Amazon PPC Auction Explained: How Bids Really Work

Priya sells a magnesium supplement. Her exact-match campaign on "magnesium glycinate 400mg" was showing up on page two, so she did the obvious thing: she raised her bid from $1.20 to $2.40.

Impressions went up 6%. Her average CPC went from $0.94 to $1.03. She was still losing top of search to a competitor she knew was bidding less than she was, because he had told her so at a trade show three months earlier.

She raised the bid again, to $3.50. Nothing moved. She had now doubled her ceiling and bought almost no additional visibility.

Priya's problem was not her bid. It was that she was treating the Amazon PPC auction like a price tag, when it works like a scoring system where price is only one input, and not even the one she had the most control over.

Almost every seller learns the tactics before the mechanism. They learn to set bids, adjust placements, split match types, and add negatives, without ever being told how the thing they're feeding actually resolves. That's why bid changes feel random, why CPC never matches the number you typed, and why competitors with smaller budgets sit above you.

Here's what actually happens in the roughly 100 milliseconds between a shopper hitting enter and your ad appearing.

The Short Answer

The Amazon PPC auction is a second-price auction. You don't pay what you bid. You pay one cent more than the minimum needed to beat the next advertiser below you.

Winners are decided by ad rank, not by bid alone. Ad rank combines what you're willing to pay with how likely Amazon thinks your ad is to produce a sale. A high bid on an ad nobody clicks loses to a moderate bid on an ad that converts.

Everything else in this article is detail on those two sentences.

What Happens Before the Auction Even Runs

The auction is the last step, not the first. Before Amazon compares a single bid, it builds the pool of ads allowed to compete for that specific search. Most sellers who think they have an auction problem actually have an eligibility problem.

To enter the auction for a given query, your ad has to clear four gates:

• You own the Featured Offer: if you've lost the Buy Box, your Sponsored Products ads stop serving on that ASIN. Not "serve worse." It stops. This is the single most common cause of a campaign that goes quiet overnight with no settings changed. Our Buy Box guide covers how to diagnose it.
• You're in stock and the listing is active: suppressed listings, out-of-stock variations, and restricted categories are pulled from the pool.
• Your targeting matches the query: match type decides whether you're in the room, not how you rank once you're there. A broad keyword and an exact keyword targeting the same query enter the same auction on equal footing.
• You clear the reserve price: Amazon won't serve an ad below a floor bid, which varies by marketplace and placement. Bids under roughly $0.15–$0.20 frequently sit in "eligible" status and never serve.

Tomas, who sells dog grooming tools on Amazon.de, spent two weeks raising bids on a campaign that had flatlined. The campaign was fine. A supplier delay had put his hero ASIN into a two-day stockout, he'd relisted at a higher price, and a reseller had taken the Featured Offer. His ads had never lost an auction. They had never been in one.

Ad Rank: Bid Is Half the Equation

Once you're in the pool, Amazon scores every competing ad and orders them. That score is what the industry calls ad rank, and it works out to roughly:

Amazon has never published a public quality score the way Google does, and anyone who quotes you a formula with exact weights is guessing. What Amazon has stated plainly, and what every seller's data confirms, is that the platform ranks ads by both bid and relevance, because Amazon gets paid on clicks but makes far more money on the sale behind the click. An ad that converts at 12% is worth more to Amazon at $0.90 than an ad that converts at 3% is at $1.80.

The performance side of that equation is built from signals Amazon already has about your listing and your ad:

• Historical click-through rate for that keyword-ASIN pair
• Conversion rate on the traffic you've already received
• Relevance between the query and your product: title, bullets, backend terms, category, attributes
• Sales velocity and BSR for the ASIN
• Listing health: reviews, star rating, price competitiveness, image quality

This is why Priya's bid increases did nothing. Her competitor's listing had 4,100 reviews at 4.6 stars and converted ad clicks at 14%. Hers had 380 reviews at 4.2 stars and converted at 6%. He could hold top of search at a lower bid because his expected value per impression was more than double hers. She was trying to buy her way past a conversion-rate gap, which is the most expensive way to solve a conversion-rate problem.

Second-Price Bidding: Why Your CPC Is Always Lower Than Your Bid

Amazon runs a second-price auction, the same mechanism auction theory calls a Vickrey auction. The highest ad rank wins the placement, but pays only what it needed to beat the ad ranked below it, plus one cent.

