Amazon PPC: How It Actually Works in 2026
Amazon PPC is Amazon’s pay per click ad system. You bid for slots on search results pages and product detail pages, and you pay only when a shopper clicks. Four surfaces run on it: Sponsored Products, Sponsored Brands, Sponsored Display, and Streaming TV. Amazon settles each click in an auction. Its own cost per click guide says the final price rests on your adjusted bid plus additional factors, and nothing more precise than that. Much of what sellers repeat about that auction has no Amazon source behind it.
What pay per click on Amazon is, in one paragraph
A shopper types a query. Some results carry a small Sponsored label. Those slots went to advertisers in an auction that ran before the page loaded. Nobody paid for the slot. They paid for the click.
Impressions cost you nothing. A click costs you money whether or not it converts. So the discipline sits in one question: which queries deserve a paid click, and what is that click worth to you?
Four surfaces sit under Amazon Ads, and one name changed recently. Sponsored TV now carries the streaming TV name, and Amazon says existing Sponsored TV campaigns keep running without interruption. Most guides still teach the old label.
| Surface | Where it appears | What it promotes | Eligibility | Typical use |
|---|---|---|---|---|
| Sponsored Products | Search results and product pages | One ASIN | Buy Box eligible sellers and vendors | Volume, the default start |
| Sponsored Brands | Top of search and video slots | Logo, headline, several ASINs, or video | Brand Registry | Category and brand defence |
| Sponsored Display | Product pages, plus off Amazon | Products and audiences | Brand Registry, some sellers | Retargeting and competitor pages |
| Streaming TV | Streaming inventory, formerly Sponsored TV | Brand video creative | Brand Registry | Upper funnel reach |
Where your ad can appear
Sponsored Products ads land in three placement groups. The console names them top of search, rest of search, and product pages. You can bid against each group separately, which is the closest thing to placement control the ad platform gives you.
Amazon’s Sponsored Products guide states that you can add placement bid adjustments up to 900%. That is a multiplier on your base bid for one placement group. A one dollar bid with a 900 percent top of search adjustment can clear at ten dollars, before dynamic bidding touches it.
Rest of search was not adjustable for years. Amazon added the rest of search adjustment on 9 January 2024. Anything written before that date is stale on this point.
How the auction sets your price
Start with what Amazon publishes, because it is short. The cost per click guide says the final price comes from an auction, and rests on your adjusted bid plus additional factors. Amazon adds that the auction considers factors beyond bid when it sets placement and price. Your charge does not go above your maximum adjusted bid.
That is the whole documented mechanism. Amazon names no formula and no runner up rule.
Now the claim you read everywhere else. The industry says Amazon runs a second price auction, where you pay one cent more than the next bidder. That sentence appears in no Amazon guide, help page, or filing. It is folklore, borrowed from how Google Ads worked before 2021, then repeated until it sounded like documentation.
Under a strict second price rule, your bid would only ever set a ceiling. Under what Amazon describes, relevance and other unnamed inputs sit inside the same calculation. Two advertisers with identical bids can pay different prices for the same query.
You can rely on the ceiling. The charge stays at or below the adjusted version of your bid. You cannot rely on a clean rule for the gap between the two.
We track that gap across the accounts we run, because it decides how hard you can bid without losing control of spend. The pattern holds inside a category, and moves once you cross into another.
We do not publish those figures, because a median pulled from one category misleads in another. Build your own: divide actual spend by clicks for each campaign, compare that against the bid you set, and repeat it per category.
Documented versus repeated
| What the industry repeats | What Amazon publishes |
|---|---|
| You pay one cent above the next bidder | Final cost comes from an auction and rests on your adjusted bid plus additional factors. |
| Ten dollars a day is the platform minimum | “We recommend starting with a daily budget of $10 or equivalent”. |
| TACOS is an Amazon metric | TACOS appears in no Amazon advertising guide. Amazon defines ACOS and ROAS. |
| There are three sponsored ad types | Four surfaces. Sponsored TV now carries the streaming TV name. |
| Your category has an average ACOS to hit | “There isn’t a definitive number for a good Amazon ACOS”. |
The three bidding strategies
Amazon gives Sponsored Products three settings. Their exact names are dynamic bids down only, dynamic bids up and down, and fixed bids.
One number in that list surprises people. Amazon’s dynamic bidding guide is blunt about up and down. It says Amazon will “increase or decrease your bids by up to 100% for all placements based on performance”. A one dollar bid can therefore clear at two dollars. Stack a placement adjustment on top and the ceiling climbs again. Sellers who budget on the number they typed get a shock in week one.
| Strategy | How Amazon adjusts | Best use | Risk |
|---|---|---|---|
| Dynamic bids, down only | Lowers the bid when conversion looks unlikely | New campaigns, thin margins | Slow data, low impression volume |
| Dynamic bids, up and down | Raises or lowers by up to 100 percent | Proven exact match terms | Real cost can double your bid |
| Fixed bids | No adjustment at all | Bid testing and brand terms | Pays in full on weak clicks |
Automatic versus manual targeting
Run automatic first, for two to four weeks, then move the winners into manual. That is the answer, and hedging it helps nobody.
