Ad Copy Formulas That Work: PAS, AIDA & Native Patterns

The ad copy formulas that hold up on native placements are Problem-Agitate-Solution (PAS), a compressed version of AIDA, and a small set of native-specific patterns built on curiosity and third-person framing: the "she discovered" story hook, the numbered-benefit listicle, and the before/after reveal. None of these are secret. What separates a formula that converts from one that gets a creative rejected is discipline: matching the headline's promise to what the landing page actually delivers, and staying inside each network's content policy while you do it.

Why generic copy formulas need adapting for native#

Native ad units sit inside editorial content, next to "you might also like" widgets on a news site or a content-recommendation feed. That placement changes what copy has to do. A search ad competes on intent, a social ad competes on scroll-stopping visuals, but a native unit competes with the article the reader is already reading. The headline and thumbnail carry almost the entire persuasive load; there's rarely body copy at the ad-unit level, so the formula has to be compressed into 40 to 90 characters and one image.

That constraint is why straight textbook AIDA (Attention, Interest, Desire, Action) rarely survives contact with a native placement unchanged. You don't have room for four distinct beats in a headline. What works is folding Attention and Interest into the headline and thumbnail, then letting Desire and Action happen on the landing page or pre-lander.

PAS, compressed into a headline#

Problem-Agitate-Solution still works because it mirrors how people actually read a feed: they skim for a problem they recognize, feel a flicker of concern, and click to see the fix. On native, the "Agitate" step often has to happen in the same clause as "Problem" because there's no space for three separate lines.

A workable native PAS headline looks like: "[Problem, stated plainly]. [Agitation clause that raises the stakes]. [Implied solution, not stated outright]." The solution rarely appears in the headline itself, since spelling it out kills the reason to click. It shows up on the landing page instead.

AIDA, compressed for a 60-character headline#

Because you can't run all four AIDA stages in the headline, treat the headline as an Attention-plus-Interest hybrid and the landing page as the Desire-plus-Action half. The headline's only job is to get the click; asking it to also build desire and close the action is what produces bloated, unreadable headlines that networks throttle for low CTR anyway.

This split also matches how hook, angle and claim function as separate layers of a native ad: the hook lives in the headline, the angle carries through the image and opening paragraph of the landing page, and the claim is whatever specific promise the offer is actually making.

Native-specific patterns pulled from live creatives#

Beyond PAS and AIDA, a handful of patterns show up constantly across the OpenAdLibrary index of 725,000+ native creatives across 49 networks (July 2026), and they're worth naming because they're genuinely different from display or social copy conventions.

• The third-person discovery hook. "Wrinkles: Most People Use Lotions. Koreans Do This Instead (It's Genius)" is a live Taboola headline pattern in the beauty vertical: state the common approach, then imply a better one exists without naming it.
• The authority-figure reveal. "Surgeons: This Simple Trick Will End Knee Pain & Arthritis Quickly (Try It)" borrows credibility from a profession rather than a named individual, which sidesteps most networks' rules against implying a celebrity or doctor endorsement that doesn't exist.
• The numbered list. "Top 5 Shampoos To Avoid" works because a number sets a concrete expectation and reads as editorial rather than promotional.
• The dollar-figure curiosity hook. "Inside television star's $9.1 million payday" pairs a specific number with a vague subject, which is one of the most consistent ad hook patterns in the finance and real estate verticals.

None of these are formulas you invent from scratch. They're patterns that survive because they get approved repeatedly and keep converting, and you can see live variants of each one by browsing the most common native ad angles or, if you sell physical product, the ad angles library for ecommerce.

A pattern worth calling out separately is the household-object reframe: "A Window Cleaner Explained Why Sprays Make Your Glass Worse" takes a mundane category and manufactures a contrarian claim about the conventional approach, which works because it doesn't require the reader to have a pre-existing problem, just curiosity about a category they already interact with. This differs from PAS in a subtle but important way: there's no agitated pain point, only an implied "you're doing it wrong" premise. It tends to perform well in home, garden and cleaning verticals precisely because those categories don't carry the same emotional weight as health or finance, so a straight problem-agitate approach can feel forced where a contrarian reframe doesn't.

