Register
Login
Back to Blog
Abstract layered ad copy cards connected by testing arrows and conversion signal dots

Ad Copywriting for Performance Ads: Formulas, Examples and Tests (2026)

Ethan Cole
Ethan ColePublished on October 13, 2026 in Tech Guides

Ad copywriting for performance ads means writing the primary text, headline and call to action so that a paid impression turns into a measurable action, not just attention. The working method is simple: pick one promise per ad, lead with the outcome in the first line, match the copy to the landing page, and test variants on cost per result rather than on taste.

What makes performance ad copy different from brand copy

Brand copy is judged on recall and sentiment. Performance ad copy is judged on a single number the campaign is optimising for: cost per purchase, cost per lead, cost per install. That changes how you write in three ways.

First, the copy has a job in the auction. On Meta and TikTok the delivery system predicts how likely a person is to take the optimised action, and that prediction is partly learned from how people react to your text and creative. Copy that attracts clicks from the wrong people trains the system toward the wrong audience.

Second, the copy has to survive the scroll. On mobile feeds, Meta's own ad specs recommend keeping primary text to about 125 characters so it shows before the "See more" fold, and headlines to roughly 27 characters so they do not truncate. Anything important that sits below the fold is effectively invisible to most viewers.

Third, the copy has to agree with the landing page. If the ad promises one thing and the page delivers another, the click is paid for and the conversion is lost. This is also where most policy problems start, which is why it pays to read how Meta evaluates ads before you write, not after a rejection. The guide to Meta ad policy approval and disapproval covers the review logic in detail.

Five ad copywriting formulas that work for performance ads

You do not need dozens of frameworks. These five cover almost every performance use case, and they are easy to brief to a writer or to a model.

Problem, agitate, solve (PAS)

Name the problem in the first line, make it concrete in the second, and present the product as the resolution in the third. PAS works well for products that fix something annoying: tools, utilities, services with a clear before and after.

Example (Meta, lead generation for a bookkeeping service): "Still reconciling invoices by hand on Sunday night? Every missed receipt is money you cannot claim back. Book a 15-minute setup call and let us close your books this month."

Before, after, bridge (BAB)

Describe the current state, describe the desired state, then position the product as the bridge. BAB is gentler than PAS and suits categories where people do not feel pain so much as aspiration, such as learning apps, fitness programmes and home products.

Example (TikTok, app install): "Six months ago I could not hold a conversation in Spanish. Now I order dinner without the menu. Ten minutes a day in the app got me there."

Feature, advantage, benefit (FAB)

State what the product has, what that lets it do, and what that means for the buyer. FAB is the safest formula for e-commerce product ads where the differentiator is a real attribute: material, battery life, compatibility, delivery speed.

Example (Meta, catalogue sales): "Titanium frame. Half the weight of steel. You stop noticing you are wearing glasses."

Proof first

Open with the evidence, then explain the offer. Proof can be a customer quote, a review count, an award, a certification or a specific result. Keep the proof literal and sourced; vague superlatives read as filler and can trigger policy review.

Example (Meta, SaaS trial): "Rated 4.8 on G2 by 2,000+ marketing teams. Start a free 14-day trial, no card required."

Only use numbers you can show on the landing page. Review counts, ratings and prices must match what the visitor sees after the click.

Objection flip

Lead with the reason people do not buy, then answer it. This is the most under-used formula and often the best performer in retargeting, where the audience already knows the product and is stalled on one doubt.

Example (Meta, retargeting cart abandoners): "Worried it will not fit? Free exchanges for 60 days, prepaid label included."

Facebook ad copywriting tips that hold up in 2026

The formulas give you structure. These tips cover the details that decide whether the structure actually performs on Meta placements.

  • Put the outcome in the first seven words. The first line is the only line most people read. "Cut your ad review time in half" beats "We are excited to announce our new feature".
  • One promise per ad. If you have three selling points, make three ads. Multiple promises in one primary text dilute the message and make results impossible to attribute.
  • Write the headline as the answer to "what do I get?". The headline sits under the creative, next to the button, and it is often the only text visible on Reels and Stories placements.
  • Match the call to action button to the verb in the copy. "Shop now" with a product ad, "Learn more" with a content ad, "Get quote" with a lead form. Mismatched buttons lower the click-through rate on the one element people actually tap.
  • Keep claims specific and verifiable. "Ships in 2 days from our Ohio warehouse" is specific. "Fast shipping" is not. Specific claims also pass review more easily; if an ad does get rejected, the fixes in how to fix a rejected Facebook ad usually come down to the copy making a claim the page does not support.
  • Avoid addressing personal attributes. Meta's advertising standards prohibit copy that asserts or implies a person's characteristics (health condition, financial status, religion and similar). "Tired of knee pain?" is a common way to get an otherwise fine ad disapproved. "Knee support designed for daily walks" is the compliant rewrite.
  • Use the description field for logistics, not for persuasion. Return policy, delivery window, price. It only shows on some placements, so nothing essential should live there.

