TLDR;
- Your Lookalike audience isn't in 'Saved Audiences' because it's a 'Custom Audience'. You select it from a different field in the ad set settings. It's a simple fix, but it's masking a much bigger strategic issue.
- Stop A/B testing a cold interest audience against a Lookalike. It's an unfair fight and a waste of money. The real goal is to test audiences at the same stage of the funnel against each other to find winners.
- The most important advice is to prioritise your audiences based on intent. Start with your hottest audiences first (retargeting cart abandoners - BoFu), then warm audiences (website visitors - MoFu), and only then cold audiences like interests and Lookalikes (ToFu).
- Your cost per lead is less important than your customer lifetime value (LTV). This letter includes an interactive LTV calculator to help you figure out how much you can actually afford to spend to acquire a customer.
- The quality of your offer is probably the number one reason your ads will succeed or fail, far more than the specific audience you choose. We'll break down how to create an offer that actually works.
Hi there,
Thanks for reaching out!
You mentioned you're having trouble finding your Lookalike audience, and it’s a really common point of confusion in Meta's interface, so you're definitely not alone. It's an easy enough fix on a technical level, which I'll walk you through.
But honestly, I think the real issue isn't where to click in Ads Manager. The bigger opportunity here is to step back and look at the whole strategy behind *how* you're choosing and testing your audiences. Getting this right is the difference between burning through cash and building a predictable, scalable advertising machine. What you're trying to do with your A/B test is a classic mistake I see all the time, and it's probably costing you more than you realise.
So, I wanted to give you some of my thoughts, based on running thousands of these campaigns, that go a bit deeper than just a quick technical answer. We'll cover the quick fix first, but then we'll get into the stuff that actually moves the needle: how to structure your testing, how to prioritise your audiences properly, and how to make sure you're aiming at the right people in the first place.
We'll need to look at the technicals first... Why you can't find your Lookalike
Alright, let's get the simple bit out of the way. The reason you can't select your Lookalike from the "use saved audience" dropdown is because of how Meta categorises audiences. It's not a rule about where they can be used, just a confusing bit of labelling on their part.
There are three main types of audiences:
1. Custom Audiences: These are audiences made from your *own* data sources. This includes people who have visited your website (from your Pixel), your customer email lists, people who have engaged with your Facebook or Instagram page, and, crucially, Lookalike Audiences which are built from these source custom audiences.
2. Saved Audiences: This is just a shortcut. It's a combination of detailed targeting (interests, behaviours, demographics), locations, and age ranges that you have saved to reuse later. You're basically just saving a *configuration*, not a list of specific people.
3. Special Ad Audiences: These are for specific categories like housing or credit and likely don't apply here.
So, your Lookalike audience is a Custom Audience, not a Saved Audience. When you're in your ad set settings, you don't use the 'Use Saved Audience' dropdown. Instead, you go to the 'Audience' section and click into the box that says 'Add Custom Audiences or Lookalike Audiences'. Your Lookalike, which is 'ready', will appear in the list there for you to select. Simple as that.
That solves your immediate problem. But please, don't just go and set up that A/B test. The test itself is fundamentally flawed, and that's what we need to talk about next.
I'd say you are thinking about A/B testing all wrong...
Your plan is to test a detailed targeting audience against a Lookalike audience. On the surface, this sounds logical. But in practice, it's like putting a heavyweight boxer in the ring with a welterweight and then being surprised when the heavyweight wins. It doesn't tell you anything useful.
A Lookalike audience, even one based on a relatively low-intent source like page engagers, is *supposed* to be better than a cold, broad interest group. It's built from people who have already shown some affinity for your brand. Meta's algorithm has analysed them and gone out to find millions of other people who share similar characteristics. Your detailed targeting audience is just a big bucket of people who happen to like 'eCommerce' or whatever you've chosen. It's a completely different level of qualification.
So when you run this test, the Lookalike will probably "win" (get a lower cost per result). But what have you learned? You've just confirmed that a slightly warmer audience performs better than a completely cold one. You haven't discovered a new, scalable winning audience. You've just spent money to prove a basic marketing principle.
