Dispatch № 3 - Lookalike Audiences
LALs, LLAs, LAAs: A Hit Pop Song...
Hello and welcome to another free dispatch with me, Jake the Ad Nerd 🤓 (< not me)
I'm really excited about this one, not only because I love the topic, but because a few well-known and experienced advertisers were gracious enough to contribute.
Spoiler alert!
It's Cory Dobbin, Andrew Foxwell, and Zach Stuck.
I know, right?! Put your tongue back in your mouth 😜
But, before we hop into all of that...
I’ve got to hit you with some disclaimers (they go in every dispatch).
If you read these last time... you have my permission to skip it and continue on, reader.
If you're new, do me a favor and read this section.
We'll get to the good stuff in no time at all!
1. There is no one unicorn source (including this one) that will teach you everything you need. If that’s why you’ve decided to be in this community, you’ll soon be disappointed.
2. What works for one advertiser, might not work for you. These are merely observations on what myself and others see is working across accounts and verticals. You need to test responsibly and at your own discretion.
⁉️ What Are We Nerding Out On?
Unfortunately, it's not a hit pop song that we're collaborating on in this dispatch but, I'm excited to cover in more detail... Lookalike Audiences!
What's the history of Lookalike Audiences?
If memory serves me correctly, lookalikes popped onto the scene in 2013. If you were jamming on Facebook Ads back then, we're automatically best friends. Deal with it.
I remember when they launched. It was the first time that Facebook Ads really allowed us to lean into leveraging the amount of data that they were processing.
As with anything on Facebook, lookalikes are not a one size fits all solution and more often than not, tactics deployed effectively in one scenario may be ineffective in another (operating under different assumptions).
How do I set them up?
I'm not entirely sure this is the place where we should break down, in detail, where to go to get things going. There are a ton of great resources out there on where this all lives within FB Ads Land (audience manager), so for this dispatch, we'll stay pretty macro in terms of setup.
If you’d be interested in a detailed post in the future about the exact setup process for Lookalike Audiences and how they use 10k data points (or some shit), drop a comment or hit me up and let me know.
It used to take around 24 hours for Facebook to create a Lookalike Audience. Now, it’s more like 1-6 hours (from what I’m seeing). So, if your lookalike doesn’t instantly appear, try not to worry and give it some time. You’ll then see that magically algorithmic created lookalike in the Audience section when building out a new ad set.
Now, a few things to note when we're talking about LLAs: not all Lookalike Audiences are created equally. It can be easy to get overwhelmed with all the different types of ads and experiments you 'could' be running right now.
Lookalikes are no different.
However, I do want lookalikes to be one of those things that you start working on, know about, and are constantly thinking about going forward. It's a great asset when utilized correctly.
🤝 Let’s Be Honest
Standby for honesty time with Jake...
Because… that's what this whole community is about. It might come back to haunt me with a few laughs and some internal shock from my team. But, if this one section helps someone, then it’s all worth it.
Plus, you as the reader deserve the honesty for spending your time here.
This newsletter will always be a place for no BS or chest-pounding.
I've eaten shit on numerous occasions... numerous. With that comes failure. Also, an insatiable desire to continue testing and to improve upon myself and my craft.
Lookalikes were once one of those things that I ate shit on (for a bit).
Why?
I was blind to a very common nuance early on when LLAs first came out.
I can see it clearly now when speaking with others who are at the level that I was at before (no shame). That is… not fully understanding…
Please pay attention to this: Success depends on how much data and how much VALUE is going into the source you’re using to build Lookalike Audiences. That little tidbit is often overlooked.
You see that lookalikes can be run and then you just go off into testing them all. Please keep that nuance in mind before any of that mess. For most of us with higher spending clients, the brands are coming to us with seasoned data.
Back to nuances… the same can be said for growing a business. There was a time when we were eating nothing but DIRT and I couldn’t see the nuances that go along with partner/client relations and growing an A+ team. It forced me to evaluate the idea that I could have walked away from something so special.
Much like the agency, lookalikes were a real kick in the teeth but, if I would have walked away from both rather than work to improve upon my understanding of them, I’d be a very sad nerd right now.
🍯 The Good Stuff
Let's kick things off with the contributors because it would be wrong of me to make you wait any longer.
I reached out to these advertisers and asked two questions.
So, we'll keep it all in an "interview" style. I didn't want to ask the standard crap, like what lookalike should I run to do x,y,z, or what lookalike will turn my account around today to get 19x ROAS...
So, I went with:
1️⃣ What has been your best performing Lookalike Audience(s) this past year?
and
2️⃣ What's your favorite Lookalike strategy of all time?
I left it all in the hands of the contributors to take each question and explore it on their own time.
Personally, I’m fired up about the responses and you’ll soon find out my reasoning behind these specific questions!
Shall we begin?
First up is Cory Dobbin, the Founder of Aaron Advertising. I reached out to Cory because I knew that he'd have something valuable for you as a reader.
He's one of the few FB advertisers that I know, who consistently gets hands-on testing with new products from FB Ads Land. Plus, he's dropping in with years of experience, over $10M in spend managed and is just an all-around genuine person.
