Amazon PPC Explained: The Auction, the Math, and the Mistakes

What Amazon PPC actually is

Amazon PPC covers four ad products, and they behave differently enough that lumping them together is where most confusion starts. Sponsored Products puts a single listing into search results and product pages, targeted by keyword or by competitor ASIN. Sponsored Brands puts your logo and a headline above search results, targeted by keyword only. Sponsored Display retargets shoppers who viewed your listing or a competitor's, on and off Amazon. DSP is programmatic display bought through a managed console or an agency, usually with a spend minimum well above what a new seller runs.

All four run on the same basic mechanic: you name a maximum bid for a target, Amazon runs an instant auction every time a shopper's search or browsing matches that target, and the ad shows if you win. You are charged only if the ad is clicked. Impressions where nobody clicks cost you nothing directly, though they still burn through your daily budget's headroom in the sense that Amazon won't keep showing an ad indefinitely if nothing ever converts.

The auction isn't a pure high-bid-wins system. Amazon factors in your bid, the relevance of your listing to the search term, and your historical click-through and conversion performance on that term. A lower bid with a stronger conversion history can beat a higher bid with a weak one. This is why two sellers bidding the same amount on the same keyword often see very different results.

How the auction sets your price — a worked example

Say you set a maximum bid of $1.20 on a keyword. The next-highest relevant bidder in that auction is at $0.85. You don't pay your bid — you pay one cent above theirs, so your actual cost per click is $0.86. This is the second-price mechanic that most explainers mention and then never do the arithmetic on.

Run that keyword for a week and say it draws 500 clicks at an average CPC of $0.86. That's $430 in spend. If 10% of those clicks convert, you've made 50 sales. At an average sale price of $28, that's $1,400 in ad-attributed revenue.

ACoS — advertising cost of sale — is spend divided by ad revenue: $430 ÷ $1,400 = 30.7%. Whether that's good or bad depends entirely on the next section, because 30.7% is either comfortably profitable or a slow bleed depending on your margin, and no guide can tell you which without knowing your numbers.

The metric that decides if 30.7% is good news

Your breakeven ACoS is your margin before ad spend, expressed as a percentage. If this product carries a 35% margin before advertising, you can spend up to 35% of ad revenue on clicks and still break even. At 30.7%, that campaign is profitable — you're keeping about 4.3 points of margin after the ad spend on every unit the ads sold.

The number most sellers stop tracking is TACoS — total advertising cost of sale, which is ad spend divided by total sales, organic included. If this listing also did $2,800 in organic sales that week, total sales were $4,200 and TACoS is $430 ÷ $4,200 = 10.2%. TACoS falling over time, even while ACoS holds steady, is the signal that PPC is building organic rank rather than just renting it. TACoS climbing while ACoS looks fine is the signal that organic sales are drying up and the ads are propping the listing up alone.

• ACoS tells you if a campaign paid for itself.
• TACoS tells you if advertising is building something durable or just buying today's sales.
• Breakeven ACoS is the number that turns either metric into a decision.

Where sellers go wrong — including mistakes we've made

The most common mistake is treating automatic and manual campaigns as competitors instead of a pipeline. Run automatic first, let it run one to two weeks, pull the search term report, move every term that converted into a manual exact-match campaign you control, and add that same term as a negative in the automatic campaign so the two stop competing for the same click. Most sellers do this loop once, get busy, and never do it again. The account keeps running on month-two targeting for the next two years.

The second mistake is chasing ACoS to zero. A campaign at 5% ACoS that gets 40 clicks a month isn't a win, it's a campaign too small to matter, and cutting bids further to protect that number often costs more in lost rank than it saves in spend. We've made this exact mistake ourselves — pulling back a keyword after three days of no conversions when the term needed ten to fourteen days of data to say anything reliable.

The third is skipping the measurement step entirely. A bid change, a new negative, a budget increase — each one needs a reason before it runs and a way to tell afterward if it worked. Every proposed change carries three things before it runs: the evidence behind it, a measurement plan, and a rollback trigger. That discipline is the difference between an account that improves and one that just moves.

What to do when the number is bad news

If ACoS is above breakeven, don't cut the bid first — check whether the conversion rate on that term is normal for the listing. A high CPC with a normal conversion rate is a bidding problem; a normal CPC with a collapsed conversion rate is usually a listing, price, or stock problem the ad spend is just exposing faster.

If a keyword has spent through 15–20 clicks with zero conversions, negate it before it spends through 40. If a campaign's TACoS keeps rising while ACoS looks stable, check organic rank on the same terms — the ads may be carrying a listing that lost its organic position.

If a bid change didn't move impression share after five to seven days, the bid probably wasn't the constraint — relevance or conversion history likely was, and no further bid increase will fix that on its own.

Where this gets managed once you're past the manual stage

Everything above is the mechanic — the loop of harvest, promote, negate, measure, adjust, repeat, across every SKU, every week, for as long as the account runs. Doing it by hand works at ten campaigns. It stops working at two hundred. Dr. PPC, from drppc.ai, reads the account, writes a strategy per product against that product's real margin, and proposes each change with the evidence, the measurement plan, and the rollback trigger already attached — the same discipline described above, run at account scale, with the reader choosing how much of it runs without a human click. It's $300 a month plus a capped 3% of ad spend, month-to-month, first 30 days free, and it's built and operated by Full Circle, which has managed over $500M in revenue across more than 100 brands. Whether or not you ever use it, the math in this page is the same math it runs — you now know what it's checking.