The best ad examples are ones where you can see what the ad got, not just how it looked.
ROASArchive groups real ads from Meta, TikTok, Google and YouTube into 7 collections.
Every example links to a result page with the measured CPA, ROAS or cost per install and the name of whoever got it.
Most "examples of ads" galleries show a screenshot and stop there.
Here, every card goes back to a logged result: the change that was made, the number before and after, and the source.
How we picked: each collection pulls from our database of 698 results (582 of them pass our indexing bar), taken from public posts by media buyers, agency case studies and platform success stories.
We label vendor-reported stories. We did not run these ads ourselves.
| # | Collection | What's in it | Best measured example | Type |
|---|---|---|---|---|
| 1 | Facebook ads | Meta ads across ecommerce, apps, lead gen, info products | −45% cost per purchase after a €10 price cut (@SamPiliero) | Platform |
| 2 | Google ads | Search, Shopping, PMax, Demand Gen | ROAS 6.96 account-wide, Search at 9.37 (@runtraffic) | Platform |
| 3 | TikTok ads | Spark, in-feed, TikTok Shop, Smart+ | −57% CPI with creator Spark Ads (Admiral Media, Petit BamBou) | Platform |
| 4 | Instagram ads | Reels, Stories and Feed placements | −30% VIP CPA from Instagram bid multipliers (Smartly, TechStyle, vendor-reported) | Platform |
| 5 | YouTube ads | Direct-response video only | ROAS 0.76 → 1.51 while spend rose $682k (@runtraffic) | Platform |
| 6 | UGC ads | Creator content on Meta, TikTok, YouTube | CPA −50%, ROAS 3.2 from a "3 reasons why" ad (Hustler Marketing, TAO Clean) | Format |
| 7 | Ad copy | Headlines, primary text, hook lines | Cost per qualified lead −62% with a sharper angle (Flighted, Ambrook) | Element |
What this section is
This is the hub for every ad example page on ROASArchive.
Each collection is a list of ads or campaign changes, and each entry follows the same pattern: what ran, the measured result, who ran it, and one thing you can copy.
The difference from a swipe file is the number.
A public ad library shows you what is running.
The Meta Ad Library, for example, lets anyone search active ads, but spend ranges are only shown for ads about social issues, elections or politics. You see the creative, not the result.
Our examples start from the result. If a buyer posted "CPA $101.29 → $60.50" after launching 20 new angles, that number is what goes on the card.
The archive is built from wins people chose to share.
Treat each example as a hypothesis to test in your own account, not a guarantee.
Examples by platform
Five collections split ads by where they ran.
Start with the platform you spend the most on.
1. Facebook ad examples
What's inside: our Facebook ad examples cover 20-25 Meta ads and account changes across ecommerce, apps, lead gen and info products, mixing UGC, static, carousel and founder-led formats.
Standout result: @SamPiliero cut cost per purchase by 45% by lowering the product price by €10, with the same ads and budget.
Another account of theirs saw CPA go from $101.29 to $60.50 (−40.27%) after 20 new angles went into a tired account.
Why it's worth a look: several of the biggest Meta wins in the archive came from the offer and the angle, not from targeting tweaks.
Start here if: Meta is your main channel and your CPA has crept up.
2. Google ad examples
What's inside: the Google ad examples page has 15-20 entries split by campaign type: Search copy, Shopping feeds, Performance Max and Demand Gen.
Standout result: @runtraffic shared an ecommerce account at ROAS 6.96 overall, with Search at 9.37 and PMax at 7.65. @thedanielbudai grew revenue from $4.5k to $505k a month at a 2-2.5x ROAS by raising budgets in 10-15% steps.
Why it's worth a look: the breakdown by campaign type shows where the return actually comes from inside one account.
Start here if: you run Shopping or PMax and want to know which campaign types deserve more budget.
3. TikTok ad examples
What's inside: the TikTok ad examples collection has 15-25 entries covering Spark Ads, in-feed, TikTok Shop and Smart+.
Stories from TikTok's own inspiration pages are labeled vendor-reported.
Standout result: Admiral Media ran creator-made Spark Ads using real in-app footage for the meditation app Petit BamBou and cut CPI by 57% while conversion rate rose 170% in France.
In a separate test, scandiweb found a 10-minute flip-through of Baker Creek's seed catalog got 233 purchases at 1.32 ROAS, against 1 purchase at 0.29 ROAS for an agency-made video.
Why it's worth a look: on TikTok, the native, low-polish creative keeps beating the produced version.
Start here if: you are an app or an ecommerce brand testing TikTok next to Meta.
4. Instagram ad examples
What's inside: our Instagram ad examples are Meta results where the placement was Instagram, grouped by Reels, Stories and Feed.
Standout result: Smartly says TechStyle used bid multipliers to raise bids on Instagram placements without splitting ad sets, and VIP conversion CPA fell 30% (vendor-reported).
