Building Lookalike Audiences From Your Actual Best Customers
- Why Most Lookalike Audiences Underperform
- Segmenting Your Customer Base for Lookalike Source Lists
- The Right Audience Size for Lookalike Campaigns
- Combining Lookalikes With Interest Targeting: When to Layer
- AskBiz Customer Cohorts as Lookalike Source Lists
- Measuring Lookalike Performance: Beyond Click-Through Rate
Most SMBs build lookalike audiences from their entire customer list. The right approach: build from your top 10-15% of customers by LTV. This single change typically reduces CAC by 25-40% for cold audience campaigns.
- Why Most Lookalike Audiences Underperform
- Segmenting Your Customer Base for Lookalike Source Lists
- The Right Audience Size for Lookalike Campaigns
- Combining Lookalikes With Interest Targeting: When to Layer
- AskBiz Customer Cohorts as Lookalike Source Lists
Why Most Lookalike Audiences Underperform#
Lookalike audiences are one of Meta's most powerful targeting tools — the ability to tell the algorithm "find me more people who look like my best customers" is genuinely valuable. But most SMBs implement them in a way that dramatically limits their effectiveness. The most common mistake: uploading your entire customer list as the source audience for your lookalike. If you have 2,000 customers on your list and 400 of them are one-time bargain hunters who bought on a promotional offer and never returned, while 200 of them are loyal advocates who buy regularly at full price, your lookalike audience is averaging these groups together. The algorithm is told to find people who look like a mix of your best and worst customers — predictably mediocre results. The fix is almost embarrassingly simple once you see it: segment your customer list by value before creating lookalikes. Your top 10-15% of customers by total lifetime spend (or by number of purchases, or by 12-month revenue) form a much tighter, more useful source audience. A lookalike of your top 200 customers will consistently outperform a lookalike of your top 2,000.
Segmenting Your Customer Base for Lookalike Source Lists#
Effective lookalike source list creation requires sorting your customer database by value. The metrics to use depend on your business type. For ecommerce businesses, total lifetime spend or total number of orders over the last 12 months works best. For subscription businesses, current monthly recurring value. For appointment-based businesses (salons, clinics, gyms), average booking value multiplied by visit frequency. AskBiz pulls this data from your POS transaction history and exports it in the exact format required by Meta's Custom Audience upload: email, phone number, first name, last name, date of birth (if captured). The export is formatted to Meta's hashing requirements automatically — personally identifiable information is hashed before upload, so the actual customer data never leaves the upload process in readable form. Create four distinct source lists for testing: top 5% of customers by LTV (your absolute best — small list but tightest signal); top 15% of customers by LTV; top 30% of customers by LTV; and new customers only (those who've made a first purchase in the last 90 days — good for finding new-to-brand prospects rather than replicating your most loyal customers). Test lookalikes at 1-2% audience size (most similar to source) from each source list in separate ad sets to determine which source audience drives the best CAC.
The Right Audience Size for Lookalike Campaigns#
Meta lookalike audiences range from 1% to 10% of the target country's population. A 1% lookalike in the UK represents approximately 430,000 people — those most similar to your source audience. A 10% lookalike represents 4.3 million people — similar in broad category terms but much more loosely matched. For most SMBs, the sweet spot is 1-2% lookalike audiences for acquisition campaigns. The 1% audience is small enough that it'll saturate quickly at moderate budgets (you'll hit frequency limits within 2-4 weeks at £2,000/month spend), so maintaining campaign freshness requires either expanding to 2% or refreshing the source list regularly. When you need scale beyond 2%, rather than expanding a single lookalike to a wider percentage, it's often better to create multiple 1-2% lookalikes from different source audiences (your top customers, your new customers, your high-frequency buyers) and run them in separate ad sets. This gives you scale while maintaining the quality signal of tighter source lists. AskBiz can automate the monthly refresh of your source lists — as new customers enter your top 15%, the export updates automatically and the lookalike retrains.
