Ecommerce Advertising Examples, By Format — With Real Numbers

The four formats that actually count as ecommerce advertising

Ecommerce advertising isn't one thing with one look. It's four distinct mechanisms, each triggered differently and each measured differently — most "examples" pages lump them into one long scroll of creative screenshots.

• Search / sponsored ads — triggered by a keyword or ASIN, shown at the moment someone is already looking (Sponsored Products, Google Shopping).
• Social creative ads — image, video, carousel, or UGC shown in-feed, triggered by audience and interest rather than intent (Meta, TikTok).
• Retail media brand placements — Sponsored Brand and Sponsored Display, competing for a shelf position on retailer-owned real estate.
• Programmatic display and video (DSP) — bought on an exchange, targeted by audience or retargeting list, shown on- or off-platform rather than against a keyword.

The pages currently ranking for this query are all built from the second category — creative swipe files, hundreds of screenshots of carousel ads, UGC, and unboxing videos. Useful for a designer looking for a layout. None of them touch retail media or DSP, which is where a growing share of ecommerce spend sits for brands that sell on Amazon, and none of them show what happens after the click.

What each format looks like in practice

Here's what a real example looks like for each format, described by mechanism rather than screenshot:

• Sponsored Product example: a cookware brand like HexClad bids on a competitor's ASIN so its ad shows on that product's detail page. Proof it worked isn't the click — it's whether the buyer was new to the brand or already shopping it.
• Social creative example: a beauty brand like BK Beauty runs a carousel built from customer UGC. Proof it worked is scroll-stop rate and cost per landing-page view, not impressions.
• Sponsored Brand example: a grooming brand like Beardbrand runs a headline search ad across three of its own SKUs. Proof it worked is whether the click bought the bundle, not just the searched item.
• DSP retargeting example: a wallet brand like Ridge serves display ads to shoppers who viewed a product page but didn't buy in the last 14 days. Proof it worked is the purchase rate of that audience measured against a holdout that saw no ad at all.
• DSP prospecting example: a gardening brand like Epic Gardening runs online video to an in-market audience that has never bought before. Proof it worked is new-to-brand rate, not view count.

The format determines what "worked" even means. A search ad proves itself on intent capture. A DSP ad proves itself on incremental lift. Judging both by the same metric — usually last-click ROAS — is where ecommerce ad budgets quietly leak.

A full worked example, start to finish

Most pages on this topic stop at the creative. Here's one example carried all the way through, because impression-to-purchase is the part nobody publishes.

Across 30 advertisers in reMKTR's Amazon DSP book in July 2026 — the whole portfolio, not the best line item — the campaigns ran 78.4 million impressions at a $4.00 CPM, with a blended $1.42 cost-per-click (the $0.41 CPC figure that circulates in this category is online-video only, not the full mix of formats). That spend produced 57,137 attributed purchases at a blended $5.49 cost per acquisition, rolling up to a 6.04x return on ad spend across the book. 20.1% of those purchases came from a shopper new to the brand — the number that matters most and the one almost never published, because it's the only one that shows whether the ad grew the customer base or just closed a sale that was already coming.

That's what a real example looks like end to end: impressions, cost, clicks, purchases, and the split between new and existing customers. A screenshot of the ad creative is the smallest part of that story.

The mistake: judging any of this on last-click alone

Last-click attribution hands 100% of the credit to whichever ad the shopper clicked last, ignoring every impression before it. Run Sponsored Products and DSP at the same time and this becomes a real problem — both channels can claim credit for the same purchase, and the dashboard shows a bigger number than what actually happened.

The fix isn't a better dashboard. It's reconciling the two in Amazon Marketing Cloud, where sponsored ads and DSP exposure sit in the same log-level data and can be de-duplicated, and testing with holdouts and matched control groups — showing the ad to one group, withholding it from a comparable one, and measuring the actual difference in purchase rate. Last-click can't prove incrementality. It never could. It only shows which ad was closest to the sale, not whether the sale needed the ad at all.

We've made this mistake ourselves: a campaign that looked strong on last-click ROAS and shrank considerably once measured against a holdout, because most of what it was "winning" was demand that would have converted anyway. That's not a reason to stop measuring — it's the reason to measure this way.

When the example looks good but the number is wrong

A few signs the number in front of you is flattering the campaign, not describing it:

• High ROAS, low new-to-brand rate. The ad is closing sales organic search or repeat customers were already going to make. That's not growth, and if new-to-brand isn't broken out, ask why.
• CPC dropped, CPA didn't. Cheaper clicks that don't convert usually mean the audience widened past the people actually likely to buy.
• ROAS looks flat across DSP and sponsored ads reported separately. If both channels reach the same shoppers, the combined number is double-counted, not additive. Reconciling in one dataset — not two reports — is the only way to see the real total.

None of these mean stop the campaign. They mean the metric answers a different question than the one you're asking.

Where this fits

Most of what ranks for "ecommerce advertising examples" is a swipe file — useful for the creative brief, silent on whether the ad actually changed anything. reMKTR runs Amazon DSP as a managed service across 109 live advertiser seats, reconciles DSP and sponsored ads in Amazon Marketing Cloud so they stop double-counting each other, and measures with holdouts rather than last-click. That's not the only honest way to run ecommerce ads — plenty of brands are well served by a creative-first agency or an in-house social team. It's the honest way to answer the question this page started with: not what does the ad look like, but did it work.