Insense reports that Cleanly's whitelisted influencer ads, including Stories Boomerangs, got 2.1X the CTR and a 2.3X lower CPM than its earlier creative (vendor-reported).
Why it's worth a look: placement is a lever on its own, separate from the creative.
Start here if: your audience lives on Reels and Stories, such as beauty and apparel brands.
5. YouTube ad examples
What's inside: the YouTube ad examples page has 10-15 direct-response ads, each with the hook, length, offer and result.
Standout result: @runtraffic scaled a YouTube account by $682k in spend while ROAS went from 0.76 to 1.51.
Power Digital rebuilt a makeup brand's YouTube for reach and frequency and measured YouTube prospecting at a 3.65 incremental ROAS (91% confidence) in matched-market tests.
Why it's worth a look: YouTube often looks weak in platform attribution.
These examples show what happened when people measured it properly.
Start here if: you already have video that works on Meta or TikTok and want a second channel for it.
Examples by format and element
Two collections cut across platforms: one by format, one by the words in the ad.
6. UGC ad examples
What's inside: our UGC ad examples cover 15-25 creator-made ads across Meta, TikTok and YouTube, each naming the creator or agency.
Case studies from UGC platforms are labeled vendor-reported.
Standout result: Hustler Marketing moved TAO Clean from mostly influencer content to UGC ads, and a "3 reasons why" winner led to a 50% lower CPA and a 3.2 ROAS, with a 49.4% hook rate. inBeat reports 73% lower customer acquisition cost for Genomelink after replacing stale image and carousel ads with an always-on UGC video pipeline.
Why it's worth a look: the winners share a format you can brief: one clear reason to buy per few seconds, said by a real person.
Start here if: you are about to hire creators and want to know which scripts to ask for.
7. Ad copy examples
What's inside: the ad copy examples page collects primary text, headline and hook-line changes with a measured result.
We quote only the copy the source showed.
Standout result: Flighted swapped soft lifestyle creative for product-forward ads with faster hooks and a "tax prep" angle for the farm finance software Ambrook.
Cost per qualified lead fell 62% while Meta spend grew 181% month over month.
Admiral Media's weekly tests mapping ChatPDF's selling points to hooks came with a 280% year-over-year rise in Meta ROAS and 45% lower CAC.
Why it's worth a look: copy and angle changes are cheap to test and show up again and again in the archive's biggest swings.
Start here if: your visuals are fine but the message feels vague.
If you want a live swipe file next to these measured examples, Foreplay and Atria both save ads from public libraries.
As of November 2026, Foreplay's Basic plan is $59 a month (or $49 a month billed yearly) with a 7-day free trial.
Atria's Core plan is $219 a month, or $129 a month billed annually.
Neither shows you another brand's results, so they work best alongside a results database like this one, not instead of it.
We may earn a commission if you sign up through our links.
How every example is chosen (real results only)
Every example in this section comes from every result in the database, and it has to clear the same bar:
- A measured number. CPA, ROAS, CPI, CPL, CTR or revenue, with a before and after or a split test. A nice-looking ad with no number doesn't qualify.
- A named source. The buyer's handle, the agency or the platform, linked on the result page.
- Numbers copied exactly. We don't round, convert or average. If the source gives two conflicting figures, the result page says so.
- Vendor stories labeled. When a platform or tool reports a win for its own feature, we mark it vendor-reported.
- Enough volume. Results based on a handful of conversions stay out of the indexed lists.
One honest limit: people post wins far more often than losses, so the archive leans positive.
Read every example as "this worked once, under these conditions."
Patterns across the winners
The offer and angle moved the most
A €10 price cut, 20 new angles and a "tax prep" angle produced cost drops of 40.27% to 62% without new targeting.
Check your offer before you rebuild your campaigns.
Show the product fast
Ambrook's winning ads revealed the product in the first 3-5 seconds.
TAO Clean's winner listed three concrete reasons to buy.
Cheap and native often beats polished
Baker Creek's 10-minute catalog video beat the agency video by 233 purchases to 1. Petit BamBou's best TikTok was a 10-second creator clip.
Measure beyond the platform dashboard
@runtraffic's brand search test showed an incremental ROAS of 8.17 but a new-customer ROAS of 1.17. Same campaign, very different story depending on what you count.
Real results from the archive
What to look for in an ad example
- A metric tied to money: CPA or ROAS tells you more than likes or views.
- A baseline: "CPA $47" means little without the number it replaced.
- The change that was made: one clear variable is easier to copy than a full rebuild.
- Who reported it: a buyer's own account, an agency or the platform selling the feature.
- Scale: spend and conversion counts tell you whether the result is noise.
- A business like yours: an app's CPI win may not carry over to a $200 ecommerce product.
If you'd rather have someone who has run tests like these work on your account, we can match you with a vetted media buyer or agency.