Copywriting for facebook ads

Lead with the hook, never the brand

The first 3-5 words of the primary text decide whether the user stops scrolling. **Brand-name-first openers consistently lose to hook-first openers** by 20-40% on CTR. 'Acme Corp helps marketing teams...' loses to 'Marketing leaders are firing one tool a quarter...' The hook earns the scroll-stop; the brand is identified later.

# Bad (brand-first):
# "Acme Corp helps Shopify brands scale their paid social..."
# Good (hook-first):
# "Most Shopify brands waste 30% of their Meta budget on..."
# Better (hook-first + specific):
# "I audited 50 Shopify ad accounts. The same 3 mistakes cost them $1.2M last year."

Match copy length to the click-through promise

Short copy (25-90 chars) wins when the visual carries the message and the click is the next obvious step. Long copy (300-1000+ chars) wins when the offer needs context (B2B SaaS, complex services, higher AOV). **Don't default to short or long - match the length to the friction of the conversion**. A $40 product needs less copy than a $4000 service.

Open the headline with a specific number or claim

Numbers and specific claims outperform vague benefit statements. **'$3,200 saved on Meta spend per month'** beats 'Lower your ad costs.' **'47 new customers in 30 days'** beats 'Grow your business.' Specificity signals real proof; vague claims signal generic stock copy.

Write in second person ('you'), not third person

Facebook ads sell to the individual scrolling, not a market segment. **'You're spending $5k/mo on Meta and not sure if it's working'** beats 'Marketing teams struggle to measure Meta ROI.' Second person creates immediacy; third person creates distance. Distance is the enemy of CTR.

Place the CTA in the last 1-2 sentences only

Don't bury 'Click here to learn more' in sentence 2 of a 5-sentence ad. The CTA goes in the final sentences - after the hook has earned the read and the body has built the case. Premature CTAs feel pushy; final-sentence CTAs feel earned.

Use line breaks to make copy scannable

Walls of text die on Meta. **Two-sentence paragraphs maximum.** Line breaks between paragraphs. The first paragraph is the hook (1-2 sentences). Subsequent paragraphs build the case in scannable chunks. The mobile preview is your friend - if it looks like a wall on mobile, it dies on mobile.

# Good structure:
#
# [Hook - 1-2 sentences]
#
# [Proof or context - 1-2 sentences]
#
# [Specific outcome / number - 1 sentence]
#
# [CTA - 1 sentence]

Tailor copy to the audience temperature

**Cold (prospecting)**: hook-heavy, no jargon, no insider language. **Warm (retargeting site visitors)**: assume product awareness, lead with offer. **Hot (cart abandoners, customers)**: assume buying intent, lead with urgency or incentive. Running the same copy across all three temperatures wastes the warm/hot audiences' specific advantages.

Test copy in isolated experiments, not bundled

Most teams change copy + visual + audience + bid at the same time and have no idea which change moved the metric. **One variable per test.** Same audience, same visual, two copy variants. Run for 7 days. Re-measure. The variant-isolation discipline is what makes copy learnings compound.

Pull copy back into your swipe file when it works

When a copy variant outperforms, save it with full context: ad-set audience, visual paired with, outcome metric, the hook archetype it represents. **Your swipe file is the compounding asset** - month 6 of disciplined collection beats month 1 by 10x because you can pattern-match new tests against the corpus.