Yes — for some books, on some keywords, at the right bids. The honest answer comes from what a catalogue’s clicks actually do. Here is what 254,473 clicks and 32,569 orders on one publisher’s KDP books show, and how to find out for yours without wasting money.
What a real catalogue’s ads did
| Across the account, April 2025 to September 2026 | |
|---|---|
| Clicks | 254,473 |
| Orders | 32,569 |
| Conversion | 12.8% of clicks |
| Average cost per click | $1.04 |
| ACOS | 52% |
| Keywords and targets | 16,326 |
| Never clicked | 5,825 (36%) |
| Ever made a sale | 3,417 (21%) |
A few keywords do almost all the work
The top 5% of keywords and targets produced 82% of all orders; the top 1% produced half. Most keywords either never got a click or never sold. That is normal, and it changes how you should think about “do ads work”: the question is not whether your average keyword works, but whether you can find the few that do, and stop paying for the rest.
When ads can pay for a book
ACOS below break-even ACOS = royalty ÷ price
- A healthy royalty. A 70% eBook or a well-priced paperback has room; a low-priced, low-royalty book often can’t pay for clicks at all. Work out yours.
- A series. Sales of book two and three aren’t in book one’s ACOS, so a series opener can run above break-even and still make money.
- Kindle Unlimited. Page reads earn money that ACOS ignores. KENP and ACOS.
- A book page that converts. Ads bring the click; the cover, price and description make the sale.
When they usually don’t
- The royalty per sale is smaller than the cost of the clicks a sale takes.
- The ads mostly show on other books’ pages: in this account those clicks converted at 5% against 14–16% from search, at 95% ACOS.
- Keywords are paused before they’ve had a fair test, so the few winners are never found.
How to run a fair test
- Know your break-even ACOS and bid before you start: break-even bid = royalty × conversion rate.
- Start with 20–30 relevant keywords plus an automatic campaign to find the searches you didn’t think of.
- Budget for the test: roughly 15 clicks per keyword. At about $1 a click, 20 keywords need about $300 to tell you anything.
- Promote the searches that sell, even after one order, and negate the ones that don’t after a fair test. our study of 2,846 keywords shows why one order is enough.
- Judge after 2–3 weeks, ignoring the last few days, whose sales are still arriving.
Common questions
Do Amazon ads work for books?
They can. Across one publisher’s 254,473 clicks they converted 12.8% of clicks into orders at a 52% ACOS — profitable or not depending on each book’s royalty. Most of the results came from a small share of keywords.
Are Amazon ads worth it for self-published authors?
When your ACOS can stay under your break-even ACOS (royalty ÷ price), or when read-through, page reads or launch rank make up the difference. For very low royalties, often not.
How much should I spend to test Amazon ads for a book?
Enough for each keyword to get about 15 clicks. At around $1 a click, 20 keywords need roughly $300 before the results mean much.
Why do most of my keywords get no sales?
Because that is normal: in one large book account, 36% of keywords were never clicked and only 21% ever sold, while the top 5% made 82% of orders. The job is finding that 5%.
- Across 254,473 book-ad clicks: 12.8% conversion, $1.04 a click, 52% ACOS.
- The top 5% of keywords made 82% of the orders; most never sold.
- Ads pay when ACOS stays under break-even (royalty ÷ price), or when series, KU or launch rank make up the gap.
- Search placements converted three times better than product pages.
- Give keywords a fair test (about 15 clicks) before judging the whole idea.