Amazon PPC Match Types: Broad, Phrase, Exact and Auto

Shivam Kumar

TL;DR
Key takeaways
Amazon PPC match types decide which searches a keyword can trigger: exact is narrowest, phrase is in between, broad is widest.
Auto campaigns use four targeting groups instead: close match, loose match, substitutes and complements, each with its own bid.
In 2026 broad match is semantic and Amazon infers search terms for off-Amazon placements, so negatives carry more of the control.
Discover with auto and broad, prove terms in phrase, then scale winners in exact at your highest bid.
Amazon PPC match types are the rules Amazon uses to decide which shopper searches your keywords and product targets are allowed to serve on. Sponsored Products has three keyword match types (broad, phrase, exact), two negative match types, four automatic targeting groups and two ways to target a product.
This guide explains how each one behaves in 2026, how we bid them, and how we decide which to use at each stage of a campaign.
What Are Amazon PPC Match Types?
A match type is a setting on a keyword that tells Amazon how closely a shopper's search has to resemble that keyword before your ad can show. Amazon's keyword match type documentation, last updated July 29, 2026, says match types "allow you to fine-tune which shopping queries your ads are eligible to show against."
Most guides stop at broad, phrase and exact. In a real Sponsored Products account, every targeting option has its own version of the same idea: how much reach you trade for how much control.

Targeting option | Match types | Where it serves | Control |
|---|---|---|---|
Keywords (manual) | Broad, phrase, exact | Search results, plus inferred searches off Amazon | You pick the keyword and bid |
Negative keywords | Negative exact, negative phrase | Blocks searches | You pick what to exclude |
Automatic targeting | Close, loose, substitutes, complements | Search results and product pages | Amazon picks targets, you bid per group |
Product targeting (manual) | Individual ASIN exact or expanded, category | Product pages and search results | You pick ASINs, brands or categories |
Negative product targeting | Exclude products, exclude brands | Blocks product pages | You pick what to exclude |
If you're still mixing up the keyword and the search term itself, start with our guide to Amazon keywords vs search terms. This post assumes that difference and goes deeper on the match types.
Keyword Match Types: Broad, Phrase and Exact
All three keyword match types account for plurals, misspellings and translations, according to Amazon's documentation. After that, they behave very differently.

Broad Match
Broad match is the widest keyword match type. Amazon says a broad query can contain the keyword terms in any order and include "synonyms, and related terms as determined by the meaning of the keyword and the context of the advertised products."
The line that matters most is the last one: "The keyword itself may not be contained in the customer shopping query." Amazon's own example is the broad keyword "sneakers" matching "basketball shoes", "cleats" and "foam runners".
So treat broad as a discovery tool with a bid, not as a looser exact. In the accounts we manage, a broad keyword typically produces dozens of distinct search terms a month, and a handful of them carry most of the sales.
Phrase Match
Phrase match shows your ad when the shopper's search contains your keyword phrase, in the same word order, plus close variants. Amazon's example: the phrase keyword "curtain rod" can match "extra long curtain rods" and "curtain rods for bedroom".
Phrase is where we prove a theme. It keeps the core meaning fixed while letting modifiers like size, color, use case and audience come through. It also produces the cleanest long-tail candidates for exact.
Exact Match
Exact match shows your ad when the search is your keyword, in the same order, or a close variant. It's the most restrictive keyword match type and, in Amazon's Sponsored Products targeting guide, "tends to result in the highest conversion rates."
"Close variant" is wider than many sellers expect. Amazon's documentation says the exact keyword "outside lights" can still serve on "outdoor lights", so check your exact keywords in the search term report too.
Broad | Phrase | Exact | |
|---|---|---|---|
Word order | Any | Fixed | Fixed |
Extra words in the search | Yes | Yes, before or after | No |
Synonyms and related terms | Yes | No (close variants only) | No (close variants only) |
Reach | Highest | Medium | Lowest |
Typical conversion rate | Lowest | Medium | Highest |
Main job | Discovery | Proving themes, long tail | Scaling proven terms |
Bid position | Lowest | Middle | Highest |
Rules That Apply to All Three
Three rules from Amazon's documentation shape how we build campaigns:
You can't change a match type after the campaign is created. Add the keyword again with the new match type instead.
The highest bid wins when match types overlap. If one keyword is live in broad and exact in the same campaign, Amazon uses whichever has the higher bid for a matching search.
Broad match modifiers exist only in Sponsored Brands. Putting "+" before a word ("+kids shoes") forces that word into the search. The old version of this post implied it works in Sponsored Products; it doesn't.
That second rule is why we keep match types in separate ad groups or campaigns. When exact and broad share an ad group, a high broad bid can quietly take traffic you meant to price in exact.
Negative Match Types: Negative Exact and Negative Phrase
Negative match types are the reverse of keyword match types: they tell Amazon which searches your ad must not serve on. Sponsored Products has no negative broad.
Amazon's negative targeting help page defines them:
Negative exact blocks the exact phrase and close variations. Up to 10 words and 80 characters.
Negative phrase blocks any search that contains the complete phrase or close variations. Up to 4 words and 80 characters.
For product pages, negative product targeting excludes specific ASINs or whole brands. Amazon says negative keywords can take up to 72 hours to apply and negative products up to 96 hours.
We default to negative exact and keep negative phrase for concepts that never convert, like "used" or "refurbished" on a new product. Our negative keywords guide covers the click thresholds and campaign vs ad group levels in depth, so we won't repeat them here.
Automatic Targeting: Close Match, Loose Match, Substitutes and Complements
Automatic campaigns don't use keywords, so they don't use keyword match types. Amazon calls their equivalent targeting groups, and there are four of them, each with its own on/off switch and its own bid.

