How to Set Amazon PPC Bids
To set Amazon PPC bids, start from what one click is worth: target ACoS times conversion rate times average order value. Then let search term data decide whether each bid rises, falls or holds.
Target ACoS is the highest advertising cost of sales (ad spend divided by the sales those ads bring) that a product can carry and still meet its margin goal. It is set product by product, from contribution margin and the job the campaign has to do.
Start with the economics, not the suggested bid
Amazon's suggested bid is useful context, but it is not your operating limit. It reflects auction behavior around a keyword or target, not your margin, product maturity, cash position, or launch objective. A seller who lets suggested bids set the account can end up spending at the speed of the marketplace instead of the speed of the business.
A practical bid model starts with four inputs: expected average order value, conversion rate, target ACoS range for that product, and the role of the campaign. A launch campaign can tolerate different economics than a mature exact-match campaign. A defensive branded campaign should not be judged the same way as a broad discovery campaign. The bid is a hypothesis about what one click can be worth under those conditions.
The common starting formula is:
starting bid = target ACoS x conversion rate x average order value
This is not a universal rule. It is a first reference point. With an average order value and a conversion rate read from the product's own data, and an ACoS limit set for that product, the formula gives the team a grounded starting point. The real decision still depends on stock, margin, keyword intent, and whether the campaign is meant to harvest demand or discover it.
The mistake is treating the formula as a machine answer. Amazon PPC does not behave like a spreadsheet because the inputs move. Conversion rate changes by keyword, placement, review profile, price, competitor activity, season, Prime eligibility, and inventory pressure. Average order value may look stable at account level while a specific ASIN behaves very differently. Target ACoS is not a moral number; it is a business constraint shaped by contribution margin, launch strategy, and how much the brand is willing to spend to earn rank, data, or market share.
A useful bid is therefore not the highest bid you can justify. It is the bid that lets you buy enough evidence without letting the auction take control of the account. The first bid should create a measurable test. The next bid should respond to what the test proves.
Separate campaign roles before comparing bids
Two campaigns can carry the same keyword and still need different bid logic. A discovery campaign, an exact-match winner campaign, a branded defense campaign, and a product targeting campaign are not doing the same job. If they are managed with one bid rule, the account loses precision.
Discovery campaigns exist to find search terms and product targets. Their job is to generate controlled evidence, so they will rarely look perfect in the first week. Bids should be high enough to enter relevant auctions, but conservative enough that a weak term does not drain budget before it has produced a meaningful signal.
Exact-match winner campaigns exist to scale terms that already proved they can convert. These can justify stronger bids because the uncertainty is lower. The team knows the search term, the product, the conversion behavior, and the economics. Bid changes here should be deliberate, because volume often becomes more expensive as the campaign climbs into more competitive auction positions.
Branded campaigns protect demand that already knows the brand. They often show strong ACoS, but that does not mean they deserve unlimited budget or aggressive bidding. A branded campaign can make the ad account look efficient while hiding dependency on demand that may have converted organically anyway.
Product targeting campaigns need separate judgment because the click context is different. A shopper on a competitor detail page is not identical to a shopper typing a high-intent keyword. Placement, price comparison, reviews, and product image strength affect whether a higher bid is worth testing.
Before changing bids, ask what the campaign is supposed to prove. If the role is unclear, the bid change will usually be a cosmetic edit rather than an operational decision.
Read search term data before you move bids aggressively
Most bid mistakes come from reacting too early. A keyword with two clicks and no orders is not a failed keyword. A target with one order from three clicks is not automatically scalable. Bids should move when the account has enough evidence to separate a weak signal from normal variance.
Use search term reports to separate four situations:
| Situation | What it usually means | Practical action |
|---|---|---|
| Clicks with no orders | Possible waste or insufficient data | Wait for meaningful click volume, then step the bid down toward the minimum |
| Orders at acceptable economics | Target is working inside its role | Hold or raise gradually if budget allows |
| Many clicks, few orders (high click-through rate, low conversion rate) | The ad attracts interest but the listing may not convert | Check price, image, reviews, and offer before only changing bids |
| Low click-through rate, low spend | The bid may be too low or the term may be weak | Test a controlled increase if intent is relevant |
The bid is only one lever. If conversion rate is weak because the product page is uncompetitive, raising bids just buys more inefficient traffic. If the campaign structure is messy, bid changes may hide a deeper targeting problem. For the metrics and campaign structure around this, see Amazon PPC advertising. For the listing and account work that decides conversion, see Amazon operations.
Search term data also protects the account from averages. Campaign-level ACoS can look acceptable while one query is wasting budget and another is carrying performance. Ad group data can hide a single product target that should be isolated. Account-level TACoS (ad spend divided by total sales, not only the sales from ads) can improve while a new product launch is underfed. Bid decisions need the lowest useful layer of evidence, not the dashboard number that feels easiest to read.
