Amazon Ads Search Term Report for Authors: Find What Shoppers Actually Typed

If you could only look at one report in Amazon Ads for the rest of your advertising career, it should be the Search Terms Report. It shows you the actual search queries, not keywords you set, but real text that real readers typed into Amazon, that triggered your ads. It shows how much you spent on each one and whether it produced any sales. It is the most direct possible evidence about what your actual book buyers search for, and every piece of optimization intelligence you need to improve your campaigns sits inside it.

Most authors either never open it or open it, feel overwhelmed by the spreadsheet, and close it again. This guide changes that.

What the Search Terms Report Actually Is

Keywords and search terms are different. A keyword or automatic target is the instruction in your campaign; a search term is the shopper query associated with a click. Phrase and broad match can therefore reveal longer or related queries that you never entered yourself, while automatic targeting can expose buyer language and product contexts you did not predict.

Amazon’s current Search term report includes search terms that generated at least one ad click. In some non-search contexts, Amazon can infer a search term from the surrounding context rather than reporting a literal typed query. That nuance matters: the report is powerful evidence about the traffic your ads attracted, but it is not a complete transcript of every impression or every word a shopper typed. Use the report to answer four questions: which search terms deserve direct targeting, which are clearly irrelevant, which need more evidence, and which reveal a mismatch between the campaign’s targeting and the book’s positioning.

How to Find and Download It

In the Amazon Advertising console (advertising.amazon.com), navigate to: Measurement & Reporting > Advertising Reports. Click Create Report. Under Report Type, select Search Term. Select your date range (use 14 days, see below). Select the campaign or campaigns you want to include, you can run the report for your entire account or for specific campaigns. Click Run Report.

The report generates and becomes downloadable as a .csv file within a few minutes for most accounts. For very large accounts with many campaigns, it may take up to 30 minutes to generate. Once downloaded, open in any spreadsheet application, Excel, Google Sheets, Numbers. The raw file has a header row and one row per search term per campaign per ad group.

Tip: create a template spreadsheet with pre-applied sorting, filtering, and conditional formatting for the columns you use most. After the first setup, each fortnightly review becomes a quick data paste rather than manual formatting work.

What Each Column Means

The Search Terms Report contains more columns than you need for standard optimization. These are the ones that matter:

Targeting Type: whether this search term was triggered by an automatic or manual campaign targeting type. This tells you which campaign structure generated the match, important for knowing where to act.

Match Type: for manual campaigns, this shows which match type (exact, phrase, broad) the triggering keyword uses. For automatic, it shows the sub-type (Close Match, Loose Match, Substitutes, Complements).

Keyword (Targeting): the keyword in your campaign that Amazon matched to this search term. This is important context, if a phrase match keyword “cozy mystery” is matching searches for “cozy mystery audiobook free download,” you may want to either tighten the match type or add a negative phrase for “audiobook” and “free.”

Customer Search Term: the actual search query the reader typed. This is the column you spend most of your analysis time on.

Impressions: how many times your ad appeared for this search term. High impressions with low clicks indicates relevance issues (the term surfaces your ad but readers do not click, cover and title presentation problem).

Clicks: how many readers clicked through to your product page. This is your sample size for evaluating each term.

Spend: total cost of all clicks from this term in the period. For author advertisers, Amazon Ads currently uses a 14-day click-attribution window for sponsored ads. That means purchases and eligible KENP reading can continue to be attributed after the click, so the most recent days in a report may still be incomplete. Use mature date ranges when judging borderline targets, and distinguish attribution lag from genuine underperformance.

Sales: revenue from those orders. ACoS: calculated ACoS for this term (Spend ÷ Sales × 100). Any term showing ACoS at or below your target with meaningful click volume is an immediate exact match candidate.

Choosing the Right Date Window

There is no single mandatory date window for every Search term decision. Author click attribution currently runs for 14 days, so the newest clicks can still gain attributed purchases or eligible KENP. That means a report ending today contains some immature rows even if the range itself covers two or four weeks.

