All guides
Amazon PPC workflow · published August 2026

How to turn an Amazon search term report into safe PPC actions

A search term report is evidence of what shoppers actually entered or what product detail page matched an ad. It is not a ready-made action list. A reliable workflow separates the shopper query from the target, gives attribution time to settle and applies one controlled action at a time.

Start with the distinction between a target and a search term

A target is the keyword, category, product or automatic targeting expression configured in the campaign. A search term is the customer query or matched ASIN that caused the ad to be eligible. One broad target can therefore produce many search terms with very different economics. Judging only the target hides both the profitable query that should be harvested and the irrelevant query that should be blocked.

Amazon explains that the Sponsored Products search term report contains terms with at least one ad click. That scope matters: campaign impressions can be higher than report impressions, and a missing term is not proof that the campaign had no traffic. Review the current definition in the official Amazon Ads search term report documentation before comparing totals from different reports.

Normalize the evidence before calculating actions

Use one marketplace, currency and date range at a time. Keep campaign, ad group, ad type, match type and target identity attached to every term. Aggregate duplicate rows only when those dimensions match. Otherwise a term from a defensive campaign can be merged with the same text from a discovery campaign and produce a misleading blended ACOS.

  • Use spend, clicks, attributed sales and attributed orders from the same report window.
  • Keep text queries and matched ASINs distinct because the available actions differ.
  • Mark the last active date so old waste does not look like current bleeding.
  • Retain the current bid and any recent action so a stabilization period can be respected.

Classify the term before changing anything

EvidenceLikely decisionSafety check
Sales at acceptable economicsHarvest into controlled targetingConfirm it is not already targeted elsewhere
Sales but ACOS above the SKU guardrailReview or lower bidAllow attribution and recent bid changes to settle
Meaningful spend or clicks with zero salesNegate or lower exposureConfirm the term is irrelevant or persistently inefficient
Too little evidenceKeep observingDo not convert uncertainty into a destructive action

Thresholds must be tied to product economics. Ten euros of spend may be decisive for a low-margin item and noise for a high-price item. Use the SKU contribution margin, target ACOS and expected conversion rate to define how much evidence is enough. A global rule without product context is simple, but it is not safe.

Harvest converters without creating internal competition

A converting search term can be moved into an exact or otherwise controlled target where its bid and budget are easier to manage. Before creating it, check whether the same term already exists in the destination campaign and whether the source campaign needs an exact negative to prevent duplicate routing. Preserve the original campaign and ad group IDs in the action log so the result can be verified later.

Do not treat every sale as a winner. A term with one order can still lose money after VAT, FBA fees, product cost and returns. Conversely, a launch term may be allowed to exceed the mature target while it builds relevance. The decision should state which economic rule was used, not only that a sale occurred.

Negate persistent waste, not temporary uncertainty

Negative exact is appropriate when one precise query is proven irrelevant or uneconomic. Negative phrase has a wider effect and should be reserved for a repeatable unwanted intent. Never negate a term only because today's order count is zero. Advertising attribution can arrive after the click, and incomplete marketplace coverage can make a recent window look worse than it is.

The safest order is: verify the date window, check current activity, confirm zero settled sales, inspect semantic relevance, preview the scope of the negative and only then execute.

Change bids as a measured experiment

A bid change should include the old bid, proposed bid, reason, maximum allowed movement and next review date. Avoid repeated changes while the previous one is stabilizing. When a target is already at its floor, another decrease cannot stop meaningful waste; the correct next decision may be a negative, pause or campaign-level investigation instead.

Use Sniper Hub to keep the report dimensions, economic rules, stabilization state and Amazon confirmation in one operational loop. For the narrower question of negative match scope, continue with the Amazon negative keywords guide.

Verify every write against Amazon

A successful API submission means Amazon accepted the request for processing; it does not prove that the final campaign state has changed. Store the submitted IDs, read the target back after processing and show confirmed, pending and failed states separately. If the readback does not match, keep the action pending and report the Amazon error rather than presenting an optimistic success.