A practical review sequence is simple: identify spend, identify orders, identify intent, then decide whether the problem is bid, targeting, listing, or structure. If the query is relevant and conversion is weak, inspect the offer before cutting the bid. If the query is weak or loosely related, step the bid down toward the minimum rather than pausing or negating it; a term with no link to the product goes straight to the minimum and gets flagged. If the query is relevant and converting, decide whether the business wants more volume at potentially worse marginal efficiency.
Bid up, bid down, or leave it alone
Bid up when a target has enough clicks, clear purchase intent, and orders inside the product's economic role. Increase in steps rather than jumps so the account can show whether volume scales or efficiency breaks. A keyword can look profitable at low volume and become expensive when it enters a more competitive auction layer.
The best reason to bid up is not simply that ACoS is low. Low ACoS can mean the bid is conservative, but it can also mean the keyword is branded, volume is small, or attribution is flattering the campaign. Bid up when there is a credible reason to believe more spend can create more useful sales, more ranking pressure, or more strategic coverage.
Bid down when spend is accumulating against weak or irrelevant intent. Whether the target has converted is only part of the check: look at whether the search term belongs in that campaign and whether the product has a realistic chance of winning the click. If the term has nothing to do with the product, step its bid to the minimum and flag it; the listing's keywords usually explain why it matched.
A bid reduction should preserve learning where possible. Pausing a target makes the next read impossible. If the target is relevant but inefficient, step down. If it keeps losing money, keep stepping it toward the minimum bid: it stays live as a cheap test and can recover when the listing, price or season changes. If the target is relevant but the listing cannot convert, send the issue to listing, pricing, creative, or offer work instead of pretending the bid can solve everything.
Leave the bid alone when the data is thin. This is harder than it sounds. Many accounts are damaged by daily micro-edits that reset learning, fragment evidence, and turn the campaign into a reaction log. Weekly or twice-weekly review is usually enough for stable campaigns; launch campaigns may need closer observation, but not constant panic changes.
Leaving a bid alone is an active decision when the account needs more evidence. It tells the system: keep collecting data under the current hypothesis. Without that discipline, the account never holds a test long enough to learn from it.
Suggested bids are market signals, not instructions
Amazon suggested bids can help identify whether your planned bid is wildly outside the auction range. They are useful when entering a new keyword cluster, estimating competitive pressure, or checking if a product is priced into a difficult category. But suggested bids do not know your contribution margin, inventory age, refund rate, or whether a keyword is strategically necessary.
Treat suggested bids as a reference band. If your calculated starting bid is far below the range, either the auction is too expensive for your economics or your assumptions about conversion are too conservative. If your calculated bid is far above the range, you may be overpaying or entering a keyword where volume can be bought more efficiently.
Suggested bids are especially dangerous when the seller is already anxious. A low-impression campaign makes the operator feel invisible, so the suggested bid becomes a shortcut to action. But visibility without economics is not control. What matters is whether entering that auction at that price serves the product's current job.
The best PPC operators do not worship the formula or the suggested bid. They use both as constraints, then let search term data, catalog maturity, and margin decide the next move.
A simple operating rhythm for bid management
A stable account needs rhythm more than drama. Daily emergency edits make sense when something is clearly broken: budget leak, irrelevant query, stock issue, campaign launch, or a sudden change in conversion. Normal bid management should run on a cadence that allows evidence to accumulate.
A weekly operating rhythm can look like this:
- Budget leaks first: campaigns or targets spending without useful intent.
- Search terms reviewed by spend and orders, not by curiosity.
- Irrelevant terms separated from relevant but inefficient ones.
- Bids raised only where more volume is strategically useful.
- Bids lowered toward the minimum where a target costs more than it returns.
- Terms unrelated to the product stepped to the minimum and flagged.
- Listing or offer problems noted separately, so PPC is not blamed for everything.
This rhythm keeps the account readable. It also makes later automation safer. AI tools and bid rules work better when the campaign structure, targets, and decision logic are already clean. Automation magnifies whatever process sits underneath it; it does not replace it.
We do not run PPC on a monthly retainer. When an account needs this review done by someone outside the team, campaign upkeep can be drawn task by task from a prepaid bag of hours. A question about bid strategy or target ACoS fits an hour of consulting. A product moving into a new marketplace needs its bids rebuilt from that market's own data. Opening a marketplace has a closed price per product, and Selling in Europe covers entry into the European marketplaces; the campaigns there are not part of that price and are set up and looked after from the bag of hours. To describe your account, contact us.
FAQ
What is the safest way to set Amazon PPC bids for a new keyword?
Use a conservative starting bid based on target economics, conversion assumptions, and average order value. Then wait for enough click and conversion data before making large changes.
Should I follow Amazon's suggested bid?
Not automatically. Suggested bids show auction context, but they do not account for your margin, product maturity, inventory position, or campaign role.
How often should Amazon PPC bids be adjusted?
Stable campaigns usually need a weekly or twice-weekly review. Launch campaigns may need closer monitoring, but daily bid changes without enough data often create noise instead of control.
Can one bid formula work for every product?
No. The formula is a starting framework. Category competition, conversion rate, price, margin, and business stage change what a practical bid can be.