Facebook Lookalike Audiences for E-commerce: Complete Setup Guide (2026)

Key Takeaways

• Value-based Lookalike Audiences from high-LTV customers deliver 20-40% better ROAS than basic purchase audiences

• Start with 1% Lookalikes for budgets under £3,000 monthly, expand to 3-5% only when 1% proves stable

• Segment customers by total spend (top 25%) and repeat purchases (2+ orders) for strongest seeds

• Focus product Lookalikes on high-margin items, not highest-volume SKUs

• Always exclude existing customers and source audiences from prospecting campaigns

• Value-based Lookalikes need 1,000+ customers and 14-21 days minimum before judging performance

Lookalike Audiences are Meta's algorithmic targeting tool that finds new customers who share characteristics with your existing best customers. For e-commerce brands running Facebook Ads, this means moving beyond generic targeting to attract high-value shoppers who actually convert.

Why Most Shops Get Lookalike Audiences Wrong

Most online shops build Facebook Lookalike Audiences from every purchaser. Bigger seed, better results? Wrong.

Here's what actually happens: your "all purchasers" seed gets dominated by one-time discount hunters, clearance buyers, and returned orders. Meta finds more people who behave exactly like that - common e-commerce marketing mistakes.

We've seen fashion shops with 5,000-person purchase audiences getting £1.80 back per pound. Competitors with 1,200 high-value repeat buyers hit £4.20+ ROAS. Quality beats quantity.

This guide shows you how to build profitable Facebook Lookalike Audiences for e-commerce based on lifetime value, repeat purchases, and margins for Shopify, WooCommerce, and UK e-commerce brands.

Building Value-Based Audiences from Customer Data

Value-based audiences target people like your best customers, not your average ones. This is where real profit lives.

Export customers from Shopify or WooCommerce with email, phone, country, total spent, and order count. Filter for £500+ total spend or 2+ orders.

Calculate lifetime value per customer. Sort by total spend and grab your top 25%. Create a CSV with email, phone, country, and customer_value column.

Upload as a customer list in Meta Events Manager. Navigate to Audiences, create Custom Audience, select Customer List, map columns. The customer_value field tells Meta which behaviours matter most.

Build a 1% Lookalike for your primary market. You need 1,000+ customers minimum. Below that, performance suffers.

Worth creating separate seeds for customers with £750+ LTV and 3+ orders, subscription buyers, or multi-category purchasers.

Product-Specific and Catalogue Lookalikes

Someone who buys designer handbags behaves differently from someone buying budget accessories. Product-level targeting captures this.

Your Meta Pixel needs proper events: ViewContent, AddToCart, Purchase with content_ids, value, and currency matching your Commerce Manager catalogue.

Build Custom Audiences for buyers of specific categories or price tiers. In Events Manager, create audiences based on Purchase events with conditions like content_ids contains "dresses" or value greater than £150. Accumulate 90-180 days of data.

Create 1% Lookalikes from high-margin items and hero products that attract your best customers, not highest-volume sellers. A £200 coat with 65% margin beats a £20 T-shirt with 30% margin.

Match creative to Lookalikes. Dress ads to dress-buyer Lookalikes. Message matching drives conversion rates up 25-40%.

Three to five product Lookalikes work better than twenty micro-segments. Too many small audiences prevent learning.

Lookalike Percentage Strategy (1%, 3%, 5%)

Right, percentages. 1% targets the 1% of your market most similar to your seed, 5% targets the top 5%. Concentration matters.

Start with 1% Lookalikes when monthly budget per country sits below £3,000. At this spend, 1% audiences (500,000-700,000 people in UK) give you best concentration.

Test 2-3% only after 1% spends most budget at target ROAS and frequency stays reasonable.

Reserve 5%+ for mature brands with large budgets. Most e-commerce brands waste money with 5% Lookalikes in Facebook Ads chasing low-intent prospects.

Multiple narrow Lookalikes typically outperform single broad ones. Three separate 1% audiences usually deliver better ROAS.

Source Quality and Exclusions

Seed quality matters more than size. 1,200 customers with £800+ LTV outperform 5,000 random purchasers.

Aim for 1,000-50,000 people per seed. Below 1,000, Meta lacks data points. Above 50,000, you're diluting the signal.

Use 90-180 day windows for recency. Behaviour shifts quite rapidly in fashion and seasonal categories.

Combine purchases with high-intent actions when volume is low. If you've got 600 purchasers, add people who initiated checkout over £100 in 90 days.

Exclude existing customers from prospecting campaigns. Nothing burns budget faster. Create a customer list audience and exclude it everywhere.

Exclude source audiences too. Don't waste impressions on people in your seed.

Common mistakes: all-purchaser seeds dominated by discount hunters, 365-day windows with ancient data, ignoring refunds. Clean data builds better audiences.