Take it away Cory!
What has been your best performing Lookalike Audience(s) this past year?
“For me, LALs that perform the best really vary a lot from client to client.
So, instead of picking one LAL type that works best for me, what I have found works great across the board is the recency used for the lookback window. Considering the dramatic economic shifts happening in recent months, it's important to consider that purchase behaviour has also changed dramatically.
With that in mind, I have been finding that using signals in the last 60 days has proven to work very well as it better reflects our current-day purchase behaviour. Using signals of anything larger than that will show a mix of behaviour from pre-COVID mixed in, which is going to present conflicting data.
All that said, try shifting your Purchase LAL seed audience, for example, from 90-180 days to a 60-day window and see how that performs.”
(Jake wonders in…)
This is such timely and really, timeless advice. Create your lookalike’s using signals that are in-tune with your purchase behaviour.
With lookalikes, you should always identify what level of behaviour is ideal for flagging intent. Love it.
Carry on Cory!
What's your favorite Lookalike strategy of all time?
“I've recently been testing a new LAL build-out in partnership with Facebook that they are calling ABC LALs. This uses the API to build LALs off of entire campaigns or ad sets instead of audiences.”
(read that again and try not to have your mind blown 🤯)
“This is very powerful because it uses multiple input signals instead of traditional LALs that use a single audience signal. Not only that, the API keeps the audiences updated by refreshing the source signals every couple hours, and also works great for horizontal scaling as it only tends to only overlap the source audiences by <10%.
I've been testing this on a few accounts with great success so far (2.4x in cold TOFU prospecting on $1,500/day in spend on one account, for example), and I am looking forward to Facebook rolling this out into Ads Manager some time in the near future.”
(Jake strolls in…)
I mean come on… you should all be fired about ABC LALs. Cory even let me in on a little a secret. He’ll be posting the results of those tests next week ;)
Also, does this mean that Facebook themselves are calling them LALs…?
I need to ask him about this…
Tyler Narducci backs this up with a twitter poll from earlier this month… So, myself and Andrew might just be the only nerds walking around slapping LLA everywhere…
But I digress.
Next up is Andrew Foxwell. Andrew Co-Founded Foxwell Digital with his wife Gracie Foxwell and has spent the last ten years advertising on Facebook and Instagram, spending over $50 million dollars in the process. He’s also a co-host on the eCommerce Influence Podcast and brings to the table years of experience and a knack for translating nuanced topics into concepts that anyone can understand. Proof of that can be found in his online courses that have become a go-to for media buyers.
You’re up Andrew!
What has been your best performing Lookalike Audience(s) this past year?
“I love a solid high value purchaser LLA.
The old fashioned way. Download your customers, sort by order and amount and then take that segment and create a lookalike, mostly for 5 or 7% to see. If it’s a smaller account try a 2%.
This old trusty is a favorite of mine always.”
What's your favorite Lookalike strategy of all time?
“A big issue I see in auditing and consulting on $15M in spend in the last six months on FB/IG is a lack of diversification in prospecting.
People rely on one or two lookalikes for too long and don’t think about how they can horizontally scale more effectively.
The strategy is to think about high and low value lookalikes in different CBOs. For example, this would be something like:
High Value CBO- 7%
PUR 180 LLA
Top 25% Last 180 LLA
180 ATC 180 LLA
High Value Customer LLA
Low Value CBO - 7%
180 View Content LLA
180 Email Opens LLA
180 Engagers Lump LLA
180 50% IGS Video View LLA
You can choose whatever size you want and whatever time window too, but the point is to diversify while also getting a read on which audiences are most effective.
Then once you find winners you can build more prospecting around them and compliment it with MOFU. For example, if you find the LLA or top 25% killing it, integrate it into your MOFU strategy more.”
💚 Pause Button…
I hate to be that guy but if you’re sleep… No, you know what… Test it out! 👏👏👏
Take a moment to think about your strategy and whether or not it’s built to scale horizontally. If you’re interested in a deep dive into this, you can take Andrew’s scaling program and get $30 off with code JAKE30.
Transparency: I get a % of sales generated from that code and I will be using any funds from it to donate directly to Trips For Kids. As a mountain biking nut, I donate specifically to the TFK, a charity that has introduced over 140,500 at-risk youth to cycling since 1988 through mountain bike rides and Earn-A-Bike programs.
Last, but not least...
Zach is the CEO and Founder at Homestead Studio, a digital growth agency that helps eCommerce businesses scale. Most of their clients spend anywhere from $20k/month to $500k/month on paid advertising. Zach can be found on Twitter (@zachmstuck) providing nothing but humble, insightful value to his followers and the FB Ads community as a whole.
Send it home Zach!
What has been your best performing Lookalike Audience(s) this past year?
“This answer is pretty boring, to be honest, but in 2020 the top-performing Lookalike is a 1% Lookalike of purchases.
The lookalike is usually value based and made up of 180 day purchasers.
