Amazon PPC Bid Strategy: Bid by Placement, Not by Panic

Bids are where sellers do the most damage the fastest. One bad week and someone slashes everything 30%. One guru video and someone hands Amazon a blank cheque with "up and down" bidding. Both are the same mistake: bidding by feeling instead of by data.

Here's the bid strategy I run across 500+ live campaigns in accounts managing $300–400K or more in annual ad spend.

Default: dynamic bids, down only

Amazon offers three bid strategies. I default to dynamic bids — down only: Amazon can lower your bid in auctions it predicts won't convert, but can never raise it.

"Up and down" sounds smarter — Amazon boosts you when conversion looks likely. In practice it lets Amazon raise your bid up to 100% at top of search, and Amazon's optimism is funded by your card. Until you have placement data proving top of search converts for that campaign, don't hand over that permission.

The one case where "up and down" earns its keep: a mature campaign with at least 60 days of clean placement data showing top of search is your most profitable slot by a clear margin. In that situation, letting Amazon compete up there automatically removes some manual work. But that's a narrow window — most accounts aren't there, and most sellers who switch to "up and down" do it before the data supports it.

Opening bids: come in strong, then settle

Timid opening bids are a slow leak. Bid too low on a new exact keyword and you get scraps of impressions on poor placements, gather no usable data for weeks, and conclude the keyword "doesn't work."

For proven terms entering an exact campaign, I open at roughly double Amazon's suggested bid. You win real placements, you get clean data in days instead of weeks, and then you settle the bid down to what conversion actually supports.

The formula for where to settle: max bid = target ACoS × price × conversion rate.

For a product priced at $40, a 25% target ACoS, and a 10% conversion rate: max bid = 0.25 × $40 × 0.10 = $1.00. That's your ceiling at steady state. I target settling to roughly 70% of break-even — leaving headroom to optimise rather than running permanently at the margin. On a break-even ACoS of 33%, that means targeting around 23% ACoS in practice.

The extra dollars spent at launch buy speed and certainty. That's usually the cheapest data you'll ever purchase.

Placement multipliers: read the report first

The placement report splits every campaign into top of search, product pages, and rest of search. The same keyword often converts at triple the rate at top of search versus product pages — which means a flat bid strategy quietly overpays for bad placements while underpaying for good ones.

Here's what that looks like with real numbers from a two-week campaign window (illustrative, but representative of what I see across accounts):

Keyword: "stainless water bottle" — base bid $1.30

Top of search: profitable at 23% against a 25% target. Product pages: bleeding at 65%. Rest of search: borderline but acceptable.

The wrong move is pausing the keyword. The right move is placement surgery.

Lower the base bid to $0.90. Set a top of search multiplier at +44% — giving an effective bid of $1.30 at top of search, unchanged. Product pages now bid at $0.90 instead of $1.30, cutting spend on the bleeding placement without touching the profitable slot. You're redirecting spend, not adding it.

If product pages continue to bleed even at a lower base bid, negative ASIN targeting in a separate sponsored products campaign can suppress that placement further.

The process:

1. Run flat (no multipliers) until you have real click volume per placement.
2. Read the placement report. Let the data tell you which slot is profitable.
3. If top of search converts well, add a multiplier there and lower the base bid proportionally.
4. If product pages bleed, lower the base bid to cut their spend — don't just add a top of search multiplier while leaving the base untouched.

Half the keywords sellers pause could be saved with placement adjustments alone.

The 10-click rule

No bid judgment on fewer than 10 clicks. Two clicks and no sale means nothing. A conversion rate of 10% means one order per ten clicks on average — judging on five clicks flips a coin and calls it analysis.

This cuts both ways: don't kill keywords early, and don't scale a keyword because it went 2-for-2 either.

Bid adjustment sizing

When you move a bid, move in steps of 15–20%, not 5% and not 50%.

A 5% cut won't move you out of a placement tier. It's noise. A 50% cut is a panic slash — you lose position, lose data continuity, and have no idea whether the results that follow are caused by the bid change or by seasonality or a competitor repricing.

Make the change. Wait seven days before judging. Run weekly, not daily.

The same rule applies to increases. If a keyword is running below target ACoS, raise the bid 15–20% and check again in a week. Stacking rapid increases is how you accidentally bid $4 on a $1 keyword.

The weekly cycle

My cadence on every account, every week:

• Pull the search term report and the placement report together — not in isolation
• Search terms with 10+ clicks, zero orders → negative exact (how negatives work)
• Keywords above break-even ACoS with real data → check placement split first; if product pages are the culprit, fix placement before touching the base bid
• Keywords below target ACoS → raise bids 15–20%, check budget caps aren't throttling them mid-day
• New converting search terms in auto/broad → harvest to exact (the structure that makes this work)
• Placement multipliers → update against current report data; multipliers set three months ago don't reflect what the account is doing now

Boring, mechanical, compounding. That's the job.

If your bids have been set by panic — or by whoever managed the account two years ago — the free audit will show you what they should be, keyword by keyword, in a short video.