Performance ad copy on TikTok versus Meta

The same formula needs a different voice on each platform.

On TikTok, the text field is short (TikTok Ads Manager caps ad text at 100 characters) and the copy is usually spoken in the video rather than read in the caption. The caption's job is to reinforce the hook and carry the call to action. Write it like a comment, not like a billboard: first person, present tense, no corporate framing. The first three seconds of the video do most of the work, and the written copy should echo that hook rather than introduce a new idea. The article on ad hooks in the first 3 seconds covers how to build that opening.

On Meta, the primary text can carry the full argument, and feed placements still reward a readable two or three line structure. The headline is more important than on TikTok because it renders on more placements. Reels and Stories on Meta behave more like TikTok, so for a campaign running across Advantage+ placements it is worth writing a short variant of the primary text and letting the system pick.

A practical rule: write the Meta version first using one of the five formulas, then cut it to the first line plus the call to action for TikTok.

Ad copywriting examples by campaign objective

Use these as templates. Replace the specifics; keep the structure.

Purchase, cold audience (FAB): Primary text: "Merino base layer, 150 gsm. Regulates heat on the climb and dries on the descent. Free returns." Headline: "Merino base layer, 4 colours" Button: Shop now

Lead, cold audience (PAS): Primary text: "Your website gets traffic but the phone does not ring. Most of those visitors leave without seeing a reason to call. Get a free 10-point audit in 48 hours." Headline: "Free website audit" Button: Get quote

App install (BAB): Primary text: "From 'I will budget next month' to seeing every subscription in one list. Takes two minutes to connect." Headline: "See every subscription" Button: Install now

Retargeting, viewed product (objection flip): Primary text: "Not sure about the size? Try it at home. Free exchange within 60 days." Headline: "Free size exchange" Button: Shop now

Retargeting, lapsed customer (proof first): Primary text: "Your last order was 90 days ago. The reformulated version has a 4.9 rating from 3,100 reviews. Reorder in one tap." Headline: "Reorder in one tap" Button: Shop now

How to test performance ad copy without fooling yourself

Copy testing fails for the same reasons creative testing fails: too many variables, too little spend per variant, and judging on click-through rate instead of the objective. The method below is the same discipline described in how to test ad creatives on Meta, applied to text.

  1. Change one layer at a time. Test headlines against the same primary text, or primary texts against the same headline. Do not test a new headline and a new image in the same variant.
  2. Keep the creative constant. If the image changes, you are testing the image.
  3. Give each variant enough results to read. A variant with four conversions cannot be ranked against a variant with six. Set a minimum number of optimised events per variant before you compare.
  4. Rank on the campaign objective. Cost per result first, then conversion rate from click, then click-through rate. A copy that wins on click-through rate and loses on cost per purchase is attracting the wrong clicks.
  5. Watch the comments. On both Meta and TikTok, confused or hostile comments on a copy variant are an early warning that the promise and the product do not line up.
  6. Promote, then iterate on the winner. Once a formula wins, write the next round as variations of that formula (different proof, different objection), not as five unrelated ideas.

For account-level hygiene around testing, the structure in Facebook ads account structure explains where copy tests should live so that they do not compete with scaling ad sets.

FAQ

What is the best ad copywriting formula for performance ads?

There is no single best formula. PAS and FAB are the most reliable for cold audiences, and the objection flip is usually strongest in retargeting. Pick the formula that matches where the audience is in the funnel, then test proof and promise variations within it.

How long should Facebook ad copy be?

Lead with the message in the first 125 characters, which is the length Meta's specs recommend to avoid truncation in feed. Longer primary text can work for considered purchases, but the first line still has to carry the promise on its own.

Should ad copy be different for TikTok and Meta?

Yes. TikTok ad text is limited to 100 characters and the message is carried by the video, so the caption echoes the spoken hook and the call to action. Meta primary text can hold the full argument, and the headline matters more because it renders on more placements.

How do I know if a copy variant is actually better?

Compare on cost per optimised result with enough conversions per variant to be meaningful, and keep everything except the copy constant. Click-through rate alone is not a reliable signal for performance ads.

Ready to Boost Your Ad Conversions?

See how DeepClick can improve your post-click performance.

© 2009, DeepClick Limited.
Email: [email protected]
Room 1508, Grand Plaza Office-Tower 2, 625 Nathan Rd, Mong Kok, Kowloon City, Hong Kong
Reflow Features
icon
Ad Fallback PageExclusive PageAudience RecoveryClaim ResolutionGreen ShieldPush NotificationsPWA Retargeting
Industry Solutions
icon
AI Social AppsGamingMeta & TikTok Advertisers
About Us
icon
Contact Sales
Join Us
Partners
Resource Center
icon
BlogAll Articles
API Document
Privacy PolicyUser Agreement