Hi there,
Thanks for reaching out. I saw your post about troubles with Lookalike audiences on Meta and thought I'd share some of my thoughts and experience. It's a common frustration, and you're right, Meta is not always great at explaining its own rules. I'm happy to give you some initial guidance based on the hundreds of campaigns we've run.
There's quite a bit to unpack here, but getting this right is the difference between burning cash and actually finding new customers.
We'll need to look at the basics of why your audience isn't showing up...
First off, let's tackle the immediate problem. The most likely reason you can't select your Lookalike audience in the 'saved audience' dropdown is almost certainly a simple but poorly communicated rule: your source audience is too small.
For Meta to create a Lookalike, the original custom audience you're building it from (in your case, 'engaged users of my Facebook page') needs to have a minimum of 100 people from a single country. If your audience is below that number, Meta simply won't have enough data to find meaningful patterns and create a new audience. The option might appear greyed out, or sometimes just wont show up at all. It's a bit of a weird system.
However, and this is the really important bit, even if you had 1,000 or 10,000 page engagers, I would strongly advise against using that as your source for a Lookalike audience if your goal is to get conversions like sales or leads. It's probably one of the lowest quality sources you can use. Think about it: an 'engager' could be anyone who liked a single post, left a comment, or just watched a few seconds of a video. That person is miles away from someone who has actually shown intent to buy something. You're telling the algorithm to find more people who like to click 'like', not people who like to enter their credit card details. This is a crucial distinction that trips up a lot of people.
To get real results, you need to be much, much more strategic about which audiences you build and in what order you test them. This is where we need to move beyond a simple A/B test and start thinking about a proper funnel structure.
I'd say you should rethink your entire audience strategy...
When I audit new client accounts, the most common mistake I see is a messy collection of ad sets testing random audiences against each other. A far better approach is to structure your campaigns based on the customer journey, or what we call the funnel: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu).
This sounds like jargon, but it's simple: you talk to complete strangers (ToFu) differently than you talk to people who have visited your website (MoFu) or people who have already added a product to their cart (BoFu). Your audiences should reflect this seperation.
Here’s the priority I would usually use when building out audiences for a new account, or an account we're looking to scale:
META (Facebook/Instagram) ADS AUDIENCE PRIORITISATION
ToFu (Cold Audiences - People who don't know you):
1. Detailed targeting (interests, behaviours, demographics) -> This is your starting point. You need to feed the algorithm data before it can do anything clever.
2. Lookalike audiences of the below (once you have enough data) -> In order of quality:
-> highest value previous customers
-> previous customers
-> purchased 180 days
-> added payment method
-> checkout initiated
-> viewed cart
-> added to cart
-> visited landing or product page
-> all website visitors
3. Broad targeting (no targeting at all) -> Only do this once your pixel has thousands of conversion events and knows exactly who your customer is.
MoFu (Warm Audiences - People who've shown some interest):
-> all website visitors (excluding purchasers)
-> visited landing or product page (excluding purchasers)
-> 50% video viewers (excluding purchasers)
BoFu (Hot Audiences - People who are close to converting):
-> added to cart (excluding purchasers)
-> initiated checkout (excluding purchasers)
-> viewed cart (excluding purchasers)
As you can see, Lookalikes of 'page engagers' isn't even on this list. It's because it's an unreliable, low-intent audience. You should start by focusing your energy on getting your detailed interest targeting right. This will generate the initial traffic and conversions (your first 100+ purchasers, for instance) that you need to build the high-quality retargeting and Lookalike audiences later on.
You probably should focus on the quality of your Lookalike source...
Let's go deeper on this because it's the core of your question and the key to scaling profitably. The performance of a Lookalike audience is 100% dependent on the quality of its source audience. Garbage in, garbage out.
Your goal is to create a source audience that contains only people who have performed the action you want more of. If you want more sales, you need a source audience of people who have already bought from you. If you want more high-quality leads, you need a source audience of your best past leads.
Look at the prioritised list I shared above. The Lookalike sources are ordered from highest intent (a previous customer) to lowest intent (a website visitor). When you have enough data (again, at least 100 conversions, but honestly you want more like 500-1000 for it to work really well), you should start building Lookalikes in that exact order.
Start with a 1% Lookalike of your 'Purchasers' list. This tells Meta: "Go and find the 1% of users in this country who look and behave most like the people who have already given me money." This is an incredibly powerful instruction. We've had so many clients see huge improvements from this one change. I remember one eCommerce client selling subscription boxes where we hit a 1000% return on ad spend, and a huge part of that was moving from broad interest targeting to highly specific Lookalikes based on their best customers.
Once that audience is working well, you can test a Lookalike of 'Add to Carts'. Then 'Website Visitors'. By testing them seperately, you can see which source provides the best performance and allocate your budget accordingly. For another B2B software client, we found that a Lookalike of people who completed a trial registration performed far better than a Lookalike of general website visitors, getting us a cost per registration of just $2.38. The source matters more than anything else.
You'll need a solid campaign structure to test this properly...
So, how do you put all this into practice? Your idea to A/B test is good, but you need a better structure to do it in a way that gives you clear answers. Don't just chuck two audiences into one campaign and hope for the best.
Here is a simplified version of the structure we use for many of our clients. It's designed for clarity and control.