For a recurring review, a 14-day or 30-day view can both be useful. The shorter range is more responsive to recent changes; the longer range helps lower-volume targets accumulate evidence. For borderline profitability decisions, consider excluding the newest portion or rechecking the term after attribution has matured. For obvious mismatch, act sooner.

Keep comparable periods when you are evaluating a change. If you changed a bid, cover, or price mid-period, note the date so you do not treat two different conditions as one clean sample.

The Weekly Review Workflow

Start by filtering the report to the spend and traffic that can change a decision. Review relevant queries with attributed outcomes, high-spend non-converters, and obvious audience mismatches. Then classify each row as promote, exclude, observe, or investigate.

Promote a query when repeat evidence justifies direct control. Exclude obvious mismatch or persistently uneconomic traffic at the appropriate negative match type. Observe plausible low-volume terms that have not yet gathered enough evidence. Investigate terms that look relevant but convert poorly, because the problem may be the detail page rather than the target.

Finally, log the action, date, campaign, and reason. A recurring workflow can be weekly, fortnightly, or more frequent for a high-spend account. The useful cadence is the one that catches material waste without turning immature attribution into constant bid churn.

Identifying Harvest Candidates

A harvest candidate is a search term whose evidence is strong enough that you want direct control over its bid and reporting. Relevance comes first; then consider clicks, attributed orders or KENP, CPC, ACoS, and the expected value of the reader. One order can be meaningful for an expensive niche title, while two orders can still be noisy for a low-cost, high-volume term.

Add a useful query as an exact target when you want to manage it deliberately. You may also choose to negative-exact it from the automatic or broader campaign if you want cleaner routing, but that is an account-design choice rather than a required “harvest” rule. Overlap does not automatically mean your campaigns are bidding against themselves in a way that inflates CPC.

The point of harvesting is to turn discovered buyer language into a more interpretable target set. Let the evidence threshold reflect the book’s economics instead of a universal two-order rule.

Identifying Negative Keyword Candidates

Negative candidates fall into two groups. The first is obvious mismatch: the wrong format, genre, audience, language, or intent. You can exclude those quickly because the relevance problem is clear. The second is plausible traffic that repeatedly spends without producing enough value. That group needs more context.

Amazon recommends allowing enough evidence before adding negative targets and provides guidance around 20 clicks, but that is not a universal profitability law. A high-CPC target can become uneconomic before 20 clicks; a very cheap and highly relevant target can deserve more observation. Compare spend with the value of a conversion and respect the 14-day attribution window for borderline decisions.

Choose negative exact when you want to block one specific query and negative phrase when every query containing that phrase is genuinely unwanted. Phrase exclusions can remove useful long-tail traffic, so keep a record of broad negatives and review them when the book or campaign changes.

The Promotion and Negation Process in Practice

When a useful query appears, add it to manual targeting in the form that matches the job you want it to do. Exact match gives the tightest control over closely matching queries; phrase can keep discovering modifiers around a proven core. Use the shopper’s actual language where it accurately describes the book rather than automatically rewriting it into a prettier phrase.

Then decide whether the originating campaign should keep access to the query. A negative exact can cleanly route that search into the manual campaign, but retaining overlap can preserve discovery and delivery. The right choice depends on budget pressure, reporting clarity, and how much you want the discovery campaign to continue exploring around that term.

Do not treat “add to manual plus negate in auto” as an inseparable two-action rule. The goal is control and learning, not migration for its own sake.

Secondary Insights in the Report

Beyond the harvest and negative keyword workflows, the Search Terms Report contains secondary insights that are valuable for understanding your book’s audience. High-impression, low-click terms: search terms generating many impressions but very few clicks indicate your book is appearing for relevant searches but readers are not clicking. The problem is at the impression level, your cover or title, as displayed in the search result, is not compelling enough to generate a click despite matching the search intent. These terms are not negative keyword candidates, they are conversion funnel signals pointing to a product presentation problem.

