Amazon Ads Targeting: How Automatic, Keyword, and Product Targeting Work

The three targeting types, in plain terms

Every Sponsored Products, Sponsored Brands, or Sponsored Display campaign runs on one of three targeting types: automatic, keyword (manual), and product (ASIN or category). You choose one when you build the campaign and can't switch it once the campaign is live — you'd have to build a new one.

Automatic targeting hands the matching to Amazon. It uses four sub-types — close match, loose match, substitutes, and complements — to decide which search terms and product pages your ad shows against, based on your listing content and past purchase behavior. You set one bid, or bids by sub-type, and Amazon fills in the rest.

Keyword targeting is manual: you choose the search terms and the match type (exact, phrase, or broad) for each one. You get full control over what you're paying for, but you have to find those terms first — usually by mining a search term report from an automatic campaign.

Product targeting lets you bid on specific ASINs or entire categories instead of search terms. This is how you show up on a competitor's product page, or across a category with price, rating, or Prime-eligibility filters applied. Negative targeting — excluding keywords, ASINs, or brands — sits on top of any of the three; it isn't a fourth type, it's the subtraction that keeps the other three clean.

Keyword match types: exact, phrase, and broad — and what each actually does

Exact match shows your ad only for the term you entered, plus very close variants Amazon considers equivalent, like plurals. Phrase match shows it for any search containing your keyword in that order, with words allowed before or after. Broad match shows it for any search containing your words in any order, plus synonyms Amazon decides are relevant — and Amazon has widened what 'broad' is allowed to mean over time, so a broad keyword can pull in less relevant traffic today than it did a year ago.

The practical rule: broad and phrase are for discovery, exact is for control. Start broad or auto to learn what people actually type, then move the terms that convert into their own exact-match campaign where you can bid precisely. Terms that don't convert become negatives — not because they're bad keywords in general, but because they're not buying-intent for your specific product.

Product targeting: showing up where keywords can't reach

Keyword targeting only works if a shopper types something. Product targeting works when they don't — when they land on a competitor's listing, browse a category, or check a complementary product. You can target a specific ASIN, a list of ASINs, or a whole category refined by brand, price band, star rating, or Prime eligibility.

Category targeting is the blunt version — cheap impressions, wide net — and it's worth running with a low bid and a defined refinement, not left open. The most common use of negative product targeting is excluding your own brand from category campaigns, so branded traffic doesn't get double-counted against a campaign meant to find new-to-brand customers.

How auto data becomes a manual campaign: a worked example

Say a new automatic campaign for a $30 kitchen product runs for three weeks. The search term report shows 40 clicks and 3 orders on the term 'stainless steel milk frother' at a $1.10 average CPC — about $44 spent against roughly $90 in sales on that one term, a 49% ACOS, versus close to 90% ACOS on everything else in the campaign.

That's the signal to act: move 'stainless steel milk frother' into its own exact-match keyword in a manual campaign, bid slightly above what auto was paying to hold placement, and add it as a negative exact in the automatic campaign so the two stop competing for the same auction. Leave the automatic campaign running — it's still the cheapest way to find the next term.

The mistakes that waste the most spend

The most common mistake is leaving a brand-new automatic campaign on Amazon's 'increase bids' dynamic bidding setting before it has enough data to tell sub-types apart. We've done this ourselves on a launch — turned on aggressive dynamic bidding day one, watched CPCs climb on loose-match and substitute placements before the algorithm had enough signal to separate them from close match, and had to walk the bids back down a week later.

The second mistake is treating negative targeting as free cleanup. A negative exact match blocks only that exact term. A negative phrase match can silently block dozens of related searches you never see, including ones that convert. Add negatives one at a time and check the search term report afterward — don't batch-add a list from a spreadsheet without reviewing what's in it.

The third is running category targeting with no refinement. An unrefined category target can spend a full day's budget on impressions next to products that share a category and nothing else — wrong price point, wrong rating band, wrong customer.

When you've done all this and targeting still isn't working

If you've moved converting terms to exact match, added negatives, and refined product targets, and ACOS still isn't moving, the problem usually isn't targeting anymore — it's the listing, the price, or the review count the ad is sending traffic to. Targeting controls who sees the ad; it can't fix what happens after they click. Check the listing's conversion rate before touching bids again.

If a keyword or ASIN target has spent budget for two weeks with zero orders, pause it rather than lowering the bid repeatedly. A bid that's too low to win auctions and a bid that's winning auctions on the wrong term look identical in the data until you check impressions.