7 Creative Testing Ideas for Meta and Google Ads
Will Phillips
Marketing Manager
You’ve tightened the targeting. You’ve adjusted the bids. You’ve split the budget three ways and watched the dashboards like a hawk. And performance is still… flat.
More often than not, the creative is the thing holding you back, not the settings. The targeting you used to control by hand is now largely automated. What you put in front of people is the part you still own — and it’s the part that decides whether they stop scrolling or keep going.
So if you want better results, that’s where to look. Below are seven paid media creative testing ideas you can run on Meta and Google this quarter, plus a quick framework for testing them without fooling yourself.
This isn’t a hunch. A meta-analysis by Nielsen and NCSolutions across hundreds of campaigns found that creative quality drives around half of a brand’s sales lift from advertising — and for digital campaigns specifically, that figure rises to 56%, far ahead of media placement. Meta cites the same Nielsen research, noting creative drives 56% of a campaign’s ROI. (Meta for Business)
Meanwhile the things teams obsess over carry less weight than you’d think. One analysis suggests click-through rate, the metric everyone watches, influences only about 4% of ROI. (Motion)
The takeaway is simple. As targeting gets handed to the algorithm, your creative becomes the main thing you can influence. Testing it properly is no longer a nice-to-have — it’s the work.
A test only teaches you something if it’s set up to give a clean answer. Four rules before you start:
Now to the tests.
Most people decide whether to keep watching almost instantly. So the thing worth testing first in a video ad isn’t the edit, the music, or the call to action — it’s the opening.
How to run it: Take one video concept and produce three versions that are identical except for the first three seconds. Version A opens with the problem (“Still getting bloated after every meal?”). Version B opens with a bold visual or motion. Version C opens with social proof (“Over 100,000 five-star reviews”). Same everything else.
Watch the thumbstop rate (the percentage of people who stop scrolling and watch) and hold rate (how many keep watching past the opening). These tell you which hook earns attention before you spend money optimising anything downstream.
A problem-led ad leads with the pain the customer feels. A product-led ad leads with what you sell. They attract very different buyers — and which one wins tells you something useful about where your audience sits.
If a problem-led hook wins, your audience probably doesn’t yet realise they have a problem you solve; you need to educate before you sell. If product-led wins, they already know the category and you’re competing on the offer.
How to run it: Same product, same format, two messages. One opens with the struggle (“Your skincare shelf is a graveyard of half-used jars”). The other opens with the product and its benefit (“One serum. Five steps replaced.”). Run them as separate ads and compare cost per acquisition.
Polished brand films have their place. But on Meta’s feed, ads that look like content — shot on a phone, a person talking to camera, a screenshot of a genuine review — often outperform the glossy stuff, because they blend into the feed instead of announcing “this is an ad.”
This is the gap between a brand’s instinct (make it beautiful) and what the platform rewards (make it feel native). The only way to settle it is to test.
How to run it: Take a concept that’s working as a polished asset and remake it as raw, user-generated-style content (UGC) — informal, hand-held, unscripted. Run them head to head. Don’t assume the pretty one wins. Often it doesn’t.
There’s a widespread belief that video always beats static. It frequently does — but not always, and video costs far more to produce. Before you pour the budget into production, find out whether you actually need to.
How to run it: Take a single message and build it both ways: one well-designed static image and one short video. Compare not just the engagement but the cost per result. If the static delivers 90% of the outcome at 20% of the production cost, that changes how you allocate your whole creative budget.
On Google’s search ads, you’re not stopping a scroll — you’re answering a question someone typed on purpose. The place to focus is your responsive search ads (RSAs), the ad format where you supply multiple headlines and descriptions and Google’s machine learning tests combinations to find what performs best.
Google’s own guidance is to provide a strong mix of distinct assets and aim for “Good” or “Excellent” Ad Strength. According to Google Ads Help, advertisers who improve an RSA’s Ad Strength from “Poor” to “Excellent” see, on average, 12% more clicks and conversions. (Google Ads Help)
How to run it: Most accounts fill their headlines with features and brand names. Instead, write a set of headlines that each answer a specific objection — “Free returns within 30 days,” “No subscription required,” “Ships from the UK.” Provide at least 8–10 headlines, with several unique ones that don’t repeat each other, as Google recommends. Then let Google test which objection-busters pull best.
“Trusted by thousands” is social proof. So is a screenshot of one detailed five-star review. So is a specific number. They’re not interchangeable — and you won’t know which lands hardest with your audience until you put them against each other.
How to run it: Build three variants of the same ad, each using a different proof format:
Same creative otherwise. The winner often surprises people: specific, human reviews tend to beat round-number claims, but it’s audience-dependent, which is exactly why you test.
You don’t always need a bigger discount. Often the same offer, framed differently, performs better — and protects your margin while it does.
How to run it: Hold the actual deal constant and change only how you say it. “20% off” versus “Save £15” versus “Buy 2, get 1 free” can be the identical economic offer dressed three ways. Or test urgency framing (“Ends Sunday”) against value framing (“Free shipping, always”). You’re testing the words around the offer, not the offer itself — which means a win costs you nothing extra to roll out.
Three traps catch even experienced teams:
When a test wins, feed the learning into the next round rather than treating it as the final answer.
Creative is the biggest difference-maker you’ve got, so test it deliberately: open strong in the first three seconds, weigh problem-led against product-led hooks, give the “ugly” ad a fair shot, check whether you really need video, rewrite Google headlines around objections, test your social proof format, and reframe the offer before you discount it. Change one thing at a time, judge on business outcomes, and let each result inform the next.
If you’d rather not build and run all this in-house — or you want a second pair of eyes on a testing roadmap that connects your paid media to the rest of your funnel — speak to the team at Brave. We help eCommerce brands turn SEO, paid media, CRO and email into one connected journey, and creative testing is a big part of how we do it.
Now go break something in your ad account. That’s usually where the wins are hiding.