Effective testing isn't about pitting apples against oranges. It's about testing apples against other apples to find the tastiest one. A better test would be:
- -> Test Lookalike A (based on page engagers) vs. Lookalike B (based on website visitors).
- -> Test Interest Stack A (e.g., Shopify + WooCommerce) vs. Interest Stack B (e.g., eCommerce publications + gurus).
- -> Test the same winning audience with Creative A vs. Creative B.
This is how you get actionable data. You isolate one variable and find out what truly works. The way you're approaching it now is a bit of a dead end. To do this properly, you need a framework for thinking about your audiences. You need to stop thinking about them as one big pool and start thinking about them in terms of temperature and intent.
You probably should prioritise audiences by their quality, not just their size...
When I take on a new client, the first thing I do is audit their account structure. Nine times out of ten, it's a mess of ad sets all competing against each other, with budget spread thinly across audiences of vastly different quality. The key to successful scaling is to structure your campaigns around the marketing funnel: Bottom of Funnel (BoFu), Middle of Funnel (MoFu), and Top of Funnel (ToFu).
Think of it as a hierarchy of value. You should always allocate your budget to the most valuable audiences first, and only once you've maxed them out should you move to the next level down. Your entire testing strategy, and indeed your entire account, should be built around this principle.
BoFu (Bottom of Funnel) - Your Hottest Audience
These people are on the verge of converting. They know you and have shown strong buying intent. This is where your highest ROAS will come from. Fund this first!
- ✅ Added to Cart (last 7-14 days)
- ✅ Initiated Checkout (last 7-14 days)
- ✅ Previous Customers (for upsells/cross-sells)
MoFu (Middle of Funnel) - Your Warm Audience
They are aware of you but aren't ready to buy yet. The goal here is to nurture them and bring them back to your site to take action.
- ✅ All Website Visitors (last 30-90 days)
- ✅ Viewed specific Product Pages (last 30 days)
- ✅ Watched 50%+ of your Video Ads (last 90 days)
- ✅ Engaged with IG/FB Page (last 90 days)
ToFu (Top of Funnel) - Your Cold Audience
These people have never heard of you. This is where you find new customers, but it's the most expensive and lowest converting part of the funnel. Your current test lives here.
- ✅ Detailed Targeting (Interests, Behaviours)
- ✅ Lookalike Audiences (from your best sources)
- ✅ Broad Targeting (no targeting, for mature accounts)
So, where does your Lookalike of "engaged users" fit in? It's a ToFu audience. It's designed to find *new* people. The mistake is treating all ToFu audiences equally. A Lookalike based on your list of highest-value previous customers is infinitely more powerful than a Lookalike based on people who liked a post. You should always build Lookalikes from the highest-intent source you have available. Here's the pecking order I use for my clients, from best to worst:
- Lookalike of Highest Value Customers: People who have spent the most with you.
- Lookalike of All Customers: Your full purchaser list.
- Lookalike of Initiated Checkouts: People who almost bought.
- Lookalike of Adds to Cart: Strong interest.
- Lookalike of Website Visitors: General interest in your brand.
- Lookalike of Video Viewers: Engaged with your content.
- Lookalike of Page Engagers: The lowest quality source, but still better than nothing.
You need at least 100 people in a source audience to create a Lookalike, but realistically you need closer to 1,000 for it to be truly effective. So, start by getting your pixel tracking perfect. Once you have enough purchasers, create a Lookalike from that list. That's a ToFu audience worth testing properly and spending money on. Your current Lookalike is a good starting point if that's all you have, but you need a plan to graduate to better sources as you gather data. But even with the best audience in the world, you'll fail if you don't know who you're really talking to.
You'll need to define your customer by their pain, not their interests...
This is probably the most profound shift you can make in your advertising. Most people build an Ideal Customer Profile (ICP) that looks like a census form: "Female, 35-54, lives in London, interested in Yoga, earns £70k+". This is utterly useless. It tells you nothing about *why* they would buy from you. It leads to generic ads that try to speak to everyone and end up resonating with no one.
