How to Read an Amazon Ads Search Term Report for Books

Before pausing a campaign that looks unprofitable, check what is inside it. A few expensive searches can outweigh the income from useful ones. The campaign total tells you there is a problem; the search-term detail helps you decide what to change.

For Sponsored Products and Sponsored Brands, follow the cost from the campaign through the ad group and keyword or product target to the reported search term. Keep the book, clicks, cost, and attributed outcomes together as you work. Amazon’s unified reporting builder lets you assemble that view. Adigy collects search-term information automatically into an updated list with search and filters, and Negative Oracle manages exclusions continuously. You can use the report to review those decisions or make them yourself.

A losing campaign can contain searches worth keeping

Consider a fictional Sponsored Products campaign for one US paperback. Each copy earns $5 royalty after printing deductions. All units in this example are attributed to clicks for that edition, and both keywords sit in the same ad group.

Scroll horizontally on a small screen to read the table.

Keywords A and B add up to the campaign total. The last two rows add up to Keyword B. These are views of the same activity, so adding every row together would double-count it.

The campaign has lost $20. A publisher might say, “This campaign is not working,” and pause it. That would also stop Keyword A, which contributed $30 after advertising during this period.

Even the keyword totals can mislead. Keyword B lost $50 overall, yet “rose pruning basics” contributed $10 within it. The other searches cost $60 without a sale. Cutting the whole keyword would also reduce traffic the publisher wanted to keep.

Review those searches before choosing the change. A clear mismatch can be excluded after checking scope. Relevant traffic needs a closer look at royalty contribution, reporting maturity, and the testing allowance before you reduce bids, continue, or stop.

The figures describe this fictional period. They do not predict the result of an edit: traffic can change, and an exclusion can affect related searches. Sponsored Brands needs the same attention to the results inside an average, with its own targeting options and attribution scope.

Campaigns hold budgets, and keywords and product targets guide bidding. In a pay-per-click campaign, the charge arises when someone clicks. Some clicks come from product pages or other contexts rather than a typed search. Check the reported value and placement context before treating a row as the shopper’s exact words.

Build the report in unified reporting

Amazon is retiring the old Sponsored ads reports page. Its migration guidance gives two deadlines:

• December 17, 2026: new report creation and editing stop in the old report centers. Existing reports become read-only.
• December 31, 2026: the old report centers close. Saved reports, schedules, and historical data remaining there are deleted.

The dates can change. Save the exports you need and migrate useful schedules before the deadlines. Sponsored ads reporting continues in the new interface.

Open Reporting in the advertising console, then choose Create report or a suitable template. Amazon’s report-builder instructions explain the controls.

Choose one advertiser account, ad product, marketplace, and currency. Start with Sponsored Products and make a separate Sponsored Brands view where relevant. Check the targeting and attribution scope before comparing them. Select campaigns for the book and edition you want to review, and check the filters: the builder can include other accounts, countries, and ad products.

Use a period of complete days and record when you export it. Recent conversions are provisional. A later download may include more attributed outcomes for the same period.

Keep the target and search term in the same row

Dimensions determine how the report groups activity into rows. Metrics measure that activity. Start with the following dimensions, using combinations the builder allows.

Scroll horizontally on a small screen to read the table.

Keep the marketplace in the filters or a compatible country dimension. Add Advertised product ID if it works with your other selections. Otherwise, check that the selected campaigns and ad groups isolate the edition. If several books remain mixed together, you cannot apply one book’s royalty to every row.

Amazon’s dimension definitions distinguish Targeting, the keyword, product, or category you selected, from Search term. Keep both for Sponsored Products and Sponsored Brands where available. Similarly named fields may have different definitions or ad-product support.

Leave Hour, Device type, Placement, and other breakdowns out until you have a question that needs them. Each added dimension can split the evidence into smaller rows. If you later group a query that appeared through several targets or campaigns, retain the original routes so you can still see where a change would apply.

Pick metrics you can use together

Start with Impressions, Clicks, Total cost, and CPC. Add compatible metrics for purchase count, units sold, and sales value. Use the attribution scope you intend to evaluate.

Where they are available together, Purchases from clicks (promoted), Units sold from clicks (promoted), and Sales from clicks (promoted) give you click-attributed outcomes for the promoted product. Check their definitions and availability in your account before saving the report.

These labels describe different things:

• Purchases count purchase events; Units sold counts units. One purchase need not mean one copy.
• Sales is attributed sales value, which differs from your KDP royalty.
• Promoted and halo cover different product scopes. An attributed product may be a different edition from the one advertised.
• From clicks and from views cover different attribution scopes. Adding an overall total to its components would count activity twice.
• Purchase rate, Purchase rate over clicks, and Purchase from click rate have different labels. Check each numerator and denominator before calling it “conversion rate.”

Amazon defines Total cost as campaign expenditure, excluding third-party fees paid by the agency. Software, production, and other publishing costs need their own treatment in your calculation.

Some selections prevent others. Adding Converted product, for example, makes traffic metrics such as clicks and impressions unavailable. Use a separate compatible conversion view to investigate purchased editions. Do not copy its totals onto each search-term row or try to force the breakdown into the traffic report.

Amazon’s data-availability guidance explains greyed-out fields, restricted combinations, and shorter history for some selections. A blank may mean the data is unavailable, rather than zero activity.

Decide what to do with five searches

Here is a separate example at a smaller scale. A fictional US rose-pruning paperback earns $5 royalty per copy after printing deductions. All units are click-attributed sales of that edition, with no other editions or KU income included.

