Ad Campaign is basically the broad goal. It’s what you’re looking to achieve from the campaign. Do you want leads, conversions, video views, comments, engagement? Don’t overthink this – Facebook has close to 10 years of data now on who does what.

Ad Sets are who your advert campaigns go out to. It’s the audience that you’re showing adverts to. You’re going to have multiple ad sets inside one campaign targeting different audiences. The campaign will stay the same, the ad set will change and grow/evolve over time.

The advert is the actual creative/copy/video that you show. Ideally you’ll have the same advert running to each ad set. That makes the tests easier to run and easier to read/analyse. While each advert might have it’s own ad set, the actual adverts themselves will be the same.

So per campaign, we’ll only ever be showing ONE video/advert but to multiple ad sets. This way, we can test if the ad set (audience) is a good fit for that video and if people are even interested in it.

Note: there is absolutely nothing stopping you from having 5 different videos per ad set and 100 different audiences per campaign. Many larger agencies run far more complex campaigns than the one I’m outlining. However I run a small team and it makes sense for us to test like this.

I typically name my campaigns something like Cold – video topic – video views i.e. Cold – how much to charge – video views. That way I can easily see what videos are getting views and conversions.

I keep budget optimisation off and right now I’m not doing an A/B split tests.

Leave the auction amount up to Facebook

At the ad set level, I’ll leave each the suggest bid completely up to Facebook. Many people panic at this point, worried that Facebook is going to start giving them $5 clicks or views. You needn’t worry because the ad set itself will have a daily budget.

Set the budget to $1 per day

I set the budget of the ad set to $1 per day. Which means no matter how much Facebook bids for views and clicks, I’ll only ever pay $1 per day per ad set/audience.

For example, lets say in my first ad set I decide to target Vijay Singh. I won’t tell Facebook how much to bid per video view. But I will tell it that I only want to spend $1 per day on this audience until I’m read to spend more.

With 5 different audiences/ad sets, I’ll be spending $1 x 5 per day = $5 per day per campaign, to start.

There’s a few different reasons for this. First, Facebook tends to fluctuate views and impressions and it tries to work out the audience to start with. I don’t want to immediately say “here’s $100, go nuts” because it’ll try to spend that as quickly as possible.

I want to level out the data and get an impression of what happens over time. Yes, it takes longer, but it also gives us more control. We can always increase the daily budget later or turn off ad sets that aren’t working.

Set up a different ad campaign for each video

As I mentioned, I’ll create a different campaign for each video. I might test the same audiences for each video and each campaign. But over time they’ll diverge and have their own audience groups per campaign.

During the audience set up, I’ll make sure to either target a Lal (lookalike) audience or an interest. I’ll put the locations I want to target into Facebook, as well as any demographics. I’ll also EXCLUDE any audiences that I don’t want to target i.e. customers, current traffic, current engaged people etc.

If I target an interest, I’ll target one and only one interest. The massive mistake I see people make with Facebook ads is trying to target too many interests per ad set level. It might make sense to group all your interests together, but it doesn’t work in practice,


Because 1. it’s impossible to know exactly which interest is working and which isn’t and 2. it creates a larger than needed audience. It might seem to make sense to target everyone who loves golf, golfing, PGA, Tiger Woods etc. But we need to be more specific than that and more surgical.

When each ad campaign is approved, duplicate the ad set and change the audience

It’s worth mentioning at the point that I only create ONE ad set per campaign when I start. The reason is that I want to make sure that my advert is approved and clears Facebooks Skynet style policies robot.

Sexual imagery, false claims, words like “other Muslims” or “other women” (targeting personal attributes) are against Facebook policies and I’ve had more bans and rejections than I care to mention. So I like to test the water and see if it’s approved first.

It’s crazy and hypocritical and bullshit (see above) – but, I’m asking to play with Facebooks toys so I should play by their rules.

If and when an advert and campaign and adset is approved, then I’ll duplicate out the ad set, change the targeting to a different interest or audience and keep the advert/video the same.

Wait for 5 – 7 days to see what results I get

Now we play the waiting game.

Leave the campaign and ad set and just wait. Let Facebook spend your money and come back and see what the results are.

