The Sponsored Products bid strategy dropdown is three options. Most accounts we audit have left it on “Dynamic – down only” because that’s the default, and most agencies we replace had switched everything to “Dynamic – up and down” because it sounded more sophisticated. Neither default is right for every campaign. Both can quietly burn 15–30% of your ad budget on the wrong campaign type.
After managing 90+ Amazon accounts through 2025 and the first half of 2026, here’s the framework we use to assign a bid strategy to every campaign — and the specific scenarios where each of the three options wins.
What Amazon’s three bid strategies actually do
A refresher because the official documentation is fuzzy:
- Dynamic – down only. Amazon lowers your bid in real time when the algorithm predicts a conversion is unlikely. Your set bid is the ceiling. Bids only move down.
- Dynamic – up and down. Amazon raises your bid (up to 100% for top-of-search, 50% for product pages) when the algorithm predicts a conversion is likely. It also lowers when conversion is unlikely.
- Fixed bids. Your bid is your bid. Amazon does not adjust it in real time. You set it, you eat it.
The marketing copy says “up and down” is the most optimized. The data we see across accounts says it depends entirely on what the campaign is trying to do.
The 4 campaign types and the right strategy for each
We classify every Sponsored Products campaign into one of four buckets. Each bucket has a different right answer.
Campaign Type 1: Branded defense (your own brand terms)
The job: own your branded SERP, prevent competitors from stealing your buyer who already searched for you.
Right strategy: Fixed bids.
Branded defense campaigns convert at 25–60% CVR for most healthy brands. Amazon’s algorithm sees that high CVR and will happily bid your branded keyword bids up to the 100% top-of-search ceiling under “up and down.” That’s a 2x CPC hike on the campaign with the cheapest organic equivalent you have. We see this misallocation routinely cost branded defense campaigns $4,000–$11,000/month of unnecessary spend on accounts at the $200K/mo revenue tier.
Fixed bids at a CPC roughly 60–70% of your category’s non-branded median is the right starting point. Cap the bid, hold the SERP, don’t let the algorithm chase top-of-search auction inflation on traffic that converts at 40% regardless of position.
Campaign Type 2: Conquest / competitor PAT (product-targeting attacks)
The job: appear on competitor PDPs and steal their consideration.
Right strategy: Dynamic – down only.
Conquest campaigns have wildly variable conversion behavior. Some PDPs convert at 8%. Some convert at <1%. The Amazon algorithm cannot reliably predict which competitor PDP your product will convert against without weeks of learning data.
“Up and down” on a conquest campaign assumes the algorithm knows where you’ll convert. It doesn’t yet. You’ll pay 80% CPC premiums on top-of-page placements on competitor PDPs where the data isn’t there to support the bid.
“Down only” lets the algorithm pull bids back where conversion is unlikely (good — saves money) while letting you keep human control of the upside. Pair this with manual placement modifiers (-50% to -100% on the product-page placement modifier for the conquest PDPs that historically tank) to surgically manage exposure.
After 30+ days of conversion data accumulates on specific ASIN targets, you can break the winners out into their own campaigns and consider promoting them to fixed bids at controlled CPC.
Campaign Type 3: Non-branded keyword harvesting (the main growth engine)
The job: discover and convert on non-branded category search terms that drive new-to-brand demand.
Right strategy: Dynamic – up and down — but only after the learning period.
This is the campaign type “up and down” is actually designed for. Non-branded category keywords have enough volume and enough volatility that the algorithm’s real-time CVR prediction earns its keep. When Amazon sees your listing about to convert on a high-intent shopper, the 100% top-of-search uplift is what wins the auction in the moment that matters.
The caveat: “up and down” is a disaster during the campaign learning period (first 14–21 days). The algorithm has no conversion data and will bid up aggressively on impressions that have no business converting. Start every non-branded keyword campaign on Dynamic – down only for the first 21 days, accumulate at least 50 conversions, then flip to up-and-down. The accounts that we see consistently outperform on non-branded growth use this two-phase pattern. The accounts losing money on non-branded use up-and-down from day one.
Track the campaign-level NTB (new-to-brand) rate weekly after the flip. If NTB doesn’t hold above 18–22% after switching to up-and-down, the algorithm is bidding on returning shoppers (your existing customers) instead of net-new acquisition. Flip back to down-only and rebuild placement modifiers manually.
