Amazon Ads for books, without the jargon

August 10, 2026 · 3 min read

Amazon Ads are the most direct lever a self-published author has, because they reach people who are already on Amazon looking for a book. Not people scrolling a feed who might one day want a book — people in the act of buying one.

They are also where a lot of author money quietly disappears, almost always for the same reason: the campaign was judged too early, or judged on the wrong number.

The ad types

Sponsored Products

Your book appears in search results and on other books' product pages. This is where nearly every author should start — it is the simplest, the cheapest to test, and the closest to the point of purchase.

Sponsored Brands

A headline banner above search results, showing several titles at once. It needs a catalogue to be worth running; with one book there is nothing to showcase.

Lockscreen ads

Shown on Kindle devices when idle. Interruptive rather than intent-driven, and generally the weakest of the three for a new author.

Targeting: books, not words

The instinct is to bid on broad phrases — "thriller", "self help book". Resist it. Those terms are expensive, enormously competitive, and reach people whose intent is far too vague to convert.

What works is targeting specific comparable titles and authors: books a reader who would like yours has already bought. Your ad appears on those product pages, in front of someone demonstrably in your market.

1. List fifteen to thirty books genuinely similar to yours — tone, length, audience, not just genre label.
2. Be honest. Targeting a runaway bestseller wastes budget on readers who wanted that book specifically.
3. Include mid-list authors. Their pages are cheaper and their readers are actively looking for the next thing.
4. Target the authors as well as the individual titles.

Reading the numbers

ACOS — advertising cost of sale — is ad spend divided by the revenue those ads produced. It is the headline metric and it is routinely misread in two ways.

First, it counts royalty revenue, not cover price. A 100% ACOS does not mean breaking even; it means losing money, because you only ever received a fraction of the price.

Second, it only counts the advertised book. If someone buys book one from your ad and then reads the next four, the ads report attributes none of that. On a series, judging book one's ACOS in isolation will tell you to switch off a campaign that is profitable.

Running a first campaign

1. Start with one Sponsored Products campaign using manual targeting on comparable titles.
2. Set a daily budget you would not mind losing entirely. The first weeks buy data, not sales.
3. Bid low initially and raise on the targets that produce, rather than starting high and cutting.
4. Wait. Sales attribution lags by days, so a campaign that looks dead on day three often is not.
5. Give it at least two weeks and a few thousand impressions before judging any individual target.
6. Then prune what has spent without converting, and increase bids on what has.

Why campaigns fail

Usually not the ads. If the cover does not signal the genre or the description does not convert, advertising simply buys traffic to a page that was always going to lose it — faster, and with an invoice attached.

Before increasing spend on a campaign that is not working, check the click-through rate and the conversion rate separately. Low clicks is a cover and title problem. Good clicks with low sales is a description and price problem. Neither is fixed by bidding more.