Search Term Report

The Search Term Report (STR) is the single most important report inside the Amazon Ads Console. It is the only place where you can see the actual shopper queries that triggered your ads, broken out from the keywords you bid on. Every other report aggregates at the keyword, ad group, or campaign level; the STR aggregates at the search term level, which is where all bidding decisions ultimately get made.

What the STR contains

Per row: the campaign, ad group, the targeted keyword (or auto-targeting type) you bid on, the customer search term Amazon matched it to, and the standard performance columns — impressions, clicks, CTR, spend, sales (1-day, 7-day, 14-day), orders, units, conversion rate, ACOS, and ROAS.

A typical mature account on a single SKU will produce a STR with dozens to hundreds of distinct search terms per week, even though the keyword list is far smaller. Broad-match keywords are the main multiplier — one broad-match seed can fan out to 40+ search terms in a single week.

What the STR is used for

Two workflows, run on the same export:

1. Harvest workflow

Filter the STR for high-converting, low-ACOS search terms surfaced by broad-match and auto-targeting campaigns. Promote each into a dedicated exact-match campaign at a math-derived bid. See Keyword Harvesting for the complete loop.

Filter recipe:

Match type = Broad OR campaign type = Auto
Orders ≥ 3
ACOS ≤ Target ACOS × 1.2
CVR ≥ campaign rolling average

2. Negative workflow

Filter the STR for high-click, zero-order search terms — the spend leak. Add each as a Negative Exact keyword to block future bidding.

Filter recipe:

Clicks ≥ 15
Orders = 0
Spend ≥ €5 (or your category's "expensive enough to care" threshold)

Both filters together typically convert one weekly STR pull into ~10–30 actionable changes. An account where the STR is downloaded but no action is taken on it is an account paying for the data and not using it.

Pulling cadence

Daily during the first 14 days after a new SKU or new keyword goes live (data is sparse but every signal counts).
Weekly as steady-state cadence — Monday for trailing 7 days is the canonical pattern.
Hourly via Marketing Stream for high-velocity accounts and during peak events. Amazon Marketing Stream doesn't replace the STR but it surfaces winning and losing terms within hours rather than days.

Data quirks to know

Lookback window matters. The STR defaults to 7-day attributed sales. If you've changed your default attribution window, the STR numbers shift accordingly. Lock the window before comparing reports week-over-week.
Aggregated terms. Amazon will sometimes report a search term as * (asterisk) when the underlying query is below their privacy threshold. These rows still spend; treat them as a single unactionable bucket.
Sponsored Brands and Sponsored Display have separate STRs in different sub-menus of the console. They follow the same harvest/negative logic but the data is partitioned by format.
Auto-targeting "match type" column. For auto campaigns, the keyword column is replaced with the targeting type: close match, loose match, complements, substitutes. Each behaves differently and can be bid independently.

Common mistakes

Pulling but not actioning. The STR is a data source, not a deliverable. The work is the harvest and negation, not the download.
Using the wrong attribution window for STR vs. campaign-level decisions. Mixing 1-day STR with 14-day campaign reports produces contradictory ACOS numbers and bad bid decisions.
Ignoring auto-campaign STRs. Auto campaigns are the lowest-friction discovery engine and their STRs are often the richest harvest source — yet most accounts ignore them in favour of broad-match.
Aggregating across SKUs. STR analysis at the parent-ASIN or brand level hides SKU-specific keyword dynamics. Pull STRs at the SKU level whenever the SKU carries its own ad budget.