To write ads that stop the scroll and compel action, you need to stop thinking about demographics and start obsessing over your customer's *nightmare*. What is the specific, urgent, expensive, or emotionally draining problem they are dealing with right now that your product or service solves? Your ICP isn't a person; it's a *problem state*.
Let's take an example. Say you sell a high-end project management tool for creative agencies.
| Aspect | ❌ Useless Demographic ICP | ✅ Powerful Nightmare ICP |
|---|---|---|
| The Profile | Project Managers, aged 28-45, working in agencies with 20-100 employees. | Sarah, a Head of Operations at a 50-person design agency. She's overwhelmed, working 60-hour weeks, and terrified of losing her biggest client because her team keeps missing deadlines. She feels like a firefighter, not a manager. |
| The Pain | Needs better workflow organisation. | The deep-seated fear of professional failure. The constant anxiety from client emails marked 'URGENT'. The frustration of her best designers threatening to quit because of the chaos. |
| Targeting (Interests) | "Project Management", "Asana", "Marketing Agencies". (Too broad, full of low-level employees) | People who follow agency growth experts like David C. Baker, members of Facebook groups like 'Agency Owners & Entrepreneurs', people who list 'Head of Operations' as a job title AND are admins of a Facebook business page. (Hyper-specific) |
| Ad Copy Angle | "The Best Project Management Tool for Agencies. Try it now!" (Generic, feature-focused) | "Tired of chasing your team for updates? Is your biggest client one missed deadline away from walking? Stop managing chaos. Get the clarity to lead. Our platform gives creative agencies one source of truth, so you can go back to being a manager, not a firefighter." (Problem-focused, emotional) |
Do you see the difference? The Nightmare ICP gives you everything. It tells you what targeting to use, what language to use in your ads, and what pain points to highlight on your landing page. You're no longer selling software; you're selling Sarah a good night's sleep. You're selling her the feeling of control and the respect of her clients and team.
Before you spend another pound on ads, sit down and do this exercise. Write out a one-page document describing your ideal customer's nightmare in excruciating detail. This document is more valuable than any audience research tool you can buy. Once you know their pain, you can start to think about what a solution is actually worth to them, which brings us to the most important metric you probably aren't tracking.
You'll need to learn the math that actually matters...
Beginners in paid ads obsess over metrics like CPC (Cost Per Click) and CPL (Cost Per Lead). They try to get them as low as possible. This is a trap. It leads you to prioritise cheap, low-quality traffic from developing countries or broad, untargeted audiences. The real question isn't "How low can my CPL go?" but rather, "How high a CPL can I afford to pay to acquire a high-quality customer?"
The answer lies in calculating your Customer Lifetime Value (LTV). This tells you the total profit you can expect to make from a single customer over the entire course of your relationship. Once you know this number, everything else falls into place.
Here’s the basic formula:
LTV = (Average Revenue Per Account/Customer * Gross Margin %) / Monthly Churn Rate %
- Average Revenue Per Account (ARPA): How much does a typical customer pay you per month or per year?
- Gross Margin %: What's your profit margin on that revenue after deducting the cost of goods sold (COGS)? For SaaS or digital products, this is often very high (80-95%). For physical products or services, it will be lower.
- Monthly Churn Rate %: What percentage of your customers do you lose each month? (If you have annual churn, just divide by 12).
Let's run a scenario for a hypothetical SaaS business:
- ARPA: £150/month
- Gross Margin: 90% (0.90)
- Monthly Churn: 5% (0.05)
LTV = (£150 * 0.90) / 0.05
LTV = £135 / 0.05 = £2,700
This means each new customer is worth £2,700 in gross profit to the business. Now you can make intelligent decisions. A healthy business model often aims for a 3:1 LTV to CAC (Customer Acquisition Cost) ratio. This means you can afford to spend up to £900 (£2,700 / 3) to acquire a single new customer.
