A negative keyword tells Amazon not to display a given ad when a shopper searches that exact term, letting a seller exclude search traffic unlikely to convert without lowering bids on the keywords they do want. Negative keywords are added in two stages: preemptively, using common sense to exclude obviously mismatched but similarly-worded searches (usually as exact or phrase match), and then reactively, by mining PPC search-term reports for keywords that are accruing spend without conversions. This step is commonly cited as saving hundreds to thousands of dollars in wasted ad spend over a campaign's life, and works alongside Match-Type Overlap Trimming & Manual Campaign Restructuring as a cost-control lever distinct from bid or budget adjustments.
Two distinct mechanisms exist for stopping spend on a keyword, and they apply to different situations:
Negative keywords get added in two distinct phases: (1) pre-launch — common-sense negatives added before the campaign goes live, filtering out obviously irrelevant terms; and (2) post-launch — ongoing negatives pulled from actual PPC search-term reports as non-converting search terms surface once the campaign is spending.
A two-stage process for building out negative keywords: (1) pre-launch, common-sense filtering — strip obviously irrelevant or similar-sounding terms before the campaign goes live; (2) post-launch, ongoing removal of keywords found to be non-converting in PPC search-term reports. Negative keywords are set to exact or phrase match, not broad, and the list is refined continuously as report data accumulates to cut wasted spend.
Negative keyword hygiene matters more as targeting gets looser: exact match campaigns need the least negative-keyword maintenance, phrase match more, broad match more still, and automatic targeting the most — since automatic campaigns let Amazon's algorithm choose search terms with no keyword-level guardrails at all. See Keyword Match Types (Exact / Broad / Phrase) for the match-type spectrum this tracks, and Automatic Campaign Four-Way Split (Close Match, Loose Match, Substitutes, Complements) for how automatic targeting subdivides.
Seed the negative keyword list from pre-launch keyword research (terms already known to be irrelevant), then keep refining it using downloaded Search Term Reports to catch irrelevant queries that actually triggered the ad. Apply negatives most heavily on broad and automatic campaigns, where match looseness invites irrelevant traffic, and least on exact-match campaigns, where the target is already tightly scoped.
In auto campaigns, negative exact and negative phrase targeting are among the few control levers a seller retains, since Amazon otherwise picks all targets itself (see Automatic Campaign Four-Way Split (Close Match, Loose Match, Substitutes, Complements)). Used to exclude irrelevant search terms surfaced by autotargeting — e.g., excluding 'metal' or 'chrome' search terms for a wooden phone holder listing.
Two distinct negative-keyword controls: negative exact blocks a specific search term, while negative phrase blocks any term containing a given word (e.g., adding 'metal' or 'chrome' as negative phrase targets for a wooden phone holder blocks all searches containing those words). For Automatic Campaign Four-Way Split (Close Match, Loose Match, Substitutes, Complements) auto campaigns, the primary source for these negatives is the auto campaign's own Amazon Search Term Report after it has accumulated data — irrelevant terms surfaced there become negative phrase (or exact) targets.
Two approaches to trimming irrelevant targets from auto/broad campaigns: proactively negative-exact-match keywords that are already targeted in other campaigns to avoid overlap, or let the campaign run and add negatives as search-term data comes in. At minimum, add negative keywords/phrases for competitor or other brand names and clearly irrelevant terms — e.g., negative phrase match "metal" on a wooden phone stand listing — then refine further over time. "Don't overthink this. You can add these as time goes on when you look at the data." This is the pruning step referenced in Automatic Campaign Four-Way Split (Close Match, Loose Match, Substitutes, Complements).
Two different philosophies for assigning negative keywords across a multi-campaign structure: (1) negate only terms known to be irrelevant to the product, leaving campaigns free to overlap on relevant terms, or (2) negate every term that is already being targeted in another campaign, so each campaign is fully segregated and no keyword can trigger more than one campaign at once. The second approach trades some flexibility for cleaner attribution and bid control per keyword.
Two distinct negation moves: phrase-negate blocks any search query containing that phrase (broad protection), while exact-negate blocks only that literal query (narrow protection). Continuously mine the Amazon Search Term Report and phrase-negate irrelevant or non-converting queries — especially in auto and broad-match campaigns, where Amazon has the widest targeting discretion (e.g. a wood product surfacing for 'titanium').
Separately, some sellers negate every keyword that's actively targeted in another campaign, purely to keep campaigns segregated (so spend/attribution for a keyword isn't split across multiple campaigns) — a different motive than negating for irrelevance.
For auto campaigns, wait for the first search term report (~24 hours after launch) before negating terms. Add clearly irrelevant search terms as negatives at that point, but avoid negating a term purely because of high ACOS on a handful of clicks — early volume is too thin to distinguish a bad term from normal variance.
Из тем: PPC Campaign Structure & Bidding