Facebook Ads Copywriting That Converts (And the Copy Tropes I've Stopped Using)

The 4-part hook structure I rotate through, why "benefit-driven" copy is overrated, and the specific patterns that consistently outperform clever wordplay.

Most Facebook ad copy fails not because it's badly written but because it's aimed wrong. The copy doesn't need to be clever โ€” it needs to do one specific job: get the right person to stop scrolling and click.

Below: the patterns I've seen consistently win across 100+ tested ads, and the "copywriting wisdom" I've quietly stopped following.

The 4-part hook structure

Almost every winning ad in my account follows the same shape:

  1. Pattern interrupt โ€” first 5-7 words. Stop the scroll.
  2. Specificity โ€” name the person or situation precisely.
  3. Promise + evidence โ€” what they get and why to believe it.
  4. Action โ€” what to do now.

Example for B2B SaaS:

"Stop running standups twice. (1)
If you're managing 3+ remote engineers across timezones, (2) our async update tool gives your team 4 hours back per week. (3) Try it free for 14 days. (4)"

It's not poetry. But it does the work. Every component has a job.

A/B Test ยท 4 hooks, same offer, same audience HOOK CTR CPA "Boost productivity with AI-powered solutions" generic benefit copy 0.4% $78 "3 hours back per week, guaranteed" specific outcome, no specificity on who 0.9% $42 "Eng managers: stop running daily standups" specific audience callout + pattern interrupt 1.6% $24 "Stop running standups twice. If you're managing 3+ remote eng..." full 4-part hook structure 2.1% $19
Same offer, four hooks. The "boost productivity" generic version performed 5x worse than the specific 4-part hook. Pattern interrupt + audience callout did most of the work.

What I've stopped doing

Long benefit lists. "Save time, save money, increase revenue, automate workflows, scale your team." Reads like a feature spec. Doesn't convert.

Question hooks. "Are you tired of...?" "Did you know...?" The reader instantly answers "no" mentally and scrolls past.

Statistics as openers. "73% of marketers say..." Stats need context. By the time you set up the context, the scroll is gone.

Personal pronouns about the brand. "We've helped 10,000 customers..." Reader doesn't care. They want to hear about themselves first.

Emojis as bullets. โœ… โœ… โœ…. Reads like a corporate LinkedIn post. The exception: emojis ARE useful as a single accent in a Reels caption.

What consistently works

Specific audience callouts. "Engineering managers," "Shopify store owners doing $20-50k/mo," "Solo creators making courses." If the reader thinks "that's me," they read on.

Concrete numbers. "4 hours/week," "14-day trial," "$340/month." Specific beats vague every time.

Negative framing. "Stop running standups twice." "Quit copy-pasting between sheets." Negativity grabs attention more reliably than positive promises.

Permission to be skeptical. "If you've tried Notion and it didn't click, here's why this is different." Acknowledges the reader's objections instead of fighting them.

Insider language. Use the terms your audience uses. SaaS founders say "churn," not "customer departure rate." Match their vocabulary.

The structure of a full ad copy

For most static/video ads:

Line 1 (hook): Pattern interrupt + specificity.
Lines 2-3 (context): The pain or situation, named precisely.
Line 4 (promise): What you offer + concrete outcome.
Line 5 (proof): One credibility marker (number, customer count, time frame).
Line 6 (CTA): What to do now.

Total: 5-8 lines, 60-120 words. Anything longer doesn't get read on mobile.

Headline + Description

The headline is what shows under the image/video. Keep it under 40 characters when possible. Repeat the core promise from your copy in fewer words.

Description (the small text below headline) is mostly cosmetic. Most users don't read it. Don't spend much time optimizing it โ€” repeat the CTA or add a credibility marker.

Testing copy efficiently

Test in batches of 4-6 hooks against the same image/video. Same offer, same landing page, same audience. Let it run until you have 20-30 conversions per variant. Then keep the top 1-2 and discard the rest.

The test that taught me the most: 4 ads with identical creative and 4 different first lines. CTR ranged from 0.4% to 2.1% โ€” same image, same offer, just different hook. The first line is doing 80% of the work.

FAQ

How long should ad copy be?

Mobile reads about 90 chars before "...See more". Get the hook + key promise in those 90. Anything past that is bonus and many won't read it.

Do emojis help?

Sparingly. One at the start of a Reels caption can pop. Bullet lists of emojis read as corporate spam.

Should I use the brand name in copy?

For unknown brands: avoid until you've hooked them. For known brands (where the brand IS the value): yes, lead with it.

Do longer descriptions hurt?

Not really. Extra description is mostly invisible. But spending hours polishing it has worse ROI than testing 3 new hooks.

Bottom line

Lead with a specific audience callout. Use concrete numbers. Negative framing beats positive in most situations. The first 5-7 words decide everything; the rest is supporting work. Test hooks ruthlessly, accept that 70% of what you write will lose, keep the 30% that wins.