Suddenly, that £50 CPL from a hyper-targeted LinkedIn campaign doesn't look so expensive, does it? If your sales process converts 1 in 10 of those leads into a customer, your CAC is £500 (£50 * 10), which is well below your £900 maximum. You have a profitable, scalable engine. Without knowing your LTV, you'd probably have turned that campaign off, thinking it was "too expensive."
To make this real for you, I've built a simple calculator. Play around with your own numbers and see what your LTV and max affordable CAC are. This will change how you think about your ad budget forever.
Interactive Customer Lifetime Value (LTV) Calculator
This is the main advice I have for you:
We've covered a lot of ground, from a simple technical fix to a complete strategic overhaul. It can feel like a lot, I know. The key is to stop thinking in terms of isolated tactics ("how do I A/B test this?") and start thinking in terms of a cohesive system where your audience, message, and offer all work together. An amazing offer to a mediocre audience will often outperform a mediocre offer to a perfect audience.
I remember one B2B software client who came to us with "failing ads". They were targeting the right people, but their offer was just 'Request a Demo'. It was high-friction and low-value. We helped them shift their offer to a free, automated 'Data Health Check' tool that solved a small, tangible problem for their ICP. Their conversion rates tripled overnight, not because we found a magical new audience, but because we fixed the offer. The number one reason campaigns fail is the offer. It's not offering enough value, or it's being shown to an audience that has no need for that value.
Here's a summary of the main points and my recommendations for you to implement:
| The Problem | Why It's Hurting You | My Recommended Action |
|---|---|---|
| Flawed A/B Testing | You're testing audiences from different funnel stages against each other, which gives you no actionable data and wastes ad spend. | Pause your current test. Only test "like for like". Test two different Lookalike sources against each other, or two different interest groups. Isolate one variable at a time. |
| Poor Audience Prioritisation | You're starting with cold ToFu audiences, which are the most expensive and lowest converting, instead of focusing on the low-hanging fruit. | Restructure your account. Create separate, always-on campaigns for BoFu (retargeting) and MoFu (nurturing). Fully fund these first before spending anything on ToFu (prospecting). |
| Low-Quality Lookalike Source | Your Lookalike is based on "page engagers", which is a very low-intent signal. The quality of your Lookalike is only as good as its source. | Focus on data collection. Make sure your pixel is firing correctly on all key conversion events. As soon as you have 100+ purchasers, create a Lookalike from that source. This will be your most powerful prospecting audience. |
| Vague ICP Definition | Without a deep understanding of your customer's 'nightmare', your targeting will be too broad and your ad copy will be too generic to resonate. | Create a 'Nightmare ICP' document. Forget demographics. Detail the specific, urgent, and emotional pain your ideal customer is in. Use this to guide all future targeting and creative decisions. |
| Focusing on CPL, not LTV | You're likely trying to minimise your cost per result without knowing what a customer is actually worth, causing you to turn off potentially profitable campaigns. | Calculate your LTV. Use the calculator provided to understand your unit economics. This will give you a clear ceiling for your Customer Acquisition Cost and empower you to invest confidently in higher-quality, more 'expensive' traffic that is actually profitable in the long run. |
I know this is a fundamental shift from just trying to get a campaign live. But this strategic approach is what separates the businesses that waste a fortune on ads from those that build a reliable growth engine. It takes more work upfront, but it pays off exponentially in the long run.
Implementing all of this correctly—setting up the account structure, dialing in the tracking, writing copy that speaks to the 'nightmare', and constantly optimising based on the right metrics—is a full-time job. It's what we do all day, every day. We've helped SaaS companies reduce their CPA from £100 down to £7 and generated over $115k in revenue for course creators in under two months by applying these very principles.
If you've found this breakdown helpful and feel you might benefit from having an expert take a direct look at your account, that's something we can do. We offer a completely free, no-obligation 20-minute strategy session where we'll go through your campaigns and give you a tailored action plan. It's often the quickest way to get clarity and find the biggest opportunities for growth.
Either way, I hope this detailed breakdown gives you a much better framework for success.
Regards,
Team @ Lukas Holschuh