Simplified to make the math visible. Assume for a moment that all four advertisers have identical relevance scores:

You bid $2.40 and paid $1.86. The $0.54 gap is not a discount Amazon gave you. It's the distance between you and the next advertiser, which means it's a measurement of how much competitive headroom you have.

Two consequences follow, and they're the practical payoff of understanding any of this.

Your bid is a ceiling, not a price

Raising a bid you're already winning at usually changes nothing about what you pay. It only matters when someone below you bids past your old ceiling. This is why "raise bids on winners, cut bids on losers" is bad advice applied blindly. Raising a bid on a keyword you already dominate buys nothing, while raising the bid on a keyword where you're consistently outranked may just raise your cost without changing position, because your ad rank gap was never about price.

Your bid-to-CPC gap is a competitive read

Track it per keyword. A keyword bid at $2.00 delivering $0.80 CPC is uncontested. You could probably lower the bid with no loss of position, or leave it as insurance against a competitor waking up. A keyword bid at $2.00 delivering $1.94 CPC is a knife fight, and every cent of ad spend there is being extracted by the auction.

Rachel, who sells silicone baking mats, went through her top 40 keywords and split them by this ratio. Eleven keywords were running at under 45% of bid. She cut those bids by 30% and lost 4% of impressions while saving $610 in the month. Nine keywords were running above 90% of bid. Those were the ones she checked her margins on, and three of them she stopped bidding for entirely because the auction price had drifted above what the product could carry.

There Isn't One Auction. There Are Several

This is where most explanations stop, and where the practical money is.

Amazon doesn't run one auction per search. It runs separate auctions for each placement group, and they have completely different competitive dynamics:

• Top of Search (first page): the carousel above organic results. Highest CTR, highest conversion rate, most expensive, most contested.
• Rest of Search: ads scattered through and below organic results. Cheaper, lower intent capture.
• Product Pages: ads on competitor and complementary detail pages. Different competitive set entirely, often the cheapest impressions in the account.

A keyword can be wildly profitable at Top of Search and a money pit on Product Pages, or the reverse. Treating them as one number hides both. Our placements guide has the full report walkthrough; the point here is that when you set one bid, you're entering three different auctions with it.

How the Effective Bid Gets Built

Your typed bid is rarely the number that enters the auction. Two multipliers stack on top of it, in a specific order.

Step 1: Placement modifier. You set this at the campaign level, 0–900%, per placement group.

Step 2: Dynamic bidding. Amazon adjusts in real time based on how likely it thinks that specific impression is to convert:

• Down only: lowers your bid up to 100% on impressions unlikely to convert. Never raises.
• Up and down: lowers as above, and raises up to 100% for Top of Search, up to 50% for other placements.
• Fixed: no adjustment. Your bid is your bid.

Worked example. Base bid $1.00, Top of Search modifier set to 50%, strategy set to up and down:

• Base bid: $1.00
• After placement modifier (+50%): $1.50
• After dynamic bidding (up to +100%): up to $3.00

That's your ceiling in the Top of Search auction, 3x the number you typed. Sellers who set an aggressive placement modifier and then switch to up-and-down without recalculating are frequently shocked by their CPC, and they blame Amazon rather than the multiplication they signed up for. The bid strategy guide breaks down when each mode is worth using.

The reverse trap is quieter and more common: down-only bidding on a keyword you need for launch velocity means Amazon can silently take your $1.50 effective bid to $0.30 on impressions it doesn't rate, and you never see the auctions you didn't enter.

Match Type Doesn't Change the Auction

Worth stating flatly, because it's a persistent myth: exact match does not get a ranking bonus over broad match in the auction.

Match type controls which queries you're eligible for. Once two of your own keywords are eligible for the same query, Amazon generally serves the one with the higher ad rank, and since exact-match keywords usually carry better historical performance for their specific query, they tend to win. That's an outcome of relevance history, not a structural advantage of the match type.

The practical implication: keyword-level bids across match types are competing with each other, not just with your competitors. If your broad keyword is outbidding your exact keyword on the same term, you're paying broad-match prices for exact-match intent. See the match types guide for the waterfall structure that prevents it.

Six Ways to Win Auctions Cheaper

Everything above points at one conclusion: bid is the least efficient lever you have, because it's the only one your competitors can match instantly.

