What is a Search Term Report?

The search term report lists the queries shoppers used before clicking your Sponsored Products or Sponsored Brands ads, with the impressions, clicks, spend, orders and sales each query produced. Your keywords and targets are what you bid on. Search terms are what shoppers actually searched for, and the two lists are rarely the same.

What the report contains

Amazon's help page for the Sponsored Products search term report and its reporting API documentation set the rules (both checked 2026-10-03):

• It covers Sponsored Products and Sponsored Brands campaigns.
• It includes only search terms that produced at least one click, so its impressions will not match the campaign manager.
• A search term can be one a shopper typed, or one Amazon inferred as the best match in a non-search context.
• Entries that look like b00ipgvvz4 are ASINs: the product page your ad appeared on. Amazon says these appear only for Sponsored Products automatic campaigns and product-attribute targeted ad groups.
• For sellers, Sponsored Products rows credit sales made within 7 days of the click and split them into promoted sales, from the advertised SKU, and halo sales, from other SKUs.
• The lookback window is 65 days for Sponsored Products. The API documents 60 days of retention for Sponsored Brands.

Every row pairs a search term with the keyword or target that matched it and its match type. That pairing is the value of the report: it shows what each keyword actually bought.

The formula: is zero orders bad luck or a signal?

A common question asked of this report is whether a term that spent money without an order should go. A fixed rule such as ten clicks and no sale ignores the number that decides it: how often the campaign converts. The chance of seeing no orders purely by luck is:

Chance of 0 orders = (1 − CVR) ^ Clicks

CVR is the conversion rate, orders ÷ clicks, of the campaign the term ran in, measured over the same window.

Worked example

The inputs are invented for the arithmetic. Suppose a campaign converts 10% of its clicks.

Our house rule treats a zero-order term as waste only once that chance has fallen to 10% or less. At a 10% conversion rate that takes 22 clicks, because 0.9^22 is about 9.8%. In a campaign that converts 20% of clicks it takes 11, because 0.8^11 is about 8.6%. The click count comes out of your own conversion rate. It is not a constant.

Now a term that sells. Term C took 40 clicks for $48 of spend and produced 4 orders of the advertised SKU, worth $120.

• ACoS = $48 ÷ $120 = 40%
• Revenue per click = $120 ÷ 40 = $3.00, and actual CPC = $48 ÷ 40 = $1.20
• With a 30% target ACoS, target CPC = $3.00 × 0.30 = $0.90

If the product's break-even ACoS is 50%, term C is profitable but bought above target. It is a harvest candidate with a bid problem, not a negative.

How to use it to make decisions

1. Harvest what sells. A relevant term with two or more orders of the advertised SKU earns its own exact match keyword, where it gets its own bid. Count promoted sales, not total sales, so halo orders do not promote the wrong term. See keyword harvesting.
2. Negate only what is irrelevant. A negative is for a query that should never show your product. Use negative exact for a single query, and widen to phrase only when you mean to block every query containing those words.
3. Treat a converting term with poor ACoS as a bid problem. Bids attach to keywords and targets, so make the change on the targeting report row that bought the term. Your ceiling is break-even ACoS, which the break-even ACoS calculator works out from your price, fees and cost, and our Amazon bid optimization guide covers stepping a bid toward target.
4. Look for shared words. Hundreds of low-spend terms can share one irrelevant word. Grouping terms by one, two and three word fragments, called n-grams, shows waste that no single row reveals.
5. Save it on a schedule. With a 65-day lookback, history you do not save is gone.

Common mistakes

• Negating on performance. A term with sales is working. Negating it for a high ACoS throws away demand you could have kept at a lower bid.
• Using one click threshold for every campaign. The evidence needed depends on the campaign's conversion rate, as the example shows.
• Judging the last few days. Sponsored Products sales are credited back to the click date for up to 7 days, so recent rows look worse than they will end up.
• Expecting impressions to reconcile. Terms with impressions but no clicks are left out, so the report understates impressions.
• Ignoring ASIN rows. They are product pages your ad appeared on. Target the good ones on purpose and exclude the poor ones with negative product targeting.
• Negating a harvested term from its source by reflex. The source campaign often keeps winning the same click more cheaply. Our house view is to negate on harvest only under a real budget constraint.

How it relates to other reports

Amazon's search term impression share report shows, for each search term, your share of ad impressions and your rank against other advertisers, with a 90-day lookback (checked 2026-10-03).

How Shurq helps

• A Search Terms view in Ad Manager. Select rows and act on them in place. Harvest promotes the chosen terms as exact match keywords into the ad group you pick, with a preview before anything is sent. Negating at the source is off by default, and the keyword negatives it adds are negative exact.
• N-gram analysis. The same selection can be broken into word fragments with every metric re-added per fragment, and you can add negatives on the fragments you choose.
• A record you can reverse. Negatives added from the Search Terms view appear in Change History, where they can be undone.
• History past Amazon's window. Shurq keeps the search term rows it collects after you connect, so your history is not cut off at 65 days.
• Honest zero-order calls. Ask Shurq's chat about terms with spend and no orders and it applies the probability rule above, and it never proposes your own brand or ASIN terms as negatives.