Targeting group | Serves on | Amazon's example for "Doppler 400-count Cotton Sheets" | How we treat it |
|---|---|---|---|
Close match | Searches closely related to your product | "cotton sheets", "400-count sheets" | Highest auto bid, best converter |
Loose match | Searches loosely related to your product | "bed sheets", "bath towels" | Lower bid, heaviest negation |
Substitutes | Detail pages of similar products | "300-count cotton sheets" | Conquesting, harvest ASINs |
Complements | Detail pages of complementary products | "queen quilt", "feather pillows" | Lowest bid, cross-sell |
The examples come from Amazon's match type documentation. The groups mirror the keyword side: close match behaves like phrase on your listing's own language, loose match like a very broad keyword.
Amazon's targeting guide says you can set one default bid or bid per targeting group. Bid per group. A single default bid pays the same for a "bath towels" click as for a "400-count sheets" click, and those rarely convert alike.
Auto campaigns also show you ASINs in the search term column. Those are the product pages your substitutes and complements groups served on, and the converting ones become manual product targets.
Product Targeting and Expanded Product Targeting
Product targeting lets you choose specific ASINs, brands or categories to show your ad on. It has its own two-level match type, which most match type guides skip.
The Sponsored Products product targeting documentation says that for an individual product "you can choose to target a specific ASIN exactly, or target a specific ASIN and other similar ASINs, including substitutes, complements, and other related products."
Individual product, exact: your ad is eligible on that one ASIN's detail page and related placements.
Individual product, expanded: the ASIN plus products Amazon judges related to it. Amazon launched expanded product targeting at unBoxed 2022 and says the expanded set is "updated dynamically."
Category targeting: a whole category, refined by brand, price range, star rating and Prime eligibility.
Expanded is to product targeting what broad is to keywords. Amazon's coffee maker example includes other coffee makers, coffee filters and coffee beans under one target.
We use expanded targets to find ASINs and exact targets to bid on them. Amazon's search term report documentation suggests the same: use the report to find expanded targets that meet your goals and promote them to individual targets.
How Match Types Behave in 2026
The definitions above are current. What changed over 2025 and 2026 is how much matching Amazon now does without a typed search, and how many features run on top of your existing match types.