Campaign 1: PROSPECTING (Objective: Conversions/Sales)
This campaign is for finding new customers (ToFu). Every ad set in here will target a different cold audience. You MUST exclude all your existing customers and recent website visitors from this campaign.
- Ad Set 1: Detailed Targeting - Theme A (e.g., targeting interests related to your direct competitors)
- Ad Set 2: Detailed Targeting - Theme B (e.g., targeting interests related to complementary products or publications your ideal customer follows)
- Ad Set 3: Lookalike 1% - Purchasers (Once you have enough data)
- Ad Set 4: Lookalike 1% - Add to Carts (Once you have enough data)
Campaign 2: RETARGETING (Objective: Conversions/Sales)
This campaign is only for people who already know you (MoFu/BoFu). The audience sizes will be smaller, but they should convert at a much higher rate. Here you only target your custom audiences of website visitors, video viewers, etc.
- Ad Set 1: Website Visitors 30 Days (Excluding anyone who added to cart or purchased) - The message here is about brand building or showing different products.
- Ad Set 2: Add to Cart / Initiated Checkout 14 Days (Excluding purchasers) - The message here is much more direct, maybe with a reminder or a small incentive to complete their purchase.
This structure gives you total control. You can clearly see which prospecting audience is bringing in the cheapest new customers. You can see how much it costs to bring back someone who abandoned their cart. You're not mixing cold and warm traffic, which gives the algorithm confusing signals. You set a budget for prospecting and a seperate, smaller budget for retargeting.
When you're testing, let each ad set run until it has spent at least 1-2x your target cost per acquisition (CPA). If your target CPA is £30 and an ad set has spent £60 without a single conversion, it's probably a dud. Turn it off and test something new in its place. This systematic approach is how you find winning audiences you can scale.
We'll need to talk about your offer and creative, too...
This is getting long, but I have to mention this. Even the best targeting in the world won't work if your ads and your offer are weak. The number one reason campaigns fail isn't actually the targeting, it's the offer. You could have the perfect audience dialled in, but if your ad doesn't grab their attention and make them an offer they can't refuse, they'll just scroll past.
You need a message that speaks directly to their pain. Don't sell the features of your product; sell the solution to their problem. Instead of "Our software has AI-powered analytics", try "Tired of guessing what's wrong with your marketing? Get a clear, actionable report in 2 minutes." It's the Before-After-Bridge. Show them the painful 'before' state, the happy 'after' state, and position your product as the bridge to get there.
And your call to action needs to be as low-friction as possible. For many B2B businesses, "Request a Demo" is a terrible offer. It screams "I'm going to waste an hour of your time trying to sell you something". Instead, offer something of genuine value for free. A free tool, a free checklist, an automated audit, a short video course. For one of our software clients, moving from a "Book a Demo" CTA to a "Start a Free Trial (no card required)" offer was transformative. It reduced their cost per lead and dramatically increased the quality, as people could see the value of the product for themselves. We've seen this reduce CPA for a medical SaaS client from £100 down to just £7. The offer is everything.
I know this is a huge amount of information to take in. Getting this right involves a lot of moving parts, from technical setup and audience research to campaign structure and creative testing. It's a full-time job.
I've detailed my main recommendations for you below as a starting point:
| Phase | Action Steps | Rationale |
|---|---|---|
| Phase 1: Foundation |
-> Ensure your Meta Pixel is installed correctly. -> Set up and verify all standard conversion events (ViewContent, AddToCart, InitiateCheckout, Purchase, Lead). -> Check the size of your current custom audiences to confirm they meet the 100-person minimum. |
Without accurate data tracking, none of the more advanced strategies will work. This is the bedrock of your entire advertising effort. |
| Phase 2: Audience Strategy |
-> Pause any campaigns using low-quality Lookalikes (e.g., from page engagers). -> Start by testing highly specific Detailed Targeting (interests/behaviours) in a new prospecting campaign. -> Once you have 100+ purchases/leads, create your first high-quality Lookalike audience (e.g., 1% of Purchasers). |
Focus your budget on what works. Start with interest targeting to gather data, then use that data to create powerful Lookalike and retargeting audiences. |
| Phase 3: Campaign Structure |
-> Create two seperate, long-term campaigns: one for Prospecting (cold audiences) and one for Retargeting (warm/hot audiences). -> In the Prospecting campaign, test each audience (interest group, Lookalike) in its own ad set. -> Ensure your Retargeting audiences exclude recent purchasers. |
This seperation gives you clear data, allows you to tailor messaging for different funnel stages, and gives you precise control over your budget allocation. |
| Phase 4: Optimisation |
-> Define your target Cost Per Acquisition (CPA). -> Monitor ad set performance daily. Turn off ad sets spending more than 1.5-2x your target CPA without results. -> Scale budget slowly (15-20% every few days) into winning ad sets. |
Paid advertising is not 'set and forget'. Constant, data-driven optimisation is required to cut wasted spend and scale your profitable campaigns. |
It's not just about pushing buttons in Ads Manager. It's about understanding the deep connection between your audience, your message, and your offer. It's about building a system, not just running an ad.
That's where professional help can make a huge difference. An experienced eye can spot opportunities and problems you might miss, and can implement these complex strategies far more quickly. If you'd like to go through your specific setup and build a proper growth strategy for your business, we offer a free initial consultation where we can review your account together and give you some concrete next steps.
Hope this helps you get started!
Regards,
Team @ Lukas Holschuh