I tend to find 5 – 7 days is long enough for Facebook to test an advert and start getting accurate data. Occasionally I’ve seen Facebook take a day or two to find any audience, in which case I’ll still leave it. Or other times it’s spent $5 in a few hours and does so really quickly.

This is the price of admission and overtime you will get better at measuring results. For the first few, just let Facebook figure it out and we’ll make edits later.

Measure and target 25% video views

My main objective for videos to a cold audience is 25% video views. I want to target people who watched at least 25% of my videos.

If I’m talking about golf swings, that is crazy boring to most normal people. But some people LOVE to talk about golf swing. So if they watch 25% of my video on “the perfect golf swing on the tee is killing your bunker shots”, then it’s a good chance they’re in my target audience.

We’re like adolescent monkeys though and our attention span is worse than ever. So I’ll forgive people if they click off or stop watching, but if I had their attention for 1/4 of a 5 minute video, that’s 1 minute 25 seconds and THAT is an eternity in Facebook world. To me, that’s someone worth following up with.

As I mentioned already I’ll create a audience of people who watched at least 25% of my video, because I’ll want to target them later. With these campaigns, for cold audiences, cost per 25% view is what I’m interested in.

Shaun from Miles Beckler’s EXCELLENT course on Facebook advertising showed me that you can create custom metrics to report on in Facebook so I can easily work out (Number of 25% views/cost)=cost per 25% video view.

Switch off “expensive” or low converting audiences

Now comes the point where we keep or kill our ad sets or even entire campaigns. What do we deem too expensive or low conversions?

For me, I wanted to just experiment and see how low I could get my views for. It’s also worth knowing your numbers. I.e. does it even mathematically work out to get 25% views for $1? That means you can only get 5 views a day and it’d take months to fill up an audience for future sales.

Really, we should be looking for $0.10 or $0.05 per 25% but I have seen lower. That’s why you want a broad range of videos and audiences to start, to see what you can get down to. If you have 5 campaigns running and some are at $0.10 and some are at $3, you’d be better off focusing on the lower number to keep and potentially stopping the higher one.

You can see in the snapshot above, for some of my campaigns the cost per 25% view was crazy high. Perhaps I left them a little too long, but that’s all part of the learning curve.

You can see now that I’m getting much better prices on newer campaigns on average. So I’ll turn the expensive views off (remember, I can always turn them back on) and delve a little deeper.

If we look at one individual campaign, you can see the numbers in more detail and it gives a clearer picture. Some ad sets/audiences are giving us very cheap cost per view at 0.03 – 0.12 per 25% view. However some are at $0.53 and $0.89. This is now starting to tell us what audiences are worth focusing on and who we can continue to target.

There’s a mix of interests and Lal’s in there. We’ve even manged to generate a couple of leads from cold.

Test and repeat

If we turn off an expensive ad set or campaign, we then have to create a new one. I’ll work through my list of interests to target and slowly find the best audiences to get in front of.

It’s then a case of looking at our data every few days and turning off what doesn’t work and trying something new. New videos, new audiences, new interests.

But what if none of my adverts work?!

It’s a valid fear. And the reality is that you are going to get ads and audiences that fail and suck. Good. You want to see adverts fail and not work quickly and move on to create other ads.

Fail fast and fail early, get the bad ideas out the way and start finding the ads that work. You’re only going to see 1/10 ads work and work well, that’s par for the course.

But with Facebook advertising it means that we can easily create and kill adverts that do or don’t work.

How I’m building adverts for 2021 using Facebook

I’m creating 5 – 15 videos (probably around 5 minutes in length) and pushing them out to around 5 interests/audiences each meaning I’m running around 25 different ad sets. I’ll be measuring audience engagement by looking for people who watch at least 25% of the video. Then I’ll target those 25% views with further ads.

7 strategies to 6 figures

I’ve published my top seven strategies below for getting a marketing funnel business to 6 figures+ in revenue. You can get the guide sent straight to your email inbox, just click the button below.

Posted in

Mike Killen

Mike is the world's #1 sales coach for marketing funnel builders. He helps funnel builders sell marketing funnels to their customers. He is the author of From Single To Scale; How single-person, small and micro-businesses can scale their business to profit. You can find him on Twitter @mike_killen.