Home · Blog

Bid Strategies Without Magical Thinking

A bid is the most you will pay for one click, not a signal that the algorithm likes your book. How to set a maximum bid from your royalty and conversion rate, what Amazon's bidding strategies do, and how to change bids without chasing noise.

The BookAdsAI team · Bidding · Published October 5, 2026.

A lot of advice about Amazon bids sounds like folklore. Bid high to "train the algorithm." Never bid below the suggested bid. Raise bids on Fridays. Most of it has no basis anyone can show you. A bid is a simpler thing than that: it is the most you are willing to pay for one click on one target. Everything useful about bidding follows from that definition.

What a bid actually does

Sponsored Products runs an auction each time a shopper searches or opens a product page. Your bid, combined with Amazon's estimate of how relevant your ad is to that shopper, decides whether you win a placement and roughly where. You usually pay less than your maximum, because the price is set by the competition you had to beat.

Raising a bid can win you more impressions and higher placements. It cannot make a shopper buy a book whose cover, title, price, and reviews do not appeal to them. If a target does not convert at $0.40 a click, it will not usually start converting at $0.80. It will just lose money faster.

Work out your maximum bid from your own numbers

The fairest maximum bid comes from two numbers you can measure: what you earn per sale and how many clicks a sale takes.

Break-even bid = royalty per sale × conversion rate

Suppose your ebook earns $3.40 per sale and, across the last few months, about 1 in 12 ad clicks has produced a sale or the equivalent in Kindle Unlimited pages. Your conversion rate is about 8.3%.

• Break-even bid: $3.40 × 0.083 ≈ $0.28 per click.
• If you want ads to keep half the royalty as profit, your target bid is about $0.14.
• If you have a series and know that a share of Book 1 buyers continue, you can include that later revenue and justify a higher bid. See read-through math for series authors.

This calculation is often humbling, because suggested bids in popular genres can be well above it. That does not mean you cannot advertise. It means you need either a better conversion rate, more income per reader through a series, or targets with less competition.

Amazon's bidding strategies

Each Sponsored Products campaign has a bidding strategy that lets Amazon adjust your bid in real time:

• Dynamic bids, down only: Amazon lowers your bid when it predicts a click is less likely to convert. It never goes above your bid. This is the safest choice and a sensible default for books.
• Dynamic bids, up and down: Amazon can raise your bid when it predicts a sale is more likely, by up to 100% for top-of-search placements and up to 50% elsewhere, and lower it when a sale is less likely. Your real cost per click can end up well above the number you typed in.
• Fixed bids: Amazon uses your exact bid every time. Useful for clean tests, less forgiving on poor placements.

You can also add placement adjustments that raise bids for top of search, rest of search, or product pages. These multiply on top of the strategy. A 50% top-of-search adjustment combined with up-and-down bidding can double or triple what you pay for those clicks. Use adjustments only when a placement has clearly earned it in your placement report.

How to change bids without chasing noise

Book ads produce small numbers, so it is easy to react to randomness. A few rules help:

1. Wait for clicks, not days. Decide on a minimum number of clicks before you judge a target. Somewhere between 10 and 20 is reasonable for most books. Five clicks with no sale tells you very little.
2. Move in small steps. Change a bid by 10–20% at a time, then give it a week or two. Large swings make it impossible to tell what caused what.
3. Judge on settled data. Leave out the most recent few days, because attributed sales and page reads arrive late.
4. Change one thing at a time. If you change bids, the cover, and the price in the same week, you will not know which one worked.
5. Write it down. A simple log of date, target, old bid, new bid, and why you changed it saves you from undoing your own good decisions.

Patterns that waste money

• Bidding to the top of the suggested range by default. The suggestion reflects what others pay, not what your book can afford.
• Raising bids on a target with no sales to "give it a chance." If it has had its fair share of clicks, lower or pause it instead.
• Pushing up-and-down bidding with large placement adjustments on a new book. Costs can climb quickly before you have any conversion data.
• Ignoring Kindle Unlimited. For a KU book, a target with no orders but solid page reads may be profitable. Count pages before you cut a bid.

A calm weekly routine for bids

Once a week, sort targets by spend over the last 30 days, excluding the most recent three days:

• Targets with enough clicks and an ACoS well below break-even: raise the bid by 10%.
• Targets with enough clicks and an ACoS well above break-even, after counting pages read: lower the bid by 10–20%.
• Targets with enough clicks and no sales or pages at all: lower sharply or pause.
• Everything else: leave it alone.

That is the whole method. It is not exciting, and it works better than any trick.

How BookAdsAI applies the same discipline

BookAdsAI's Autopilot follows rules like these on campaigns you allowlist. It acts only after a target has at least five reported clicks and some spend. It lowers a bid by up to 10% when that target's ACoS is above 40%, and raises it by up to 10% when ACoS is below 15%. It moves bids at most once a day, keeps them between $0.20 and a $2.50 ceiling (or the current bid, if that is already higher), and writes each change to a log. It uses only numbers Amazon reported, and it does not change budgets or bidding strategies. The details are on the Spend-Cap Autopilot Method page.

The short version

Your bid is a price, not a message. Set a maximum from royalty times conversion rate, prefer down-only dynamic bidding while you learn, change bids in small steps after enough clicks, and keep a log.