Broad, phrase, exact — match types, with the gotchas Amazon doesn't print.
Three match types, three different breadths of query coverage, three different roles in a campaign structure. The full match-type rule-set with the close-variant behaviour that catches everyone the first time.
Match types are the keyword version of how flexibly should Amazon interpret what I typed. Three options on Sponsored Products and on Sponsored Brands manual keyword campaigns: broad, phrase, exact. Plus a fourth — Broad PLUS — that exists only on Sponsored Brands and gets its own episode.
Exact match
Matches the keyword and its very close variants — plurals, common misspellings, accented forms. Does not match reordered words. Does not match additional words before or after. Highest predictability, lowest reach. Exact match is the home of your harvested-winner manual campaigns: keywords you have already proven convert.
Phrase match
Matches the keyword as a contiguous phrase plus close variants, with additional words allowed before or after. "running shoes" phrase will match "best running shoes" and "running shoes for women" but not "shoes for running". Phrase is the middle bucket — useful for query coverage on modifiers you have not yet harvested.
Broad match
Matches the keyword in any order plus close variants plus synonyms. "running shoes" broad can match "sneakers for jogging". Maximum reach, minimum predictability. Broad is the discovery bucket and is dangerous without a strong negative list — its primary failure mode is matching against tangentially-related queries that drain budget.
The close-variant rule that catches everyone
Amazon applies close-variant matching at all three match types. That means exact match is not exactly exact, phrase is not exactly phrase. Close variants include plurals, misspellings, and abbreviations Amazon has decided are equivalent. You cannot turn close-variant matching off. The implication: even exact-match negatives sometimes do not fully block a search term, because Amazon judges the served term to be a close variant of a different keyword.
How the three match types should sit together
The three roles, in order:
- Broad and auto are the harvesting layer — they discover converting queries.
- Phrase is the consolidation layer — it covers the modifier variants of harvested winners.
- Exact is the conversion layer — proven keywords with the highest bid headroom.
Money typically flows broad/auto → exact over the life of a campaign. The diagnostic that says a campaign structure is healthy: most of the spend in the long run sits in exact match, with broad/auto getting just enough budget to keep finding new winners.
Negatives across match types
Negative keywords come in negative exact and negative phrase only. There is no negative broad — and you would not want one; it would over-block. The standard pattern: every harvested winner that moves from broad to exact gets added as a negative exact on the broad campaign, so the same query does not get auctioned twice.
Watch Episode 13: Broad, phrase, exact (German)
The German walkthrough — match types compared, with the close-variant trap.
Run match types as a system, not a setting.
AMALYZE shows where broad, phrase and exact overlap on the same search terms — so the next negative you add closes the leak, not the campaign.