Sales & Conversion
Personas
Ecommerce
Time to ROI
Short-term (< 3 months)
When I first started managing Facebook Ads for ecommerce clients, lookalike audiences felt like magic. Feed Facebook your best customers' data, and it would find you thousands more just like them. The promise was seductive: let the algorithm do the heavy lifting while you focus on creative.
But after running campaigns for multiple Shopify stores and testing every variation of lookalike targeting, I discovered something that challenged everything I thought I knew about Facebook advertising. The results weren't just disappointing—they were consistently underperforming compared to a completely different approach.
This isn't another "Facebook Ads are dead" story. It's about a fundamental shift in how targeting actually works in 2025, and why the most successful campaigns I've run recently barely use lookalike audiences at all.
Here's what you'll discover:
Why lookalike audiences became less effective after iOS 14.5 changes
The creative-first targeting strategy that's replacing traditional audience building
How I restructured campaigns to focus on creative testing over audience segmentation
The simple framework that improved ROAS across multiple client accounts
When lookalike audiences still make sense (and when they don't)
Reality Check
What Facebook Ads gurus still preach about lookalike audiences
Walk into any Facebook Ads course or agency presentation, and you'll hear the same playbook repeated endlessly:
"Build your lookalike audiences like a pyramid." Start with 1% lookalikes of your best customers, then test 2-3%, maybe even 5-10% for broader reach. Layer in interest targeting. Create separate lookalikes for purchasers, email subscribers, and website visitors.
The conventional wisdom goes something like this:
Upload your customer list (preferably 1000+ customers)
Create 1% lookalike audiences in your target countries
Test different seed audiences (customers vs. email subscribers)
Scale the winners and exclude the losers
Rinse and repeat with different audience sizes
This approach made perfect sense in the pre-iOS 14.5 world. Facebook had rich data on user behavior, could track cross-device activities, and their lookalike algorithm was genuinely impressive at finding similar users.
But here's where this breaks down in 2025: Facebook's ability to accurately model lookalike audiences has been severely compromised. Privacy changes mean less data for training. Cookie restrictions limit tracking. And the "similar users" Facebook finds often share surface-level characteristics rather than actual purchasing intent.
Yet most advertisers are still following this playbook, wondering why their cost per acquisition keeps climbing while their audience "quality" seems to deteriorate.
Consider me as your business complice.
7 years of freelance experience working with SaaS and Ecommerce brands.
I was managing Facebook Ads for a B2C Shopify store that sold fashion accessories. Beautiful products, solid margins, perfect for social media advertising. Following best practices, I built an elaborate lookalike audience structure:
1% lookalikes of purchasers, 2% lookalikes of email subscribers, interest-based audiences layered with lookalike targeting. The setup looked textbook perfect.
For months, we were getting decent results. Not amazing, but acceptable ROAS around 2.5x. The client was happy enough, and I felt like I was doing solid work.
Then something forced my hand. Our main lookalike audience got permanently disabled. Facebook's system flagged it during one of their periodic policy reviews. No explanation, no appeal process—just gone.
Instead of rebuilding the lookalike audience immediately, I decided to test something different. What if I focused entirely on creative testing with broad audiences instead of trying to find the "perfect" lookalike match?
The theory was simple: if Facebook's targeting isn't as precise as it used to be, maybe the answer isn't better targeting—maybe it's better creative that naturally attracts the right people.
I had read about "creatives as the new targeting" but hadn't fully committed to the approach. This forced experiment became my testing ground for a completely different Facebook Ads philosophy.
Here's my playbook
What I ended up doing and the results.
Here's exactly what I implemented when I pivoted away from lookalike audience dependency:
The New Campaign Structure:
Instead of multiple campaigns with different lookalike audiences, I simplified everything into one core setup:
1 campaign with broad targeting (age and gender only)
Multiple ad sets with different creative angles
3 new creative variations launched every single week
The Creative Testing Rhythm:
This became the backbone of the entire strategy. Every week, without fail, I produced:
One lifestyle-focused creative (showing the product in use)
One problem-solving creative (addressing specific pain points)
One social proof creative (customer reviews or testimonials)
Each creative acted as its own "audience filter." The lifestyle creative attracted aspirational buyers. The problem-solving creative pulled in people actively seeking solutions. The social proof creative converted fence-sitters who needed validation.
Why This Works Better Than Lookalikes:
The insight that changed everything: Facebook's algorithm is still incredibly smart at optimizing for actions. It's just not as good at predicting who will take those actions based on demographic similarities.
By giving the algorithm diverse creative signals instead of restrictive audience parameters, I was essentially letting Facebook find the people who responded to each message—regardless of whether they "looked like" my existing customers.
The Weekly Testing Process:
Every Monday, I'd analyze the previous week's creative performance. Winners stayed, losers got paused, and three new tests went live. This constant rotation meant fresh content was always feeding the algorithm new signals about who was actually engaging and converting.
The client noticed the difference immediately. Not just in performance metrics, but in the quality of customers we were attracting. Comments on ads became more engaged. Customer lifetime value improved. We weren't just reaching more people—we were reaching better people.
Targeting Simplicity
Broad demographics only - let creative do the segmentation
Audience Refresh
3 new creative angles every week without fail
Algorithm Signals
Multiple creative types give Facebook better optimization data
Campaign Structure
One campaign, multiple ad sets, continuous creative rotation
The transformation was dramatic and measurable. Within 6 weeks of implementing the creative-first approach:
ROAS improved from 2.5x to 3.8x - significantly better than our previous lookalike audience campaigns.
More importantly, the consistency improved. Instead of the typical feast-or-famine pattern where some lookalike audiences would work brilliantly one week then completely fail the next, we maintained steady performance week over week.
The client's customer acquisition cost dropped by approximately 40%, and they started seeing higher-quality customers who made repeat purchases more frequently.
But the most telling result came when I A/B tested this approach against rebuilding our original lookalike audience strategy. The creative-first campaign outperformed the lookalike campaign by 60% in conversion rate and 45% in cost per acquisition.
This wasn't a fluke—I've since replicated these results across multiple client accounts in different industries.
What I've learned and the mistakes I've made.
Sharing so you don't make them.
Creatives are the new targeting - Your message determines your audience more than demographics ever will
Simplicity beats complexity - Broad targeting with great creative outperforms elaborate audience hierarchies
Consistency is everything - Weekly creative testing trumps monthly audience optimization
Quality over quantity - Three well-thought-out creative angles beat twenty random variations
Let the algorithm learn - Facebook is better at finding converters than predicting them
Focus on signals, not segments - Give Facebook diverse creative signals instead of restrictive audience boxes
Test the fundamentals - Sometimes the biggest breakthroughs come from questioning basic assumptions
If I were starting fresh today, I'd spend 80% of my time on creative strategy and 20% on targeting, rather than the traditional 50/50 split most advertisers still use.
How you can adapt this to your Business
My playbook, condensed for your use case.
For your SaaS / Startup
For SaaS companies looking to implement this approach:
Focus creative angles on different use cases rather than user types
Test problem-focused vs solution-focused messaging weekly
Use testimonials from different company sizes as creative variations
For your Ecommerce store
For ecommerce stores adopting creative-first targeting:
Rotate between lifestyle, product-focused, and social proof creatives
Test seasonal messaging alongside evergreen product benefits
Use customer photos and reviews as regular creative content