Home · Blog

A Weekly Book Ads Review Ritual

Thirty focused minutes a week is enough to keep Amazon book ads healthy. A step-by-step routine for KDP authors: settled data, spend check, search terms, bids, and a change log.

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

Most authors look at their ads in one of two ways: constantly, refreshing the dashboard several times a day, or almost never, until a credit card statement forces the issue. Neither works well. Constant checking leads to reacting to noise. Neglect lets waste pile up. A fixed weekly review, the same time and the same steps, is calmer and more effective than either.

Why weekly

Book ads produce small numbers. A day is too short to see anything but randomness. A month is long enough for waste to grow. A week strikes the balance: enough new data to act on, and frequent enough that no problem runs for long.

Pick a time when you are not tired or rushed. Many authors choose a weekday morning before writing. Block thirty minutes on the calendar.

Before you start: use settled data

Amazon keeps updating attributed sales and Kindle Unlimited page reads for days after a click. If you include the last few days, recent results will always look worse than they really are. For every step below, use a window that ends three or four days ago, such as the last 30 days ending on Wednesday when you review on Sunday.

The routine

Step 1: Spend and royalties (5 minutes)

Write down four numbers for each book you advertise:

• Ad spend for the last 7 settled days
• Royalties for the same days from the KDP Reports dashboard, including Kindle Unlimited
• Ad-attributed orders and pages read from the Amazon Ads console
• Your running royalty TACoS: ad spend divided by total royalties

You are looking for direction, not perfection. Is royalty TACoS stable, rising, or falling? Is total income moving with spend or not? See what ACoS and TACoS actually measure for why this matters more than any single campaign's ACoS.

Step 2: Anything broken? (3 minutes)

Check each advertised book's product page quickly:

• Is every format available and correctly linked?
• Is the price what you expect?
• Has anything changed in reviews that needs attention?

If something is wrong, pause that book's ads until it is fixed. See when to pause a book ads campaign.

Step 3: Search terms (10 minutes)

Open the search term report for the last 30 settled days and sort by spend. Work through the top twenty or so:

• Add clear wrong-intent searches as negatives.
• Move converting searches from auto or broad campaigns into exact match, with a matching negative exact in the source campaign.
• Note ASINs that convert, for product targeting.

The full method is in how to read an Amazon Ads search term report.

Step 4: Bids (7 minutes)

Look at keywords and product targets with enough clicks to judge, perhaps ten or more in the window:

• Well below break-even ACoS after counting page reads: raise the bid about 10%.
• Well above break-even: lower the bid 10–20%.
• No sales and no page reads after a fair number of clicks: lower sharply or pause the target.
• Everything else: leave alone.

Resist the urge to touch targets without enough data. Most targets should be left alone in a typical week.

Step 5: Log what you changed (5 minutes)

Keep a simple log with one line per change: date, book, campaign, target, what changed, and why. It can be a spreadsheet or a notebook. The log is what turns a weekly habit into learning. When results improve or fall a few weeks later, you can look back and see what you did.

Once a month, add a longer look

On the first review of each month, add fifteen minutes for bigger questions:

• Compare organic and ad-attributed sales over the last eight to twelve weeks. Are ads building momentum or just adding sales on top? See organic rank and ads feedback loops.
• Check cost per click by campaign against last month. Rising costs with steady conversion usually mean seasonal competition.
• Review your placement report if you use placement bid adjustments.
• Look at the calendar for the coming month and plan budget shifts. See seasonality for fiction and nonfiction ads.

What not to do between reviews

• Do not refresh the dashboard every few hours. The numbers will not be settled.
• Do not change bids because of one day's results.
• Do not launch new campaigns on impulse. Write the idea down and consider it at the next review.

The exception is a real problem, such as a broken listing or a day that spends far more than you can afford. That is why a daily spending limit you trust is useful: it lets you ignore the dashboard between reviews without worrying.

A sample review in practice

A thriller author with two books and four campaigns sits down on Sunday morning. Spend for the settled week is $84 and total royalties are $260, so royalty TACoS is about 32%, a little better than last week. Both product pages look normal. In the search term report, two searches with no sales and clearly wrong intent get negative phrases, and one converting search moves into the exact campaign. Three keywords have enough data: one is raised 10% and two are lowered 15%. Five lines go in the log. Total time: twenty-six minutes.

Nothing dramatic happened, and that is the point. Over months, small, well-judged changes add up.

Where BookAdsAI fits

BookAdsAI handles the two parts of this routine that benefit from automation and fixed rules. Its Autopilot checks Amazon's reported spend every few hours and pauses a book's allowlisted campaigns when the day reaches the cap you set, then turns those same campaigns back on the next morning, Pacific time. Once a day, it can move allowlisted keyword bids by at most 10% based on reported ACoS, only after at least five clicks, and it writes every change to a log with the before and after values. Search terms, negatives, listing checks, and the monthly look remain yours. See pricing or create an account to try it.

The short version

Same time each week. Settled data only. Check spend and royalties, check listings, work the search term report, adjust bids that have enough data, and log every change. Thirty minutes, done.