Facebook Ad Creatives Guide: What I've Learned From Burning $500k on Tests

The creative formats that consistently win, the ones I've stopped making, and the production system that scales without losing quality.

Creative is 70-80% of paid social performance. You can have perfect targeting, perfect bidder, perfect LP — if the creative doesn't hook, none of it matters. After running thousands of variations across accounts, here's what consistently moves results.

The format hierarchy that actually wins

Across e-commerce, SaaS, lead-gen, and app accounts, the same five formats keep producing winners:

1. UGC review video (9-15s, vertical) — looks like a regular user posting; works in 80% of accounts
2. Founder POV video — talking head explaining the problem and solution; great for SaaS and education
3. Demo screencast — actual product in action; SaaS, apps, tools
4. Static image with bold text — works for high-volume testing because of fast production
5. Carousel with 3-5 cards — best for products with multiple SKUs or feature comparisons

What I've stopped making

High-production "commercial" video. Looks like an ad. Users skip ads. UGC outperforms 2-3x at 1/10 the cost.

Logo-first openings. Lose 80% of viewers in 3 seconds.

Generic stock photos. Convert worse than a phone-shot photo of the actual product.

"Solutions for modern teams" abstract imagery. Means nothing to anyone.

Long demo videos (60+ seconds) for cold traffic. Cold viewers don't earn 60 seconds of attention. Tighten or save for retargeting.

The hook structure

Every winning creative I've seen has the same first 3 seconds:

1. Visual interrupt — unexpected motion, color, or angle
2. Audience identifier — text or speech that says who this is for
3. Promise of value — what they'll get if they keep watching

Example for a SaaS tool: cut to UGC of someone holding their phone showing 47 unread Slack messages, on-screen text "If your team has 8+ Slack channels, this is for you," voiceover starts "Here's how I cut my notifications by 70%."

Production at scale

Single creative is fragile. To run any account >$5k/day reliably, you need a pipeline producing 5-15 new variations per week.

How I structure it:

• One creative concept per week (the "big idea")
• 5-8 variations of that concept (different hooks, different opening shots)
• Multiple cuts per variation (vertical for Reels, square for Feed)
• Iteration on winners — take last week's top performer, make 3 variations

Working with UGC creators

For most accounts, hiring 2-3 UGC creators on retainer beats trying to make everything in-house.

• Rates: $80-300 per video for established creators
• Brief them with: hook angle, key product points, do/don't list
• Let them speak naturally — the unscripted feel is the value
• Iterate on what works; double-down with creators producing winners

FAQ

How many creatives do I need to launch?

4-6 different angles minimum. Single-creative accounts are fragile to fatigue and disapproval.

Should I A/B test creatives within an ad set?

Use Facebook's Experiments tool for clean comparison. Multiple ads in one ad set lets the algorithm pick favorites and pollutes your read.

Bottom line

UGC video, 9-15 seconds, vertical, hook in first 2 seconds. Build a production pipeline producing 5+ variations/week. Stop spending on studio polish that converts worse than phone-shot authenticity.