Matching image and headline so the formula actually lands#

A formula only works if the thumbnail reinforces the headline's promise rather than fighting it. If your headline implies a personal discovery ("She found this in her garden") but the image is a stock product shot on a white background, the mismatch reads as an ad immediately, and readers who've learned to filter native placements out of their attention will do exactly that. The strongest-performing creatives in the OpenAdLibrary index consistently pair a candid, unstaged-looking photo with a headline that references something specific in that photo: a garden, a bathroom fixture, a hand holding an object. Generic stock imagery undercuts even a well-written PAS or AIDA headline, because the whole point of the native format is to look like it belongs in the feed the reader is already scrolling.

This is also where testing multiple images against a single strong headline formula tends to produce more reliable lift than testing multiple headline formulas against a single image. The headline formula sets the psychological structure; the image is what determines whether the ad reads as native or as an obvious interruption.

Matching copy to compliance and network policy#

The gap between "curiosity-driven" and "misleading" is where most rejected creatives live. A headline that implies a doctor endorsed a product without one existing, or that promises a specific result ("cures arthritis in 3 days") crosses from curiosity into a claim the network and the FTC will both flag. The FTC's disclosure rules for advertorials apply regardless of which network is running the ad, and each network layers its own moderation on top: Taboola and Outbrain are generally stricter on health and finance claims than mid-tier networks, and what clears MGID moderation won't necessarily clear Taboola's.

Practical rule: if the headline states a fact, that fact has to be roughly true. If it implies a comparison ("Koreans do this instead"), that's framing, not a factual claim, and it survives review more consistently.

How to test formulas without wasting budget#

Don't guess which formula fits a new vertical from scratch. Pull up live creatives in the vertical you're targeting through a tool like OpenAdLibrary's native ad spy tool and look at which headline patterns are running longest, since longevity is a rough proxy for profitability on networks where advertisers can pause underperforming creatives freely. Then write 3 to 4 headline variants using the pattern that's already surviving in that vertical, rather than 3 to 4 variants of a pattern you like personally.

Run those variants against the same image and landing page, let the campaign clear its learning phase, and cut anything that isn't pulling its weight on CTR after a real sample size, not after 200 impressions.

Common mistakes#

• Reusing one pattern across every vertical. A curiosity-gap hook that works in beauty often falls flat in B2B software, where readers respond better to a direct benefit statement.
• Overselling the headline relative to the landing page. If the click doesn't deliver on the implied promise within the first screen of the pre-lander, bounce rate climbs and the algorithm throttles delivery.
• Ignoring network-specific moderation. A pattern that's fine on Revcontent can get a creative rejected on Outbrain. Check each network's current policy rather than assuming uniformity.
• Skipping the creative analysis step. Scoring your own hook, angle and CTA against what's already winning tells you which formula element is actually weak before you burn budget finding out the hard way.
• Writing the headline before deciding the angle. A formula is a structure, not a substitute for having an actual angle. PAS with a weak angle underneath it still produces a weak ad; the formula only amplifies whatever premise you feed into it.
• Treating a formula as permanent. Every pattern above fatigues with repeated exposure inside the same feed. A headline structure that pulled a strong CTR for three weeks can flatten out once the same audience has seen enough variants of it, which is a separate problem from the formula being wrong, it's just worn out for that specific audience.

Formula selection by vertical#

Not every pattern performs evenly across verticals, and matching the formula to the category saves a round of wasted testing:

Use this as a starting point, not a rule. The fastest way to confirm which formula actually fits your specific offer is watching what's already running and surviving in your exact vertical rather than assuming the general pattern holds for your product.