Key takeaways: Review bids weekly, change them at keyword level, move them 10 to 20% at a time and judge each change on two weeks of data. Set every bid from the keyword's own conversion rate and the product's target ACoS, not from a blanket rule. Daily bid changes react to noise; monthly changes leave money on the table.
Amazon bid optimization works best on a weekly cycle. Most keywords need a week or two to collect enough clicks for a decision, and Amazon's attribution window means the last few days of any report are incomplete. Change bids at keyword level, in increments of 10 to 20%, and read the result before changing again. A bid is a price you are willing to pay for a click on a specific term for a specific job; it should be set from that term's data.
The target bid formula
Every bid decision starts from the same calculation, whether you do it in a spreadsheet or a tool does it for you.
Target bid = target ACoS × average order value × conversion rate
A product selling at $40 with a target ACoS of 25% and a keyword converting at 12% supports a bid of $40 × 0.25 × 0.12 = $1.20. If the current bid is $1.80 and the keyword's ACoS is 38%, the bid is above what the term can pay back, and the formula tells you by how much.
Three things about this formula matter in practice.
The conversion rate must be the keyword's own, over a long enough window to trust. Ten clicks tell you almost nothing. Fifty clicks tell you something. Two hundred clicks tell you the conversion rate.
The target ACoS depends on the keyword's job. A rank-building term for a launch can run at break-even or above on purpose. A profit term on a mature product runs well below break-even. One account-wide target ACoS is the most common reason bids are wrong. How we assign a job to each keyword is described in B16 · How we set bids.
The formula gives a ceiling, not a bid. Amazon's auction is second-price with adjustments, so the actual cost per click is often lower than the bid. Start at the formula and adjust on the observed CPC.
How often
| Cadence | What it catches | What it misses or breaks |
|---|---|---|
| Daily | Budget exhaustion, sudden CPC spikes | Reacts to noise; attribution for the last 2–3 days is incomplete, so most daily changes are made on wrong numbers |
| Weekly | Keywords drifting outside their ACoS band, terms ready to scale, terms to cut | Very little, for most accounts |
| Monthly | Long-term trends | A keyword can overspend for four weeks before anyone acts |
Weekly is the cadence we run. Daily checks are for budgets, out-of-stock alerts and anything that has clearly broken; they are not for bids. Amazon's sales attribution can update for several days after a click, so a bid changed on Tuesday for Monday's numbers is usually changed on incomplete data.
Within the weekly review, use a rolling window that fits the keyword's traffic. High-volume terms can be judged on 14 days. Low-volume terms need 30 to 60 days before their ACoS means anything.
How much
Bid changes of 10 to 20% are large enough to move placement and small enough to read. A 40% cut can drop a keyword off page one entirely, at which point its impressions collapse and you have no data on whether a smaller cut would have held the placement at a better cost. A 5% change is often absorbed by normal CPC variation and teaches you nothing.
The direction depends on the keyword's ACoS relative to its target and on whether it has volume to give.
Read this flowchart as text
- Weekly bid review, keyword with 30+ clicks
- ACoS vs target for its job?
- Converting at all?
- Negative or pause, not a bid change
- Lower bid 10–20%. Check placement report first
- Hold. Note impression share
- Losing impressions or budget-capped?
- Raise bid 10–20% or raise budget
- Re-read after 14 days
The placement check before lowering a bid is important. If a keyword's high ACoS comes almost entirely from top-of-search while its rest-of-search placements are fine, the fix is the placement modifier, not the base bid. Cutting the base bid punishes the placements that were working. That interaction is covered in Amazon Placement Modifiers: When Top of Search Is Worth It.
Dynamic bidding and what it does to your numbers
Amazon's campaign bidding strategies change your bid in real time. Dynamic bids, down only, lowers bids when a click is judged less likely to convert; down and up raises them by up to 100% for top-of-search and 50% elsewhere when a conversion is judged more likely. Fixed bids do neither.
Down only is the safe default for control campaigns. Down and up is worth testing on proven keywords with strong conversion rates, where paying more for a likely sale is fine, but it makes the observed CPC harder to interpret and can double what you thought you were bidding. Fixed bids suit defense campaigns where you want a predictable position on your own brand terms.
Whichever you choose, keep it consistent within a campaign so that bid changes mean the same thing across its keywords.
When to break the weekly rule
Some situations justify a bid change outside the cycle.
A stock-out or a listing suppression: bids and budgets come down the same day, because every click is wasted until the listing is back.
A price change: conversion rate will move, so the target bid moves with it. Recalculate as soon as the new price is live.
A competitor event: if a major competitor goes out of stock, a temporary bid increase on shared terms can capture their demand. Set a review date so the increase does not become permanent.
Peak periods: in the run-up to Prime Day or Black Friday, CPCs rise and conversion rates change hour to hour. The weekly cycle is shortened to daily during the event itself, with pre-agreed ranges rather than open-ended changes.
Frequently asked questions
How often should I change my Amazon PPC bids?
Once a week for most keywords, on at least two weeks of data. Check budgets daily, but leave bids alone unless a keyword has clearly broken or an event such as a stock-out or price change demands it.
How much should I raise or lower a bid at a time?
Ten to twenty percent. Smaller changes are lost in normal CPC variation. Larger changes can drop or leap a placement entirely, and you lose the information about whether a moderate change would have been enough.
Should I use Amazon's suggested bid?
Treat it as a reference for the range other advertisers bid in, not as a target. It does not know your margin, your conversion rate or the job of the keyword. Calculate your own target bid and compare; if the suggested bid is far above what your keyword can pay back, the term may not be one you can afford to rank on.
Is it better to lower bids or add negatives when ACoS is high?
If the term converts, lower the bid. If the term has spent past the break-even cost of one sale with no orders, negate it. Negating a converting term loses its sales; lowering the bid on a non-converting term just wastes money more slowly.