Campaign Type 4: Auto / discovery (the keyword research engine)
The job: surface new converting search terms and ASIN targets you didn’t know about.
Right strategy: Dynamic – down only with low bids.
Auto campaigns are research tools. You want broad surface area at low cost. “Up and down” defeats the purpose — it concentrates your spend on whatever the algorithm thinks is your highest-CVR query, which is usually a branded variation that you don’t need to discover.
Set the bid at roughly 30–40% of your non-branded median CPC. Cap daily budget at $20–$50. Run all four match types (close match, loose match, substitutes, complements) for the first 30 days, then prune to the two that produced converting search terms.
The auto campaign’s job is to feed your manual campaigns harvested search terms. It’s not a performance campaign. Stop optimizing it like one.
The bid strategy decision matrix
For quick reference across an account audit:
| Campaign type | Bid strategy | Why |
|—|—|—|
| Branded defense | Fixed | Algorithm has no upside left to find — CVR is already ceiling |
| Conquest / PAT | Down-only | Conversion behavior too variable for real-time uplift |
| Non-branded — learning (days 0–21) | Down-only | Algorithm needs conversion data before uplift earns ROI |
| Non-branded — mature | Up and down | Real-time CVR prediction is high-value at this scale |
| Auto / discovery | Down-only with low bid | Job is surface area, not optimization |
The two patterns we see burn the most ad spend
After auditing accounts that come to us with ACOS problems, two bid strategy mistakes account for most of the waste:
Pattern 1: Everything on Dynamic up-and-down.
Branded defense, conquest, auto, mature non-branded — all on the same setting. Branded campaigns bid up to the 100% ceiling on traffic that would have converted at half the CPC. Conquest campaigns burn through daily budget on top-of-page placements that don’t convert. Auto campaigns concentrate spend on the queries you don’t need to discover. We typically see 18–32% wasted spend across the account.
Pattern 2: Everything on Dynamic down-only because “it’s safer.”
Safer in isolation, but the mature non-branded keyword campaigns lose the auction during high-intent shopping moments to competitors using up-and-down. Top-of-search impression share on the keywords that drive new-to-brand acquisition collapses. The account looks “efficient” on a TACoS basis while organic non-branded sessions shrink quarter over quarter. The slow leak is worse than the fast burn.
The right answer is assigning a strategy to each campaign based on its job, not setting one default account-wide and forgetting it.
How to audit your account in 20 minutes
Bid strategy is not a “set it and forget it” account-level decision. It’s a per-campaign assignment based on what each campaign is trying to do. Accounts that treat it that way run consistently 4–8 points lower on ACOS than accounts that don’t.
FAQ
How often should I revisit bid strategy on existing campaigns?
Quarterly at minimum, monthly during high-volume periods (Q4, Prime Day windows). Strategy that worked in steady-state can become wrong when category demand spikes and the algorithm’s CVR prediction model has to recalibrate.
Does this framework apply to Sponsored Brands and Sponsored Display the same way?
Partially. Sponsored Brands has the same three bid strategy options and roughly the same logic. Sponsored Display uses a different bidding model (vCPM-eligible for some placements) — we cover that separately.
What about the new “Rule-based” bidding option Amazon rolled out in late 2025?
Useful for ROAS-floor enforcement on mature campaigns with consistent conversion patterns. Not a default. Use it after a campaign has 60+ days of conversion data and you’ve identified the ROAS floor that you’d never want to dip below.
Do these recommendations change for AI-driven bidding tools like Trellis, Pacvue, or Perpetua?
Yes — and that’s a longer conversation. Most algorithmic bidders override Amazon’s native bid strategy entirely. Whether they outperform depends on the campaign type, account scale, and how well the tool has been configured for your specific goals. We’ve seen them win and we’ve seen them lose. Same auditing discipline applies.
Is “up-and-down” really capped at 100% top-of-search, or did Amazon change that?
As of mid-2026 the documented ceilings remain 100% top-of-search and 50% product pages. Amazon has not raised them. If your CPC reports are showing higher multiples than that, check whether you have placement modifiers stacked on top.
—
If you’re looking for a team that manages every lever — creative, advertising, and operations — Velocity Sellers works with brands doing $100K+/month on Amazon. Contact us for a free account audit.
