The Amazon Ads System I Use to Optimize My KDP Books in 30 Minutes a Week

Most KDP authors are either ignoring their ad data or overwhelmed by it. Here’s the rule-based system that fixed both for me.

I built this system because I had no choice.

I have two kids under five, a full-time job, and a KDP catalog I’m growing on the side. I don’t have hours to spend in Amazon’s ad console every week. I don’t have the bandwidth for trial and error. When something costs me money without a clear reason why, I need to fix it fast and get back to whatever chaos is happening in my house.

So when I spent four months losing money on ads, $683 over Q1, with books that were genuinely selling, genuinely ranking, genuinely getting reviews, I got systematic about it. Not because I love spreadsheets. Because I needed this to work in the margins of a very full life.

The problem wasn’t the bids. It wasn’t the keywords. It was that I had no system, just instinct, occasional panic bid changes, and a vague hope that the algorithm would sort itself out.

So I built one. A rule-based optimization framework I run every week in about 30 minutes, usually after bedtime. It reviews every target, applies the same logic every time, and outputs a clean action list grouped and ready to paste directly into Amazon’s tools.

Why Most KDP Ad Accounts Bleed Money

Two mistakes I made myself for months, and see constantly in other accounts:

Mistake 1: Paying for clicks you’d get for free.

When you’re Rank 1 organically on a search term, shoppers would find you without an ad. But if your ad is also running on that term, some of those shoppers click the ad instead and you pay for a click you would have gotten anyway. I found $53/week in spend on terms where I was already Rank 1, spending on clicks I didn’t need to buy. The spend was going out. The incremental benefit wasn’t there.

Mistake 2: Incomplete campaign structure.

Auto campaigns find search terms. Phrase campaigns capture mid-intent searches. Exact campaigns lock in proven converters. Product targeting hits competitor listings. Most authors run one or two of these and wonder why their reach is limited. Each campaign type has a specific job, and when they’re missing or overlapping, you’re either leaving reach on the table or paying for the same thing twice.

The Framework: Five Decision Categories

Every week I pull four reports from Amazon Ads: Targeting, Search Term, Campaign Performance, and Impression Share. Then I run every target through the same decision tree.

Every target lands in one of five buckets:

Dead targets, two types. The first has had real exposure but nobody clicked. The second spent money without producing a single order. Both get paused, but the thresholds are different. Treating them the same way leads to cutting things too early or holding on too long.

Gold mines. Low click volume, high conversion rate, healthy ACOS. These look unimpressive in a dashboard and get paused by mistake all the time. My rule: once something lands here it stays running indefinitely, regardless of impression count. These are the targets that compound quietly over time.

Graduates. When a search term in an Auto or Phrase campaign has proven itself with real orders at an acceptable ACOS, it moves to an Exact campaign at a higher bid. It also gets negated in the source campaign so you stop paying for the same term in two places.

Bid up, bid down, monitor. Three separate buckets based on ACOS range and click volume. Each gets a consistent adjustment applied the same way every single week. Consistency here matters a lot because changing bids too often makes it impossible to read what’s actually happening in your account.

Leave it alone. Most targets, most weeks, need nothing. Knowing what to skip is half the work. This one took me a while to trust.

The Cannibalization Check

This runs after the main framework and is the part most people skip. It’s also where I found most of my waste.

I cross-reference my Impression Share report against my active targets every week looking for three things:

Self-bidding across campaigns: Two of my own campaigns competing on the same search term. Whoever wins the auction, I pay, but I’ve also bid against myself to raise the cost of that click. One campaign needs a negative exact added.

Dominant position waste: Rank 1 with strong impression share, ad spend going out when it could sell for free. The fix is a controlled bid pullback and watching whether organic holds.

Portfolio overlap: Two different titles in my catalog both bidding on the same keyword. Pick the right book to own each term and negate in the other.

In a single week this check found $53 in likely wasted spend, around $212/month. At the margin most indie KDP publishers operate at, that’s the difference between a losing month and a breakeven one.

The Output Format

I produce one spreadsheet, grouped by the decision categories above, with the data copy-paste ready, organized by campaign and bid amount. The goal is to open it, execute, and be done. No re-reading, no re-interpreting. I have a toddler. I need it to be that simple.

What Running This Weekly Actually Looks Like

After several months on this system: TACOS dropped from around 56% weekly to 34.6%. The cannibalization check alone recovered $212/month in spend that was going out with nothing coming back. Q1 ended trending positive after three straight losing months.

Amazon ads are volatile, especially at low KDP volumes, so any given week can look bad regardless of what you’re doing right. The value of a consistent system isn’t that it fixes every week. It’s that it keeps you from making reactive decisions on noisy data, which is very easy to do at 9pm when you’re tired and your ACOS looks terrible.

Want Me to Run This on Your Account?

I’m offering a small number of one-time optimizations for KDP authors with active ad spend. You send me your reports, I execute the full framework and deliver your action list. Flat fee, no retainer, no ongoing commitment.

If you’re spending on ads and not sure what’s working, drop a comment or email me directly at lilabloomkids@gmail.com with “KDP Ads” in the subject line.

KDP publisher, full-time employee, mom of two under five. Building this thing one nap time at a time.