Campaign Structure for Lookalike Audiences

Structure campaigns so value-based and product-specific Lookalikes sit separately for clean measurement.

Allocate 50-60% of prospecting budget to your strongest value-based 1% Lookalike, 20-30% to product-specific Lookalikes, 10-20% to tests.

Start with "highest volume" or cost-cap bidding. Switch to ROAS goals after 50+ weekly conversions.

Value-based Lookalikes need 14-21 days minimum. Volume starts lower because Meta's finding higher-quality prospects. Don't kill campaigns after three days.

Creative rotation matters more than most people realise. Cost per click jumps 15-30%, conversions drop up to 50% when creative goes stale.

Use product-specific landing pages. If your ad promises 20% off trainers, show exactly that.

Monitor frequency closely. Above 3 weekly impressions, performance drops. Expand creative, adjust budget, or test the next percentage.

Platform Integration: Shopify and WooCommerce

In Shopify, navigate to Customers, filter for total spent above threshold or order count above 2, export to CSV. Add customer_value column, upload to Meta.

WooCommerce needs WP All Export to pull customer data with order totals. Filter for order count ≥2 or total value over £500.

Map columns carefully. Email to email, phone to phone, country to country, customer_value to "value". Get this wrong and optimisation fails.

Native integrations automate refreshes. Shopify's Meta integration updates customer lists automatically. For WooCommerce, Zapier pushes updated segments on schedule.

GDPR compliance matters. Privacy policy must state you use customer information for marketing.

Refresh segments monthly. Customer behaviour changes, new high-LTV buyers appear, old customers churn.

Common Mistakes That Kill Performance

Building seeds from all purchasers instead of high-value segments. Include discount hunters, get more discount hunters.

Judging value-based Lookalikes too quickly. Give them 21 days minimum with £50+ daily. We've seen Lookalikes that looked terrible day 5 deliver brilliant ROAS by day 20.

Creating too many tiny product Lookalikes. Twenty audiences with £10 daily each prevents learning. Consolidate into three to five strong segments.

Not excluding existing customers. Exclude customer lists from every prospecting campaign, without exception.

Using stale data. Refresh value segments monthly using 90-180 day windows.

Forgetting to match creative to audience. Product-specific audiences need product-specific creative.

Expanding percentage too early. Prove 1% before broadening. Quality over volume.

Measuring E-commerce Profitability from Lookalike Audiences

Platform dashboards lie. Always have, probably always will. Meta reports ROAS based on attribution windows that inflate performance 2-5x.

Calculate true breakeven ROAS. If you keep 35p per pound after all costs, you need £2.86 back just to break even on your Facebook Ads spend.

Track blended ROAS across all channels. Value-based Lookalikes might show £3.50 in Meta whilst driving £4.80 blended.

Use post-purchase surveys. Ask "How did you first hear about us?" The gap between customer answers and Meta's claims reveals attribution inflation.

Monitor 90-day customer lifetime value by source. Value-based Lookalike customers should show 25-50% higher LTV than broad prospecting, reducing your customer acquisition cost significantly.

Set up Conversions API for server-side tracking. Browser-based Pixel misses 20-40% of conversions due to iOS privacy and ad blockers.

Profit beats vanity metrics. Do the proper maths.

Caio for now

Ali Puglianini

Frequently Asked Questions

What are Facebook Lookalike Audiences?

Facebook Lookalike Audiences use Meta's algorithm to find new people who share characteristics with your existing customers. You provide a source audience (like high-value purchasers), and Meta identifies similar users across Facebook and Instagram.

How many customers do I need for Lookalike Audiences?

Minimum 1,000 people in your seed. Below this, Meta lacks data points to identify patterns. If you've got 600 customers, combine high-value purchasers with high-intent actions to reach 1,000.

Should I start with 1% or 5% Lookalikes?

Always start with 1% for budgets under £3,000 monthly. The 1% represents closest match to your seed - roughly 500,000-700,000 people in UK. Only expand after 1% spends most budget at ROAS target.

How long before judging Lookalike performance?

Give value-based Lookalikes 14-21 days minimum with £50+ daily budget. Meta needs time to find high-value patterns. Campaigns that look poor after 5 days often deliver excellent ROAS by day 20.

What ROAS improvement should I expect?

Value-based Lookalikes from high-LTV and repeat customers typically deliver 20-40% better ROAS than basic "all purchasers" Lookalikes, once the seed has 1,000+ customers and 14 days of data. Expect lower volume but higher average order value - you're targeting quality over quantity.

How often should I refresh seed audiences?

Monthly minimum. Customer behaviour changes, new high-LTV buyers emerge, older customers churn. Build seeds using 90-180 day windows, then re-upload to Meta every 30 days.