Amazon Ads keyword bidding works best as a controlled testing process rather than a one-time guess. This guide explains how to set starting bids for broad, phrase, and exact match keywords, account for placement controls, and use spend, orders, CPC, conversion rate, advertising cost of sales (ACOS), and impression data to make repeatable bid decisions.
Overview
For Sponsored Products, a keyword bid is the maximum amount you are willing to pay when an eligible shopper click is won through an auction. The amount you ultimately pay can differ from your bid, and performance depends on more than the keyword alone. Product relevance, retail price, listing quality, competition, placement, budget, and conversion rate all influence the result.
Match types provide different levels of control. Broad match is generally useful for discovery because it can reach a wider set of related shopper searches. Phrase match narrows the relationship between the keyword and the search while still allowing variation. Exact match provides the tightest control and is often used when a search term has demonstrated relevant traffic or conversions. These are operating roles, not permanent labels: a term may begin in a discovery campaign and later receive a dedicated exact-match bid.
Do not treat suggested bids as automatic answers. They can be useful as a market reference, but the appropriate bid must also reflect your economics and campaign purpose. A discovery campaign may accept a higher short-term ACOS while collecting search-term data. An efficiency campaign may require a lower bid or stricter placement controls because its job is to protect profitability.
| Target type | Primary role | Starting-bid approach | What to review |
|---|---|---|---|
| Broad match | Discover related queries and new demand | Use a controlled test bid, usually below the bid reserved for proven exact terms | Search terms, relevance, spend without orders, and emerging converting queries |
| Phrase match | Balance reach with more query control | Start between broad and exact when the term has clear relevance | Query variations, conversion rate, CPC, and overlap with exact targets |
| Exact match | Prioritize a known, relevant search term | Set from break-even economics and observed performance, then test cautiously | Orders, ACOS, conversion rate, impression share signals, and placement results |
How to estimate a starting bid
Begin with the maximum CPC your unit economics can support. A useful break-even formula is:
Break-even CPC = contribution per order × conversion rate
Contribution per order should represent the amount available for advertising after the costs you choose to include, such as product cost, Amazon fees, fulfillment, discounts, and other variable costs. Conversion rate should be expressed as a decimal. For example, a 10% conversion rate is 0.10.
Suppose a product has $12 available for advertising per order and converts at 8%. Its break-even CPC is:
$12 × 0.08 = $0.96
This is an economic ceiling, not necessarily the recommended launch bid. A new broad-match target may need to start below that amount while you confirm relevance. An exact-match target with reliable order history may justify testing closer to the ceiling, provided the resulting ACOS remains acceptable and the campaign has enough budget to learn.
You can also estimate the CPC implied by a target ACOS:
Target CPC = average order value × target ACOS × conversion rate
If the average order value is $30, the target ACOS is 25% (0.25), and conversion rate is 8% (0.08), the target CPC is $0.60. This calculation creates a planning reference. It should be checked against actual CPC because auction conditions, placement, and query mix can change the result.
Use these calculations alongside the campaign objective:
- Discovery: prioritize useful search-term data and relevant traffic, while limiting uncontrolled spend.
- Efficiency: prioritize orders at or below the acceptable ACOS or above the required ROAS.
- Product launch: use a defined test budget and avoid making large bid changes before enough impressions and clicks accumulate.
- Seasonal promotion: recalculate using the promotional price, expected conversion rate, and available margin.
Inputs and assumptions
Before changing Amazon Ads keyword bids, export or record a consistent reporting window. At minimum, capture the keyword or targeting expression, match type, campaign, ad group, placement, impressions, clicks, spend, CPC, orders, sales, and ACOS. Add conversion rate when the reporting view provides it, or calculate it as:
Conversion rate = orders ÷ clicks
ACOS = ad spend ÷ attributed sales
Use the same attribution and date settings when comparing periods. A short window can be distorted by low traffic, delayed reporting, promotions, or a small number of orders. A longer window may hide a recent listing change or seasonal shift. The right window is long enough to produce a useful sample and short enough to reflect the decision you are making.