The Cost column uses Total cost. Royalty less cost is calculated in the worksheet, not taken from an Amazon search-term profit column.

Scroll horizontally on a small screen to read the table.

“Pruning roses for beginners” leaves $4 after advertising, before other business costs. You could continue testing or set up a more focused targeting test. Two sales are too few to establish a dependable conversion rate, and an exact target can still match variations.

The calendar query fits if the book includes a useful pruning calendar. Eight clicks without a sale leave the result uncertain. Further testing needs an allowance you can afford.

“Gardening” needs an intent check. Someone looking for a general gardening book may want more than this specialist manual provides. Review the surrounding searches before reducing the parent target’s bid or adding an exclusion that could also remove useful longer queries.

“Artificial rose arrangements” asks for something a manual about living plants does not cover. You can recognize that mismatch without buying a fixed number of clicks. Check the exclusion scope before saving it.

The disease query produced a sale, but does the book meet that reader’s expectations? If disease identification is barely covered, review the book’s promise and targeting before increasing the bid.

Together, these rows cost $20.40 and produce $15 in assumed royalties, a $5.40 shortfall. Two rows contributed more in royalties than they cost, so inspect the individual searches before making a broad reduction.

The small amounts make the arithmetic easy to follow. Set your own testing allowance from royalty contribution, plausible conversion, reporting maturity, and the loss you can fund. The affordable-CPC guide explains the calculation.

Keep Kindle Unlimited income separate

An eligible ebook may have attributed reading activity even when a row shows no purchase. Where compatible, inspect KENP read, Kindle Unlimited qualified borrows, and Estimated Kindle Unlimited royalties. Select the click, promoted, or halo variants that fit your review.

Pages read, borrows, and royalty estimates measure different things. A borrow count is not a dollar amount. Amazon’s qualified-borrow metric reference defines estimated KU royalties as estimates attributed to ad interactions. Label them as estimates and keep them separate from settled KDP earnings.

Avoid counting the same reading twice. Do not add a total to its promoted/halo or click/view components. If you have estimated royalties for a set of reading activity, do not also add a page-count calculation covering that activity.

If the builder cannot combine the KU detail you need with Search term, keep it at the level available. Campaign income cannot be assigned to individual query rows as if it had been observed there.

The report-reconciliation guide explains differences between advertising attribution and KDP totals. Halo outcomes alone do not establish series read-through or prove that an ad created an additional sale.

Record the action beside the evidence

Keep a decision record for the rows you review.

Scroll horizontally on a small screen to read the table.

For the calendar query, you might write: “Calendar is included. Recent outcomes remain provisional. Allow up to $4 additional cost, review after comparable complete reporting days, and reduce exposure earlier if the allowance is approached.” You have chosen how much more to risk; the extra $4 may still leave the result uncertain.

Before excluding a relevant term, check for overlap with the negative-keyword audit. Use the clicks-without-sales checklist when relevant traffic repeatedly fails to produce purchases.

When combining exports, keep daily rows separate from summaries of the same activity. Calculate combined rates from their totals instead of averaging row percentages. Combined CPC, for example, is combined applicable cost divided by combined clicks.

More books mean more rows to review

Separate campaigns for books, formats, and markets can make results easier to interpret. They also give you more campaigns to check. The structure helps you locate a decision, but you still have to make it.

20 books advertised in two formats across three marketplaces create 120 book, format, and marketplace combinations. That is before multiple campaign types, discovery tests, and the search terms inside each target. It is a workload example, not a recommended campaign count; some books and formats will not suit every market.

Putting all the rows into one spreadsheet leaves you with the job of keeping book and campaign identities intact, separating currencies and royalty assumptions, avoiding overlapping exports, and checking how recent the evidence is. Waste can be hard to spot in a campaign total or among thousands of unsorted rows.

You need to find the traffic worth reviewing, keep its context, make a decision within the right scope, and return to it as results change. The catalog-management guide explains the boundaries between books, formats, and markets.

How Adigy handles the ongoing review

A downloaded report contains the data available when it was generated. Rerunning or scheduling it refreshes the export. Someone still has to combine the results, find the searches that need attention, and apply decisions.

Adigy automatically collects search-term information in an updated list with search and filters. You can inspect the traffic behind campaign and target averages without rebuilding a large workbook each time, while delegating the routine work of finding and managing weak traffic.

Its performance filtering algorithm gives more weight to recent data and dynamically discounts older data. A target with strong historical results may have declined recently; weak older results may also hide an improvement. The weighting helps prioritize the evidence used in ongoing management instead of treating every historical click as equally representative of current performance.

Negative Oracle applies exclusions automatically when there is sufficient performance evidence. It can also assess relevance when data is sparse. Performance filtering and Negative Oracle help prune unsuitable or uneconomic traffic without requiring a publisher to spot a poor campaign average and stop everything inside it.

Authors keep the review, reversal, protection, and aggressiveness controls described in the negative-keyword guide. Recent conversion data may still be provisional. Review relevance judgments too, and use the controls to correct a mistaken exclusion or protect a term you want to keep.

Begin with one clear mismatch or a material unresolved cost, and record what should happen next. If repeating that review across books, formats, Sponsored Products and Sponsored Brands campaigns, and marketplaces takes more time than you want to spend, Adigy’s Amazon Ads management software provides the search-term view and ongoing management.