Amazon Ads Fundamentals
Retail media: a different category from search and social
Amazon Ads is the largest example of what's now called 'retail media' — advertising sold directly by a retailer, served within or around its own shopping experience, targeting shoppers who are already on the retailer's platform actively browsing or searching for products to buy. This is a structurally different proposition from search ads (capturing intent expressed through a query, on a platform that isn't itself the point of purchase) or social ads (interrupting a content-consumption session with an ad, on a platform where purchase intent has to be created rather than already present).
On Amazon, every ad impression happens inside a purchase-ready context: the shopper is either actively searching for a product category or already looking at a competing or adjacent product's page. This proximity to the actual purchase moment is retail media's core advantage, and it's the reason Amazon Ads has grown into one of the largest ad businesses globally despite being a comparatively recent entrant next to Google and Meta.
The three core ad types
Sponsored Products are the workhorse format: ads for individual product listings that appear directly within search results and on competing product detail pages, visually near-identical to organic listings except for a small 'Sponsored' label. They're keyword- or product-targeted, bid on through an auction similar in mechanics to search ads elsewhere, and almost always the first format a seller runs because they drive directly to a purchase-ready listing.
Sponsored Brands are higher up the search results page — a banner-style placement featuring a brand logo, a custom headline, and a small selection of products, designed to build brand presence within the search results rather than promote one SKU alone. They require a registered brand (via Amazon Brand Registry) and work well for sellers with a multi-product catalog who want to establish brand recognition across several related searches rather than compete SKU-by-SKU.
Sponsored Display extends reach beyond Amazon's own search results — onto product detail pages, and critically, off-Amazon entirely across other websites and apps, using Amazon's own shopper and purchase-behavior data to retarget. This is Amazon operating as something closer to a DSP: using its unmatched first-party purchase-intent data to follow a shopper's buying signal outside its own marketplace.
Amazon's three core ad formats
| Format | Placement | Targeting | Primary use |
|---|---|---|---|
| Sponsored Products | In search results, on product pages | Keyword or product (ASIN) targeting | Direct SKU-level sales, always the starting format |
| Sponsored Brands | Top/middle of search results, banner style | Keyword targeting, requires Brand Registry | Multi-product brand presence in search |
| Sponsored Display | Product pages, off-Amazon retargeting | Audience/interest, product, and remarketing targeting | Retargeting shoppers who viewed but didn't buy, on and off Amazon |
ACOS: the metric retail media is measured on
ACOS (Advertising Cost of Sale) is Amazon Ads' primary efficiency metric, calculated as ad spend divided by attributed sales revenue, expressed as a percentage: spend $20 to generate $100 in attributed sales, and ACOS is 20%. This is effectively the inverse framing of ROAS (Return on Ad Spend) used elsewhere — a lower ACOS means better efficiency, the opposite direction from ROAS where higher is better, which trips up advertisers moving between Amazon and other platforms if they don't consciously flip the mental model.
The formula: ACOS = (Ad Spend ÷ Attributed Sales) × 100. A 'break-even ACOS' is the ACOS percentage at which ad-driven sales exactly cover ad spend given the product's margin — a product with a 40% gross margin breaks even at roughly 40% ACOS before ad spend eats into profit entirely, making margin awareness essential before setting any ACOS target.
ACOS = (Ad Spend / Attributed Ad Sales) × 100
Example: $500 spend → $2,500 ad-attributed sales
ACOS = (500 / 2,500) × 100 = 20%
TACOS = (Ad Spend / Total Sales [ad + organic]) × 100
Example: $500 spend → $2,500 ad sales + $7,500 organic sales = $10,000 total
TACOS = (500 / 10,000) × 100 = 5%
Same campaign, same spend — ACOS looks moderate (20%),
TACOS shows the true account-level picture (5%),
because organic sales dwarf ad-attributed sales here.TACOS: zooming out from the campaign to the business
TACOS (Total Advertising Cost of Sale) divides ad spend by total sales revenue — both ad-attributed and fully organic — rather than ad-attributed sales alone. This answers a different and arguably more important question than ACOS: what percentage of the entire business's revenue is being spent on ads, period, regardless of whether any individual sale gets attributed to an ad click.
The reason this distinction matters: a seller can have a high ACOS on a specific campaign while their TACOS stays low and healthy, if that advertising is driving meaningful organic sales lift alongside its direct attributed sales — ads improving a product's search ranking and visibility, which then generates organic sales the ACOS calculation never credits the ad spend for. Sellers who only watch ACOS can wrongly conclude a campaign is inefficient and cut it, when TACOS would show the campaign is healthily supporting a much larger organic sales base than its direct attribution reveals.
Falling TACOS as a brand grows is the real success signal
How retail media differs from search and social at a strategic level
Beyond the mechanical differences in ad format, retail media changes what advertising is even trying to accomplish. On Google Search, an ad competes for a click that then has to travel to an external site and complete a separate conversion funnel — the ad platform and the point of sale are different systems. On Amazon, the ad, the product listing, the reviews, the price, and the checkout are all one continuous, frictionless surface the shopper never leaves. This collapses much of the traditional funnel: there's no separate landing page to optimize, no checkout abandonment across a domain boundary (the cross-domain tracking problem largely doesn't exist within Amazon's own ecosystem), and conversion rate benchmarks run dramatically higher than equivalent off-platform display or social campaigns simply because the purchase friction is near zero.
The tradeoff is control: a seller doesn't own the customer relationship, the checkout experience, or the post-purchase data the way they would on their own site — Amazon owns the customer, the seller rents access to them, and that rent (referral fees plus ad spend) is a structurally different cost of doing business than owned-site e-commerce with its own paid acquisition stack.
Attribution windows differ from other platforms and inflate perceived performance
What's next
The e-commerce event structure covered in E-commerce Tracking in GA4 is worth comparing against Amazon's own attribution model — seeing how a self-hosted store's conversion tracking differs from a closed retail-media ecosystem's built-in attribution clarifies why the two require entirely different measurement approaches.
Next: E-commerce Tracking in GA4 →
I build these systems professionally.
Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.