Automatic targeting hands keyword selection to Amazon. You get reach fast, plus a search term report full of the real queries that produced your clicks. Manual targeting reverses the trade. You pick the terms, you set a bid per term, and you own the outcome.
The loop between them is the actual work. Your automatic campaign discovers terms. Terms that convert graduate to manual exact match with their own bid. Terms that spend without converting go back into the automatic campaign as negative exact. Repeat weekly. The automatic campaign gets cheaper while the manual campaign gets sharper.
Split budget so the automatic campaign stays funded after graduation, or discovery stops. Judge a term on orders, not clicks, before you promote it.
Match types and negatives
Manual campaigns give you five targeting shapes. Broad match catches the seed keyword in any order, with extra words and close variants. Phrase match keeps the word order and allows text on either side. Exact match takes the term and its close variants only. Product targeting aims at one ASIN. Category targeting aims at a browse node, with brand and price filters.
Bid them differently, because they carry different risk. Take one seed keyword, “stainless steel water bottle”, across three campaigns.
- Broad, low bid, wide net, heavy negative upkeep.
- Phrase, middle bid, tighter intent, steadier cost.
- Exact, top bid, the term you already know converts.
Negatives work the same way in reverse. Negative exact blocks one term and leaves everything around it alive. Negative phrase blocks a whole family of queries. Sellers reach for negative phrase because it feels efficient, then kill profitable long tail traffic they never audited.
Use negative exact when the search term report shows one bad query. Use negative phrase only for a word that can never suit your product, like “kids” on an adult range.
A question the forums ask and nobody answers: yes, you can undo a negative keyword. Open the campaign or ad group, go to the negative keywords tab, and archive the entry. Archiving is permanent, so the term will not come back to that list, but the query becomes eligible again straight away. There is no restore button, so add it back by hand.
The metrics, defined the way Amazon defines them
ACOS is ad spend divided by ad revenue, times one hundred. Spend $200, make $1,000 in ad sales, and your ACOS is 20 percent. Amazon’s ACOS guide adds the line every competitor page ignores. There isn’t a definitive number for a good ACOS, and it depends on your industry, company size, and campaign frequency.
ROAS runs the other way, revenue divided by spend. Amazon’s ROAS guide cites a Nielsen benchmark of 2:1, which it calls a bit over the current industry average. It then says a brand would ideally want ROAS closer to 3 or 4. Those are Amazon’s words about advertising in general, not a promise for your SKU.
TACOS, total ad cost of sale, divides ad spend by total sales rather than ad sales. It is useful. It is also not an Amazon metric. The term appears in none of Amazon’s published advertising guides, and no console report calculates it for you. Track it yourself, as an industry convention that shows whether paid spend pulls organic sales up with it.
Break even ACOS, from your own fee stack
Break even ACOS is the only ACOS number that means anything. It comes from your fees, not someone else’s average. Take the money left after every cost except advertising, divide it by the sale price, and that percentage is where a campaign stops adding profit.
Work one product through it. A bottle sells at $29.99. Landed cost of goods is $7.50. The referral fee runs at 15 percent, the rate that applies across many categories, so $4.50. The FBA fulfilment fee for its size tier is $5.60. What remains is $12.39, and $12.39 divided by $29.99 gives a break even ACOS of 41.3 percent.
Those figures are an example, not a benchmark. Pull your own from the Seller Central fee preview. Referral rates differ by category, and fulfilment fees follow size and weight.
| Sale price | COGS | Referral fee | FBA fee | Contribution | Break even ACOS |
|---|---|---|---|---|---|
| $29.99 | $7.50 | $4.50 | $5.60 | $12.39 | 41.3% |
At 20 percent ACOS that bottle keeps about half its contribution on every ad sale. At 41 percent it breaks even. Above that you are buying rank, a fair choice during a launch and an expensive habit afterwards.
Budgets and what Amazon actually requires
Amazon recommends starting at $10 per campaign per day. Read the sentence carefully. It says “we recommend”, not “the minimum is”. You can run a campaign below that figure, and Amazon publishes no platform floor for Sponsored Products daily budgets.
The distinction matters for small tests. A $5 daily budget on one tight exact match campaign gathers data slowly, and rarely enough clicks per term for a decision. Give a low budget test a longer window.
Then the question nobody answers well: why did spend go past the daily budget? Amazon’s public advertising guides do not document a daily spend cap or an averaging rule. Anyone quoting an exact overage percentage is guessing. In practice, two things happen. Clicks arrive faster than the ad server meters them near the cap, and reporting settles later once invalid clicks drop out. Watch spend across a full week.
What changed in Amazon Ads during 2026
Unified reporting reached general availability on 8 June 2026. One report can now pull campaigns across multiple manager or advertiser accounts, ad products, countries, metrics, and dimensions. Agencies and multi marketplace brands stop stitching exports together by hand.