1. Fix conversion rate first. It's the biggest single input in the ad rank equation, and it compounds. Better CVR wins more auctions at lower cost, which lowers ACoS, which frees budget. A move from 8% to 11% CVR is worth more than any bid change you can make. Start with the conversion rate optimization guide.
2. Improve CTR with the assets in the auction. Your main image, title, price, review count and star rating are your ad creative in Sponsored Products. There is no separate ad to optimize. A main image that reads clearly at thumbnail size on a phone is a bid multiplier you only pay for once.
3. Bid by placement, not by keyword. Pull the placement report, find keywords that convert on Product Pages but not Top of Search, and shift the modifiers. Most accounts have keywords they've written off entirely that are profitable in one placement.
4. Kill the auctions you shouldn't be in. Every irrelevant query you're eligible for drags down the CTR and CVR history attached to that keyword, which lowers your ad rank on the queries you do want. Negatives aren't just spend control; they're relevance hygiene.
5. Compete where competition is thin. Long-tail queries and product-page placements have shallower bidder pools. The bid-to-CPC ratio tells you where those are in your own account, without any third-party tool.
6. Adjust continuously, because the field does. The auction re-runs on every search. Your competitors' budgets exhaust mid-afternoon, new sellers enter on Monday, Q4 pulls in bidders who weren't there in September. A bid you set in March is competing against an October field.

A 20-Minute Auction Audit

Run this the next time performance feels random:

1. Export the last 60 days at keyword level. Add a column: average CPC ÷ bid.
2. Sort ascending. Anything under 0.50 is uncontested. You're overbidding for the position you're getting. Cut those bids 20–30% and watch impressions for a week.
3. Sort descending. Anything over 0.90 is contested. Check whether the resulting ACoS still clears your break-even. Some of these should be abandoned, not optimized. The ACoS guide has the break-even math.
4. Pull the placement report. Find the three keywords with the widest performance gap between Top of Search and Product Pages, and set placement modifiers accordingly.
5. Check Buy Box percentage on every advertised ASIN. Anything under 90% is quietly removing you from auctions you're paying to compete in.
6. Compare your CPCs against category norms in our PPC cost benchmarks. If you're paying 40% above your category median, the problem is relevance, not bidding.

Frequently Asked Questions

Is Amazon PPC a first-price or second-price auction?

Second-price. The winning advertiser pays one cent more than what was required to beat the next-ranked ad, not the full amount they bid. This is why your average CPC is nearly always lower than your maximum bid.

Why is my CPC lower than my bid?

Because your bid is a ceiling. In a second-price auction you pay only enough to hold your position against the advertiser below you. A large gap between bid and CPC means the keyword is lightly contested.

Does raising my Amazon bid guarantee a higher position?

No. Position is determined by ad rank, which combines your effective bid with expected performance: click-through rate, conversion rate, listing relevance, and sales history. A competitor with better conversion metrics can outrank you at a lower bid.

Does Amazon have a quality score like Google Ads?

Not a published one. Amazon has never released a numeric quality score or its weighting. It has confirmed that ads are ranked on both bid and relevance, and account data consistently shows that CTR and CVR history affect how much you must bid to hold a position.

How much can dynamic bidding raise my bid?

With "up and down," Amazon can increase your bid by up to 100% for Top of Search placements and up to 50% for other placements. That's applied after your placement modifier, so a $1.00 bid with a 50% Top of Search modifier can reach $3.00.

Do exact match keywords beat broad match in the auction?

Not structurally. Match type determines which searches you're eligible for. When two of your own keywords are eligible for the same query, the one with the higher ad rank serves, which is usually the exact-match keyword, because it has stronger performance history for that specific term.

How often does the Amazon auction run?

On every single shopper search. There is no daily or hourly batch. Every impression is a fresh auction against whoever is eligible and funded at that moment.

What to Take From This

The auction rewards two things: willingness to pay, and evidence that the click will turn into a sale. Sellers spend nearly all their attention on the first and almost none on the second, because the first has a box you can type a number into.

Bid is the lever everyone can pull. Relevance is the lever that takes work, and it's the one that makes every auction after it cheaper.

Once you understand that, the strange behavior stops being strange. Bid increases that do nothing are auctions you were already winning. Competitors above you at lower bids have better conversion data. CPCs far below bid are uncontested keywords you're overbidding on. Campaigns that die overnight are Buy Box losses, not auction losses.

None of that is visible from the campaign manager's default view. All of it is visible from the numbers you already have. For the official mechanics, Amazon's own Sponsored Products documentation and the Seller Central PPC primer are worth ten minutes.