Write Facebook Ad Copy That Actually Converts

The four-part anatomy of a Facebook ad

Every Facebook ad is built from four text components that behave differently. The primary text sits above the creative and does the heaviest lifting; the headline sits below the image or video in bold; the link description is a smaller line that Facebook frequently hides on mobile; and the CTA button is a preset label you choose from a short list.

Treat primary text as the hook and the case, treat the headline as a second, shorter hook that reinforces the first, and let the CTA button do the asking so your copy doesn't have to repeat it.

• Primary text The text above the creative. This is where your hook lives, and it's the only field long enough to actually make an argument before the reader has to tap to expand it.
• Headline A short bold line under the image or video. Treat it as a second chance at the hook — restate the benefit in different words rather than repeating the primary text verbatim.
• Link description A smaller line below the headline that many placements hide entirely, especially on mobile feed. Never put anything essential here; use it for a secondary detail at most.
• CTA button A preset label — Shop Now, Learn More, Sign Up — chosen from a fixed list, not custom text. Pick the one that matches what actually happens on the landing page, not the most exciting-sounding option.

Character limits and where the fold cuts you off

Facebook doesn't publish a hard character limit for primary text, but the visible portion before the "See more" link collapses is short — roughly the first 125 characters. Everything after that point is invisible until someone taps to expand it, and most people won't.

Headlines and link descriptions get truncated outright rather than collapsed, so an idea that runs long simply gets cut off mid-word. Write the complete thought inside the visible budget for each field, then treat anything past that as a bonus for the reader who taps through.

• Primary text — about 125 characters visible This is your entire hook budget on feed placements. Front-load the specific benefit or the question, and put your proof points and details after the fold, not before it.
• Headline — about 40 characters Enough for a short benefit statement or a number, not a sentence. Draft it separately from the primary text instead of trimming a longer line down to fit.
• Link description — about 30 characters Often hidden on mobile News Feed entirely and only shown in some desktop and right-column placements. Treat anything you put here as optional reading.
• Stories and Reels cut earlier Full-screen vertical placements show even fewer visible characters of primary text than feed does, so a hook tuned for feed can get chopped mid-sentence there — check the preview panel per placement.

Matching copy to audience temperature

The same product needs different copy depending on how familiar the audience already is with you. A cold audience has never heard of your product, so the copy has to earn attention and explain the offer from zero. A warm or retargeting audience has already engaged, and repeating the introduction wastes the words you have.

Writing one ad and running it against every audience is the single most common reason performance looks fine on impressions but weak on conversions: the copy is asking a stranger and a returning visitor to take the same leap with the same amount of context.

If you can only write one variant, write it for the cold audience — that copy generalizes down to warmer audiences far better than warm-audience copy generalizes up to cold ones.

Policy-safe phrasing that still converts

Meta's ad policies reject copy that implies it knows something personal about the reader, even when the claim is true. "Struggling with your credit score?" addresses a personal attribute directly. The fix is rephrasing around the outcome instead of the person: "A simpler way to track your credit score" makes the same pitch without the accusation.

Unverifiable superlatives are the second most common rejection trigger — "guaranteed results" or a specific percentage you can't substantiate will get flagged. Swap absolute claims for honestly hedged language: "designed to help" instead of "guarantees."

Before-and-after formats, fake countdown urgency, and unverifiable testimonials sit in the same risk category. None are worth the delay: a rejected ad restarts the review clock and can flag the account for extra scrutiny on future submissions.

Five hook techniques that stop the scroll

The hook is the first sentence of your primary text. It has one job — make the specific reader in front of you feel like this ad is unusually relevant — and it fails whenever it tries to also sell, explain, and persuade in the same breath.

Pick one per ad rather than stacking two or three, since combining hook types usually just produces a longer, vaguer opening line.

• Pattern interrupt Open with a statement that contradicts the expected framing of the category, such as naming the assumption most people make and then disagreeing with it in the same line.
• Direct question Ask the exact question the reader is already asking themselves, using their own likely phrasing rather than marketing language, so it reads like it was written for them specifically.
• Specific number A concrete figure — a timeframe, a count, a measurable detail — reads as more credible and more scannable than a vague adjective like 'fast' or 'easy.'
• Social proof lead Open with what other people are already doing or saying, which lowers perceived risk before you've made any claim about the product yourself.
• Problem-first Name the specific frustration in the reader's own words before introducing the product at all, so the ad reads as recognition rather than a pitch.

The pre-publish checklist

Most weak ads fail because a rushed final pass skipped a step that would have caught the problem. Run every ad through the same short list before it goes live, regardless of how confident the draft feels.

This isn't a style guide; it's a set of concrete, yes-or-no checks. If any answer is no, fix that one thing and re-check rather than rewriting the whole ad from scratch.

Formatting rules that make copy easy to skim

Facebook ad copy is read in a feed, thumb-scrolling, often with sound off. Dense paragraphs get skipped even when the writing itself is good, because the format itself signals effort required, and effort is exactly what a scrolling thumb is trying to avoid.

None of the rules below shorten copy for its own sake — they make the same information take less effort to absorb.

• Short lines, not short copy Break after a complete thought rather than writing one dense block — a line break gives the eye a resting point even when the total word count stays the same.
• One emoji, used as punctuation A single emoji can replace a bullet or draw the eye to the CTA line; more than one or two per ad starts to read as low-effort rather than friendly.
• Front-load every line Put the most important word of each sentence first, since the fold and the truncation both cut from the right, never the left.
• One CTA, stated once Decide on a single action and repeat it in the same words if you mention it twice — switching between 'Try it,' 'Get started,' and 'Sign up' in one ad makes the ask feel unclear.
• Write like you'd text it Contractions, sentence fragments, and a conversational rhythm consistently outperform formal marketing phrasing in feed placements, where the surrounding content is casual by default.

Testing and iterating without starting over

Ad fatigue is real and measurable — the same audience seeing the same creative repeatedly clicks less over time, regardless of how good the copy was on day one. The fix isn't writing an entirely new ad from scratch each time; it's changing one variable at a time so you actually learn which change moved the number.

Keep a simple log of what changed between versions. Without it, a win or a drop a week later is just a guess about which edit caused it, and you end up re-testing the same idea by accident.

• Swap the hook before the whole ad If a proven ad's performance is declining, rewrite only the first line first — the body and CTA are usually not the problem, and a full rewrite throws away a working structure.
• Test one variable per version Change the hook technique, the CTA label, or the audience temperature — not all three at once — so a lift or a drop can be attributed to something specific.
• Refresh on a cadence, not a panic Rotate primary text every one to two weeks for actively spent audiences rather than waiting for a visible drop, since fatigue shows up in cost trends before it shows up in obvious performance collapse.