This lookalike has been a staple in almost every single ad account, whether it's a brand new account (spending under $10k per month) or a continuously scaling ($100k+ spend per month) account.”
(Jake pops his head in…)
Noticing a theme here for the best performing lookalike? 😉
What's your favorite Lookalike strategy of all time?
“My favorite lookalike strategy of all time has to be creating lookalike "stacks".
My current favorite is a 10% lookalike version of this.
This means we create 180 day audiences of the following events:
Website Custom Audiences (WCA), View Content (VC), Add To Cart (ATC), and Purchasers (PUR) and make 10% lookalike audiences of each and compile them into one ad set.
This has been a consistently high performing audience that allows us to test different bid strategies in the account whether it be lowest cost, cost caps, or bid caps.
Running this stack with bid caps or cost caps has allowed us to keep account ROAS in line while testing various other things like: creative assets, copy benefits/hooks, offers, placements (Instagram stories only or feeds only).
My response above was consolidated from my personal thoughts plus thoughts from a few of our team's paid social buyers (Riley Trotter (@_RileyTrotter) + Tim Aton (@Tim_Aton) + Connor Rolain (@CROL44)).”
🤤 Are You Hungry Yet?
If the last three sections didn’t kickstart your appetite and get you hungry to work on your own lookalike strategy... I quit.
In all seriousness, I'm beyond grateful for the time and insight that Andrew, Corey, and Zach offered up. There’s some serious inspiration embedded in their contributions to this dispatch.
As I look to wrap this up, I wanted to add a skosh more info. (fun fact: The word skosh comes from the Japanese word sukoshi, which is pronounced "skoh shee" and means "a tiny bit" or "a small amount").
Until I can get a full understanding of who's in this community, I just wanted to toss a few ideas out there for anyone that might be outside of eCommerce because, heck, ya never know... there could be some value there.
For example, what if you're working with a local business account or running lead-gen accounts and don't have Cart/Checkout/Purchase data?
Let's hop into that for a quick (and I mean quick) second...
For local businesses, I'd personally recommend you start with a 10% lookalike and then geotarget it down from there. Or, if you have a high-value customer list, use that at 4% and then geotarget from there. All the while, keeping in mind the nuances mentioned above.
If working on lead-gen accounts, one of my favorite things to do is modify pixel data.
^^^ That doesn't sound as sketchy as you think and it’s something that very few talk about. Yet, I’ve seen great success with it in the past when we had lead-gen accounts.
What I mean is, test out modifying your pixel code or use google tag manager to fire off Initiate Checkout on the first step of your lead form and then send a purchase and value over with a completed lead when a form is completed. This allows you to create the lookalike audiences that we’ve referenced here and it allows you to provide more data points to FB, rather than just Viewed Content and Lead.
I’ll probably cover that last bit in a podcast or thread in the near future 🤓
🏁 In Closing
Be smart about the information that you've read here and realistic about what you believe may or may not work for your account(s).
Allow time for testing.
Be cognizant that you may not see immediate results from your audiences depending on where the business or account is at (data and budget-wise).
I’m sure you caught on to the congruent theme of what has been performing best.
There’s a reason I wanted more than one contributor to answer that specific question because I had a very strong feeling that the responses would align.
If you missed it…
Value based Lookalike Audiences built from Purchasers are gold.
It’s a staple and a backbone for more advanced FB advertisers. Especially when using signals that are in-tune with the customer intent because it’s literally one of your most valuable audiences.
With that, make sure that you’re parsing out and building a strategy with your high and low-value lookalikes in separate campaigns.
As with most macro-level strategies on accounts, you should also be strategizing as to whether or not your lookalikes are built to scale horizontally (best ads with different audiences).
Be aware that your audiences within a campaign should be close in audience size and the data behind the lookalike is truly important.
Lastly, remember that tactics deployed effectively in one account scenario may be ineffective in another (operating under different assumptions).
With gratitude.
Hopefully, some of these approaches will help you use the Lookalike Audience feature in new and specific ways that are relevant to you so that you can effectively grow and scale your accounts.
If anything, I hope it provides inspiration.
Until next time.
Cheers,
Jake the Ad Nerd 🤙
Did you enjoy this?
Click the “like” button! It’s not because I live off likes — it helps me know if I should keep making more stuff like this. This community is for you, as much as it is for me.
Also, consider sharing this dispatch if you found value in it.
A Sincere Thank You goes out to everyone who shared the previous dispatches. This little substack has now grown to over 300 subscribers and I’m truly blown away!
The Footer
📖 Things I've Enjoyed Reading
The Subtle Art of Not Giving a F*ck by Mark Manson [Link]
JTBD - Jobs To Be Done [Link]
’Problem’ w/ Cliffs by CommonThreadCo [Link]
🧐 Interesting & Random Things
Explore an Ancient Greek City [Link]
Dayparting: Go Home. You’re Drunk [Link]
🤓 Nerd Link(s)
FB Analytics (some might not know it exists 🤷♂️) [Link]
IAB U.S. 2020 Digital Video: COVID in Context [Link]
Your Gmail Habits/Trends in DataStudio [Link]