Audience vocabulary patterns: the full list of converting search terms is a map of your actual book buyers’ language. Read it as a creative brief. If converting terms cluster around “recipes” even though your cozy mystery only has one recipe element, that is a signal that prominently featuring that element in your description (and potentially adding more recipes) would improve relevance and conversion. If converting nonfiction search terms repeatedly include “practical” and “actionable,” your buyers are pragmatic, your description should reflect that vocabulary.

Match type effectiveness: filter the report by match type and compare the average ACoS per match type for the same keyword roots. If your phrase match keywords consistently outperform broad match at lower CPCs, that is a structural signal to reduce broad match proportion in your campaigns and invest more in phrase match.

How Often to Run This Process

Match review frequency to spend velocity. A high-spend launch can justify a quick weekly scan for large mismatches because expensive errors accumulate fast. A low-volume niche book can need longer before the same report contains enough traffic to support target-level decisions.

Keep attribution maturity separate from review cadence. You can review recent data every week while postponing borderline conversion decisions until the 14-day author attribution window has had time to fill in. Obvious relevance problems do not need to wait.

Amazon’s unified reporting can support longer historical comparisons, but maintain a simple change log as well. Reports tell you what happened; the log tells you what you changed before it happened.

Reading Mistakes That Lead to Bad Decisions

Author-ad attribution currently uses a 14-day click window, so recent searches may acquire additional attributed outcomes after you first download the report. When a decision is close, use a mature date window rather than treating the newest days as complete.

Acting on terms with fewer than 5 clicks. A term with 3 clicks and zero orders has not had enough traffic to evaluate. Negative keyword decisions based on this low a sample result in blocking searches that may in fact convert with sufficient traffic, you simply have not seen enough of them yet.

Sorting by ACoS and acting on terms with only 1-2 clicks. A term with 1 click, 1 sale, and 100% ACoS looks terrible on ACoS, but 1 sale on 1 click is a 100% conversion rate. Do not optimize on ACoS until there are at least 10+ clicks. Sort by orders and spend first, then let ACoS inform decisions only for terms with meaningful click volume.

Ignoring the Match Type column. A broad match keyword matching irrelevant searches is a match type problem as much as a keyword problem. If changing the keyword to phrase match would solve the irrelevance without removing it, that is the right fix, not adding a negative that might block good traffic.

Running the harvest cycle in one campaign without checking others. The same search term may appear across multiple campaigns if you run both automatic and manual with overlapping keyword pools. Check whether terms you are harvesting or negating appear in other campaigns too, and apply consistent treatment across all of them.

For the full context of the harvest-and-scale cycle this report powers, see our automatic vs manual targeting guide. The KDP Rank Fuel tools at rankfuel.vappingo.com include the Amazon Ads Generator, which produces complete campaign structures, including keyword lists and starting negative keyword sets, that you can deploy before the first Search Term Report cycle, minimising early-stage waste. Manuscript proofreading from Vappingo ensures that the readers your optimized campaigns deliver convert at the rates your ACoS targets require, a professionally proofread book earns stronger reviews and repeat readers who mention it in their searches.

Turn One Report into Four Different Decisions

The Search term report can answer at least four questions. What should I target more deliberately? Look for relevant terms with repeat value. What should I block? Find clear mismatch or spend the economics cannot recover. What language should change my listing or positioning? Repeated shopper phrasing can reveal how readers describe the problem, trope, audience, or outcome. What campaign is doing the wrong job? If a tightly controlled campaign keeps matching into unexpected contexts, inspect the targeting setup rather than endlessly adding negatives.

Save snapshots instead of overwriting the same spreadsheet. Amazon’s unified reporting makes longer historical comparisons easier, but your own decision notes still matter. A term that looked weak before a price or cover change may behave differently afterward, and you need the timeline to know whether the campaign or the product page changed first.

Frequently Asked Questions

Make Search Terms the Feedback Loop

The report becomes powerful when each row creates a possible action: promote, exclude, observe, or rethink the campaign. Repeating that loop is how a raw discovery campaign becomes a controlled acquisition system.