Campaign Manager beta, which began rolling out in late 2025, puts sponsored ads and Amazon DSP in one console. Retail and upper funnel reporting stop living in separate tools.
At unBoxed Toronto on 25 February 2026 Amazon announced five things worth knowing. Creative Agent builds ad creative. Ads Agent inside Amazon Marketing Cloud takes plain English requests for analysis. The Amazon Ads MCP Server lets AI agents connect straight to an ad account. Sponsored Brands reserve share of voice holds visibility on brand queries. Sponsored Brands collections curates product combinations.
Treat the agent tools as drafting help with a human check, not an autopilot.
When Amazon PPC is not worth running
Some products should carry no paid placements at all. Saying so costs us work we would rather not take.
Skip it when contribution margin is thin. If break even ACOS lands under roughly 15 percent, one unconverting click eats the profit from a sale, and no bid discipline fixes arithmetic.
Skip it when the listing is not ready. A product with a handful of reviews, one image, and no A+ content converts badly. Paid traffic then buys clicks that a stronger page would have earned for free. Our note on Amazon SEO covers what comes before any bid.
Skip it when inventory is short. Ads that run into a stockout waste spend and hand your rank to a competitor who stayed available.
Skip it when category cost per click already exceeds your contribution per unit. Check that number before you fund a campaign, not after.
Frequently asked questions
What is Amazon PPC in simple terms?
It is advertising where you pay only when someone clicks your ad. You pick which searches or products to target, set a maximum bid, and Amazon runs an auction for each slot. Winning ads carry a Sponsored label. Impressions are free, clicks cost money, and sales are the point.
Does Amazon use a second price auction?
Amazon has never published that it does. Its cost per click guide says only that the final price comes from an auction, rests on your adjusted bid plus additional factors, and weighs factors beyond bid. The one cent above the runner up claim appears in no Amazon document.
How much does an Amazon click cost?
Amazon publishes no average. Cost per click moves with category, keyword, placement, season, and competition, and it settles inside the auction rather than at a list price. Your own search term report is the only reliable source. Compare actual cost per click against your break even ACOS, not a benchmark.
What is a good ACOS on Amazon?
Amazon states there is no definitive number, and that it depends on industry, company size, and campaign frequency. The useful comparison is your break even ACOS, worked from sale price minus cost of goods, referral fee, and fulfilment fee. Below it, ads add profit. Above it, you buy rank by choice.
Is TACOS an official Amazon metric?
No. Total ad cost of sale appears in none of Amazon’s advertising guides, and no standard console report calculates it. Sellers built the metric because ACOS ignores organic sales. It works as a trend line for whether paid spend lifts total revenue. Treat it as industry vocabulary, not an Amazon definition.
Should I start with automatic or manual campaigns?
Start automatic. For the first two to four weeks it collects real search terms faster than you can guess them. Once the report shows which queries convert, move those into manual exact match at a higher bid. Add the wasteful queries as negative exact, and keep both campaigns running.
What is the minimum daily budget for a Sponsored Products campaign?
Amazon recommends starting at $10 per day per campaign. That is a recommendation, not a platform floor, and campaigns can run below it. A smaller budget still gathers data, just more slowly, so allow a longer test window. Few campaigns at a workable budget beat many starved of clicks.
Why did my campaign spend more than my daily budget?
Amazon’s public guides do not document a daily cap or an averaging rule, so any exact overage percentage you read online has no Amazon source. In practice, clicks can arrive faster than spend meters near the cap, and reporting adjusts later as invalid clicks fall out. Review spend across a week.
Can my cost per click be higher than the bid I set?
Higher than the number you typed, yes. Higher than your adjusted bid, no. Dynamic bids up and down can raise a bid by as much as 100 percent, and placement adjustments reach 900 percent. Both multiply your base figure. Amazon charges at or below that adjusted ceiling, never above it.
Does running PPC improve my organic ranking?
Amazon publishes no direct link between ad spend and organic position. What ads do produce is sales velocity and conversion history on a listing, and those signals matter to the retail algorithm. The effect is indirect and carries no guarantee. Anyone promising rank for ad budget is selling a claim Amazon never made.
How long should I run a campaign before judging it?
Give a new campaign at least two weeks, and judge on click volume rather than the calendar. A keyword with 20 clicks and no order tells you little at most price points. Wait for enough clicks to cover your expected conversion rate, then decide. Cutting on day three kills terms that needed data.
When should a seller skip paid ads altogether?
Skip when the numbers say so. Very thin contribution margin, a listing with almost no reviews or weak images, low or unstable inventory, and categories where cost per click already exceeds contribution per unit. In each case the money works harder on the listing, the price, or the stock.
Where to go from here
Want a second pair of eyes before you change anything? We run a free audit and start with the arithmetic, not the bids. Our Amazon PPC management work comes with one named account manager, a written monthly report, and no lock in. You can also see what an ACOS reduction looks like in a real account, where we cut ACOS from 35 percent to 10 percent in six months.
