Keyword cannibalization happens when more than one of your own campaigns, ad groups or products is eligible to show for the same search term, so they compete against each other in the same auction. The common version is a broad-match keyword in a discovery campaign and the same term in an exact-match campaign, both live with no negatives between them. Another is two similar ASINs, or two SKUs on one ASIN, targeting the same keywords with no plan for which should win.
Why it matters
Amazon shows one Sponsored Products ad per advertiser per placement, so cannibalization does not double your impressions. What it does is let your campaigns bid up the price of your own click, and it scatters the performance data across several places so no single campaign has enough to make a bid decision on. The exact-match campaign built to control a keyword ends up controlling nothing, because the broad campaign keeps taking a share of its traffic at whatever bid it happens to have.
Cannibalization between products is a strategy question rather than a structural error. Two of your ASINs on the same keyword can be right if both convert, but only one should be the ranking priority, and the other should not be bidding against it for top of search.
What moves it
- Missing negatives. Every term promoted to an exact campaign should be negated in the campaign it came from.
- Match-type mixing inside one ad group, which hides the overlap.
- Duplicate campaigns created over time by different people or tools.
- Automatic campaigns left running with no negatives, which will eventually match every term the manual campaigns target.
- Variation families advertised as both parent-level and child-level ad groups.
How we use it
Overlap is one of the seven audits we run before changing a bid: every keyword is listed with each campaign it appears in and each campaign's match type and bid. Structure is then rebuilt so every campaign has a single purpose, terms move through the ladder from discovery to control with negatives at each step, and one ASIN is named as the ranking priority for each keyword group.