Separate placement performance where possible. Top-of-search, other search, and product-page placements can produce different CPCs and conversion rates. If a campaign uses placement-based bid adjustments, evaluate the combined result and the placement-specific result before raising the base keyword bid. Otherwise, you may increase every placement when only one placement needs more exposure.
Keep discovery and efficiency targets easy to distinguish. Avoid creating several campaigns with the same keyword and identical settings without a clear purpose, because overlap can make search-term analysis and budget control harder. When a broad or phrase campaign discovers a valuable query, consider adding that query as an exact target in the campaign designed for efficiency. Then use negative keywords where appropriate to reduce unwanted duplication or irrelevant traffic.
Worked examples
Example 1: setting a first bid
A seller estimates $10 of contribution available for advertising per order and expects a 7% conversion rate. The break-even CPC is $0.70. The seller launches a broad-match discovery target at a deliberately controlled bid below the ceiling, then uses a phrase target at a moderate test level. An exact target is held for a search term with stronger evidence of relevance.
The purpose is not to predict the winning CPC precisely. It is to prevent an unbounded starting bid and establish a reference against which actual CPC, orders, and ACOS can be judged.
Example 2: reviewing an exact-match target
During the review period, an exact target receives 40 clicks, records 4 orders, spends $28, and generates $112 in attributed sales.
- CPC: $28 ÷ 40 = $0.70
- Conversion rate: 4 ÷ 40 = 10%
- ACOS: $28 ÷ $112 = 25%
If the seller's target ACOS is 25% and the product remains profitable at that level, the bid may be acceptable. A modest increase could be tested if impression volume is limited and the target is strategically important. If the target is receiving adequate impressions but ACOS is above goal, reduce the bid in a measured step or investigate placement performance before making a larger change.
Example 3: handling a broad-match target
A broad target generates many impressions and clicks but no orders, while the search-term report reveals that only a small portion of queries closely match the product. The correct response is not automatically to raise the bid. First review search terms, add appropriate negative keywords, improve targeting structure, and confirm the product detail page supports the query's intent. If relevant terms still fail to convert after a meaningful test, lower the bid or pause the target rather than paying more for weak traffic.
Use this compact bid-review template:
| Field | Record | Decision question |
|---|---|---|
| Match type and target | Broad, phrase, or exact | Is this target serving discovery or efficiency? |
| Impressions and clicks | Volume and traffic | Is there enough activity to judge the bid? |
| Spend and CPC | Total cost and average click cost | Is CPC within the economic reference? |
| Orders and conversion rate | Outcomes and efficiency | Is traffic turning into relevant sales? |
| Sales and ACOS | Revenue efficiency | Does performance meet the campaign goal? |
| Placement | Search or product-page location | Is one placement driving the result? |
| Action | Raise, lower, hold, isolate, or negate | What single change is most justified? |
When to recalculate and review bids
Review active campaigns on a regular schedule, but avoid reacting to every daily fluctuation. A weekly review can identify wasted spend, emerging search terms, budget constraints, and obvious bid outliers. A monthly review is useful for reallocating keywords between discovery and efficiency campaigns, checking economic assumptions, and comparing placement performance.
Recalculate sooner when the product price, margin, coupon, fulfillment cost, inventory position, listing content, or conversion rate changes. Revisit bids when a product launches, enters a seasonal promotion, experiences a catalog change, or faces a meaningful shift in demand or competition. Also review after Amazon Ads changes reporting fields, placement controls, suggested-bid behavior, or campaign terminology; interface labels and available controls can change over time.
At each review, make one clear decision per target: hold, raise, lower, isolate into another campaign, add as an exact target, or exclude with a negative keyword. Record the date, old bid, new bid, reason, and reporting window. Wait for a sufficiently informative sample before judging the change, while continuing to monitor budget and irrelevant traffic.
A practical workflow is to calculate the break-even or target CPC, compare it with actual CPC, inspect search terms and placements, then adjust the smallest number of variables necessary. This keeps Amazon PPC optimization tied to evidence rather than to suggested bids or isolated clicks.