Ad Auction

The Amazon ad auction is the real-time mechanism Amazon runs to decide which Sponsored ads to serve on each individual ad impression — for every search query, every detail-page view, every eligible placement. It runs millions of times per second across the marketplace and is the single mechanic underneath everything that happens in Amazon Advertising.

Understanding the auction — even at a simplified level — separates operators who tune bids by gut from those who tune them by mechanism.

The auction in one paragraph

For each impression, Amazon collects all eligible bids from advertisers whose targeting matches the query/page. It modifies each bid by ad relevance and predicted performance (CTR, CVR), applies the advertiser's dynamic-bidding policy and any placement modifiers, ranks the resulting effective bids, and awards each available slot to the highest-ranked ad in descending order. The price paid is set on a second-price basis — the winner pays just enough to beat the next-highest ad, not their own bid.

The simplified bid math

For a single ad in a single auction:

Effective Bid = Base Bid
              × (1 + Placement Modifier)
              × Dynamic-Bid Shading (0.5×–2.0×)
              × Relevance Score (Amazon internal)

Then ranked against other effective bids. The slot allocation is determined by rank; the price paid is second-price (next-highest effective bid ÷ your relevance, plus €0.01).

This is why two advertisers with identical base bids can have very different CPCs and win rates — relevance and dynamic shading drive the wedge.

Auction inputs

Input	Source	Controllable?
Base bid	Your campaign setup	Yes
Placement modifier	Your campaign setup	Yes (0–900% per placement)
Dynamic-bid shading	Your bidding strategy + Amazon's CVR prediction	Partly (you choose policy)
Relevance score	Amazon's keyword-to-ASIN relevance model	Indirectly (listing quality, search-term match, CTR history)
Competing bids	Other advertisers' choices	No

Most operators only think about the first input. The other four explain most CPC variance.

The second-price mechanism

A simplified worked example. Three advertisers compete for one slot:

Advertiser	Bid	Relevance	Effective rank
A	€1.20	1.0	1.20
B	€1.50	0.7	1.05
C	€0.90	1.0	0.90

Advertiser A wins. They pay just enough to beat B's effective rank of 1.05 — so A's CPC is approximately €1.05 / 1.0 + €0.01 = €1.06, not their €1.20 bid.

The key implication: your bid is a max-willing-to-pay, not the price you'll pay. Bid optimisation that ignores second-price mechanics tends to over-shade bids downward defensively, leaving impression share on the table.

Auction pressure

Auction pressure = competitive density of bids on a query/placement. High pressure → high CPCs, low impression share at any given bid. Low pressure → cheap clicks, high impression share.

Pressure varies massively:

Seasonal: Q4 auctions are 1.5–3× more expensive than Q1 in most categories.
Diurnal: Auction pressure peaks evenings/weekends in DACH; off-peak slots are cheaper for the same effective CVR.
Query type: Branded queries are low-pressure (you usually win at floor); top category terms are high-pressure (every player bids).
Placement: Top-of-search auction is 2–4× more expensive than rest-of-search.

Common mistakes

Bidding the click cost you want to pay. Bid is the max; second-price means you'll usually pay less. Under-bidding loses impression share for no CPC saving.
Ignoring relevance as an input. A high-quality listing wins auctions at lower bids than a low-quality listing — listing optimisation is bid optimisation.
Treating dynamic shading as opaque. Down Only and Up & Down move bids by up to 100%; the chosen policy can multiply CPC variance.
Bidding the same on top-of-search and rest-of-search. The auctions are different pressure levels; the same base bid + ToS modifier is the standard fix.