Combining Lookalikes With Interest Targeting: When to Layer#
Meta offers two main approaches for cold audience targeting: interest/behaviour-based targeting (targeting people based on what they've liked, followed, or engaged with) and lookalike audiences (targeting people who resemble your existing customers). These can be used separately or in combination. For SMBs with small source lists (fewer than 500 customers in the upload), interest layering on top of a lookalike can help maintain quality while restricting the audience size. For example, a UK craft beer brand might target a 2% lookalike of their top customers and also require that those people follow beer-related pages or have shown interest in craft beverages. This narrows the audience but improves the signal quality when the source list is thin. For businesses with 1,500+ customers in their source list, clean lookalike audiences (no interest layering) typically outperform layered combinations. The Meta algorithm has enough signal from a strong source list that additional interest constraints just reduce reach without improving relevance. Test both approaches with your specific source list size.
AskBiz Customer Cohorts as Lookalike Source Lists#
Beyond simple LTV segmentation, AskBiz enables more sophisticated source list creation based on customer behaviour patterns. Some of the most effective lookalike source lists aren't just "highest spending" — they're customers who exhibit specific purchase behaviours that correlate with long-term value. For example: customers who made a second purchase within 30 days of their first (strong early loyalty signal); customers who buy across multiple product categories rather than being single-category purchasers; customers who have never used a promotional code (full-price buyers tend to be higher LTV); and customers who were originally acquired through referrals (referred customers typically have 25-30% higher LTV than acquired customers). AskBiz can segment your customer database by any of these behavioural criteria and export the resulting lists for Meta Custom Audience upload. For a UK fashion retailer with 1,800 customers total, running a lookalike from the 240 customers who've bought across at least three different product categories might outperform a straight LTV-based lookalike — because multi-category buyers are genuinely more engaged with the brand. Test it: the upload takes five minutes.
Measuring Lookalike Performance: Beyond Click-Through Rate#
The mistake most SMBs make when evaluating lookalike audiences is measuring click-through rate and immediate conversion rate. These metrics favour your existing retargeting audiences (people who already know you) over cold lookalike audiences (people discovering you for the first time). Comparing them directly leads to the wrong conclusion that lookalikes are underperforming. The right metrics for lookalike campaigns: CAC (how much are you paying per new customer); 30-day and 90-day LTV of acquired customers; and new-to-brand percentage (what proportion of converters are genuinely new customers rather than existing customers who appeared in the lookalike). AskBiz tracks all three through its integration with your Meta Ads account and POS transaction history. Best-in-class lookalike campaigns for UK SMBs deliver CAC of 20-40% below broad interest targeting with comparable creative. If your lookalike CAC isn't at least 15% below your interest targeting CAC, either your source list needs refinement or your creative isn't optimised for cold audiences. Try both before drawing conclusions. AskBiz connects your ads to actual sales. Try free at askbiz.co and build your first LTV-segmented lookalike audience this week.
People also ask
How do I create a Facebook lookalike audience from my customers?
Effective lookalike source list creation requires sorting your customer database by value. The metrics to use depend on your business type. For ecommerce businesses, total lifetime spend or total number of orders over the last 12 months works best.
What percentage should I use for Facebook lookalike audiences?
Meta lookalike audiences range from 1% to 10% of the target country's population. A 1% lookalike in the UK represents approximately 430,000 people — those most similar to your source audience.
Why is my Facebook lookalike audience not converting?
Meta offers two main approaches for cold audience targeting: interest/behaviour-based targeting (targeting people based on what they've liked, followed, or engaged with) and lookalike audiences (targeting people who resemble your existing customers).
How many customers do I need for a good Facebook lookalike audience?
Beyond simple LTV segmentation, AskBiz enables more sophisticated source list creation based on customer behaviour patterns.
Should I use my whole customer list or my best customers for a lookalike audience?
The mistake most SMBs make when evaluating lookalike audiences is measuring click-through rate and immediate conversion rate.
Our team combines expertise in data analytics, SME strategy, and AI tools to produce practical guides that help founders and operators make better business decisions.
AskBiz connects your ads to actual sales. Try free at askbiz.co
AskBiz connects to your existing tools and surfaces insights like these automatically — no spreadsheets, no analysts, no waiting.
Connects to Shopify, Xero, Amazon, QuickBooks, Stripe & more in minutes