Amazon Ads Bidding Strategy, Explained With the Actual Math

There isn't one 'bidding strategy' — there are two different systems

"Amazon ads bidding strategy" means two different things depending on which product you're running. Inside Sponsored Products and Sponsored Brands, it's a specific setting in Campaign Manager: fixed, dynamic bids down only, or dynamic bids up and down. Inside Amazon DSP, there's no equivalent dropdown — you pick an optimization goal and an algorithm sets the bid on every impression in real time.

Most searches for this phrase land on the first meaning, so that's where we'll start. But if you're running both sponsored ads and DSP for the same brand — which is where most of the spend in this category ends up — the second half matters just as much, because the two systems don't talk to each other the way you'd expect.

The three Sponsored Products bid strategies, and the math behind each

All three bid strategies start from the same number: the bid you type in. What differs is how far Amazon is allowed to move it before the auction happens.

• Fixed bids: your bid is your bid. Amazon never raises or lowers it based on predicted performance. What you set is what enters the auction, every time.
• Dynamic bids — down only: Amazon can lower your bid for impressions it predicts are less likely to convert. It will never raise it above what you set.
• Dynamic bids — up and down: Amazon can raise your bid, by up to 100%, for impressions it predicts are more likely to convert — and lower it for impressions it predicts are less likely to. A $1.00 bid can become a $2.00 bid on a placement it thinks will convert, and well under $1.00 on one it doesn't.

That 100% ceiling is the number people miss. If you set a $1.00 bid and select dynamic up and down, plan for the possibility of paying $2.00 for some clicks — not as a glitch, but as the strategy working as designed. If that number would break your unit economics, fixed or down-only is the safer starting point, not up-and-down with a lower base bid to compensate.

When each one is actually the right call

Fixed bids earn their keep on new campaigns and on any test where you need a clean read of what a bid actually costs. If you're trying to isolate whether raising a bid from $0.75 to $0.95 changes your win rate, dynamic strategies add noise you can't separate from the change you made — Amazon is moving the bid too, for reasons you can't fully see.

Dynamic down-only fits defensive campaigns: branded terms, campaigns with a proven conversion history, situations where the job is protecting margin on traffic you already know will show up. Dynamic up-and-down fits the opposite goal — sales volume — and works best once a campaign has enough conversion history for Amazon's prediction to be worth trusting. A campaign with three clicks and no sales gives the algorithm nothing to act on; for the first couple of weeks it will behave close to fixed regardless of which dynamic option you picked.

That's the mistake we see most often: choosing dynamic up-and-down on a brand-new ASIN because it sounds like the smarter option, then concluding two weeks later that it "isn't working." It isn't broken. It has no data yet.

What to do when the bid strategy isn't the actual problem

If ACOS is climbing, the instinct is to switch bid strategy. Often that's the wrong lever. Check the search term report first — wasted spend on broad or auto match types shows up there, and no bid strategy fixes a match-type problem. Check placement bid adjustments too; they stack on top of whatever the bid strategy is already doing, and it's easy to have both pulling in the same direction without meaning to.

We've made this mistake ourselves: tightening a campaign onto dynamic down-only to bring CPC down, on a campaign that was already thin on impression share. It worked — CPC dropped — but revenue fell faster than cost did, because the campaign lost top-of-search placements it needed to convert at all. The fix wasn't a strategy switch back to fixed. It was raising the base bid and giving dynamic up-and-down room to actually work, instead of starving it.

Bidding strategy means something else entirely in Amazon DSP

DSP doesn't have a fixed-vs-dynamic choice per keyword because there's no keyword. You choose what the campaign is optimizing toward — reach, conversions, return on ad spend — and the algorithm bids per impression across the available pool, adjusting in real time as it learns which impressions are converting. There's no equivalent of typing in a $1.00 bid and watching it double.

What that means in practice, from a live book: across 30 advertisers we ran in July 2026, the portfolio delivered 6.04x return on ad spend, measured across the whole book rather than one best-performing line item. That same set of campaigns ran 78.4 million impressions at a $4.00 CPM and a blended $1.42 cost-per-click — the $0.41 CPC figure that circulates in this category is an online-video-only number, not a whole-book number, and the two get conflated constantly. Blended cost per acquisition was $5.49 across 57,137 attributed purchases, and 20.1% of those purchases came from a shopper new to the brand.

The harder question with DSP bidding isn't "what's my CPM" — it's "did this actually cause a sale, or just get credit for one that was happening anyway." Last-click attribution can't answer that; it hands credit to whichever touchpoint happened last, regardless of cause. Holdout groups and matched controls can, because they compare exposed shoppers to a comparable group that wasn't exposed. That's a different measurement problem than bid strategy, but it's the one that actually determines whether the bidding is working.

Where this fits into a bigger media plan

Setting a Sponsored Products bid strategy is something any advertiser can do themselves in Campaign Manager — it doesn't require an agency. Where a managed service earns its fee is upstream of that: choosing what DSP should be optimizing toward, and proving with a holdout or matched control that it actually moved sales rather than just re-claiming credit from a click that was going to happen anyway. reMKTR runs that as a managed service across 109 live DSP seats, reconciling in Amazon Marketing Cloud so DSP and sponsored ads stop double-counting each other. It's part of what the wider Full Circle group does across 70+ live brands and $500M+ in managed Amazon spend — worth knowing whether or not this is the page where you decide to call anyone.