Guide
Offer-led ads are the hardest kind to learn anything from. The deal is loud enough to carry mediocre creative to a decent click-through rate — and then you scale it, and it stops working, and nobody can tell you why.
"50% OFF" on a coloured card in the first frame feels like a hook because it's a strong claim. It isn't. A hook is something that interrupts a scroll — it works by being unexpected, or by opening a loop the viewer needs closed. A discount is neither. It's the single most common thing a person sees in a feed, which makes it, mechanically, wallpaper. The viewer has been trained by ten thousand identical claims to classify it and keep moving before they've read the number.
A discount is a reason to buy. Reasons to buy are enormously valuable — to someone who is already paying attention. The order matters: earn the attention, then present the offer. Lead with the offer and you're asking a stranger to care about your pricing before they've been given any reason to care about your product. The mechanics of what actually does the interrupting are covered in how to write a TikTok hook — almost none of them are a price.
Here's the specific way offer-led ads lie to you. A strong deal lifts click-through even when the creative is dull, because the subset of people who were already in-market see the number and act. Your dashboard shows a healthy CTR. You conclude the creative works. You make five more ads in that style, spend into colder audiences where nobody is already in-market, and the whole thing collapses — because the creative was never doing the work, and outside the warm pocket there's no in-market subset to bail you out.
The inverse failure is just as common and looks like the problem described in high CTR but low conversions: a giveaway pulls enormous engagement and zero revenue, because entering costs nothing and selects for people who want free things rather than people who want your thing. Both cases share a root cause. You ran one test with two variables in it — offer and creative — and the numbers can't tell you which one moved.
There are two independent things to learn, and they need different tests. Does the offer move people? Hold the creative constant and vary the deal — same ad, 20% off versus free shipping versus bundle. That's a live test; it needs real money and real buyers, because it's a question about willingness to pay and no amount of pre-testing predicts it. Does the creative earn attention? Hold the offer constant and vary the ad — and this one you can screen before spending, because it's a property of the file.
The mistake is doing both at once: new offer, new creative, one ad set, one result. You'll get a number and no knowledge. Change one thing per test, which is the discipline behind every workable creative testing framework for small budgets.
If the offer isn't the opening, where does it go? Not at the very end, taped on as a card after the ad has already finished emotionally — by then a large share of viewers are gone, and the ones remaining are watching a logo. The offer wants to arrive while attention is still alive, close to but not at the finish, which is the argument in CTA placement in video ads. A discount announced to an empty room is not a discount.
This matters most in the seasonal crush, when everyone's running the same percentage at the same time and the deal has zero differentiating power on its own — see seasonal ad creative planning. When every competitor is shouting a number, the number is not the differentiator. The ad around it is.
PreTestAds predicts how a piece of creative holds attention across its runtime and ranks it against a benchmark of top-performing short-form ads. On an offer-led ad that gives you something the live numbers actively obscure: a read on the creative independent of the deal attached to it. If the discount-led cut and the story-led cut score far apart, you've learned which one would survive without the offer propping it up — before you spend into a cold audience to find out.
And the honest limit, because it's a big one here: nothing about this scores the offer. It cannot tell you whether 30% off beats free shipping, whether your margins survive the promotion, whether a giveaway attracts buyers or freebie hunters, or whether the deal closes anyone. Those are economic and audience questions that only live testing answers. What the screen does is stop you from concluding your creative is fine because a discount was carrying it — and make sure the ad is worth attaching a good offer to in the first place.
Upload the cut and see whether it holds attention on its own — first analysis free.
Score Your AdA discount is a reason to buy, not a reason to stop scrolling — and those are different jobs. Percentage-off claims are the most common opening in paid social, which makes them the least surprising thing a viewer can encounter. The offer usually works better as the payoff after the creative has earned a few seconds of attention than as the first thing on screen.
Because a giveaway selects for people who want free things, not people who want your product. Entry costs nothing and requires no intent, so engagement inflates while purchase intent stays flat. That's a targeting and offer-design issue rather than a creative one — the ad may be doing its job perfectly and still attract the wrong crowd.
Hold one constant and vary the other. Run the same offer behind two genuinely different creative treatments, or the same creative with two different offers — but not both at once, which is the most common mistake. Screening the creative for predicted attention before launch also helps, because it reads the ad independently of the deal attached to it.