Broad Match Is Semantic, and Amazon Says So
We found no dated Amazon announcement of a new broad match algorithm in 2025 or 2026. What we did find is the documentation itself: broad matches by "the meaning of the keyword and the context of the advertised products," and the keyword need not appear in the search at all.
In practice, broad now behaves closer to an auto campaign built around one idea. Budget it like discovery and review its search terms weekly.
Some Search Terms Are Inferred, Not Typed
Amazon's match type page opens with a new sentence: when a customer search term isn't available, "Amazon will infer a search term from the context that best matches your keywords." The search term report page says the same about non-search placements, such as social sites.
That matters because Sponsored Products now serves off Amazon by default. PPC Land reported that from August 10, 2026, campaigns were enrolled in creator placements at existing bids and budgets, and that top of search and product page bid adjustments don't apply off Amazon.
Inferred terms can be negated like any other. Before you do, check whether an odd-looking term came from an off-Amazon placement rather than a real search.
Prompts Run on Your Keywords
Sponsored Products and Sponsored Brands prompts opened in beta in November 2025 and became generally available in the US on March 25, 2026. Amazon says campaigns are enrolled automatically and prompts use existing campaign targeting, billed as part of your CPC.
Your match types now decide where prompts appear as well as where ads appear. Our post on Amazon Sponsored Products and Sponsored Brands AI prompts covers the prompts report.
Keyword Groups: Amazon Picks the Keywords
Sponsored Products keyword groups target sets of related keywords chosen from your ASINs and "continuously refreshes them over time." Amazon describes them as "simplified broad-match keyword management."
They're in beta in the US only. You can mix keyword groups with normal keywords in an ad group, but not with product targets.
Our view: test keyword groups in their own campaign with a capped budget, like any discovery source. Harvest what they find into exact rather than letting them replace your exact campaigns.
Amazon Ads Agent Recommends Targets
On September 29, 2026, Amazon announced Amazon Ads Agent, which includes "AI targeting" recommendations across products, categories and keywords, and Sponsored Products recommendations for targeting, bids and budgets that apply in one click.
The recommendations still land as keywords and product targets with match types. Review the match type of each suggestion before you apply it, because a recommended broad keyword behaves like every other broad keyword.
Match Type Strategy by Campaign Stage
The right mix of match types changes as a product matures. This is the split we start from, then adjust by category and margin.

1. Launch (Weeks 1 to 4)
At launch you don't know which searches convert, so most of the budget buys information. We run an auto campaign with all four groups on, a broad campaign on 5 to 15 core keywords, and exact on brand terms and the two or three keywords we're certain about.
Amazon's targeting guide suggests running an automatic campaign for about two weeks before building manual campaigns from its search terms. Our product launch strategy covers the rest of the launch plan.
2. Growth (Months 2 to 6)
Now the search term report has data. Harvest terms with two or more orders under target ACOS (advertising cost of sales) into exact, and negate them as exact at the source.
Add phrase campaigns on the themes that keep appearing, such as "water bottle for kids" or "32 oz water bottle". Start product targeting on the ASINs your auto campaign converted on.
3. Mature (Month 6 Onward)
Most spend should sit in exact and product targeting, where you control the price of every proven search. Keep auto and broad running at lower budgets, because shopper language keeps changing and they are how you hear about it.
At this stage we also run exact on share of voice keywords, the head terms you need to be seen on, even when their ACOS is above target.
Stage | Main goal | Match types doing the work | What to watch |
|---|---|---|---|
Launch | Find converting searches | Auto (all groups), broad | Search terms, first orders |
Growth | Prove and harvest | Phrase, exact, product targeting | ACOS per term, harvest rate |
Mature | Scale and defend | Exact, product targeting | Impression share, TACOS |
Our campaign structure guide shows how to lay out these campaigns so each match type has its own budget.
How to Bid by Match Type
Amazon's targeting guide recommends "setting the highest bid on the exact match, a lower on the phrase match, and the lowest bid on the broad match," and testing for at least one to two weeks before pausing anything. We agree, and we set the gap from each match type's own conversion rate.
The formula we use for a starting bid is: target ACOS × average order value × conversion rate. It returns the most you can pay per click and still hit your target.

Here's the worked example in the image, with illustrative numbers:
A $30 product with a 30% target ACOS can spend $9 per order on ads.
Exact converts at 12%, so the bid is 0.30 × $30 × 12% = $1.08.
Phrase converts at 9%, so $0.81. Broad converts at 6%, so $0.54.
Do the same per auto targeting group. In this example close match at 8% gets $0.72, substitutes at 6% get $0.54, loose match at 5% gets $0.45 and complements at 4% get $0.36.
Use your own conversion rates once a match type has 20 or more clicks a week. Until then, start from the suggested bid and the ladder above. If you don't know your target ACOS yet, our break-even ACOS guide sets the ceiling.
Two more settings change what you actually pay per match type:
Bidding strategy. Amazon's dynamic bidding guide says "dynamic bids - up and down" can raise or lower bids by up to 100% based on likelihood to convert. We use it on exact and "down only" on broad and auto, where a raised bid on a loose match is the last thing we want.
Placement adjustments. Amazon's Sponsored Products best practices allow placement bid adjustments up to 900%. Exact terms usually earn a top of search boost; broad rarely does. Our placement modifiers guide has the method.
Our wider Amazon PPC bidding strategies post covers bid changes after launch.
Which Match Type Should You Use? A Decision Framework
We answer one question per keyword: how sure are we that this exact search converts for this product? The answer picks the match type.

The search has converted (two or more orders at or under target ACOS): add it as exact at your highest bid, and negate it as exact where it came from.
You know the theme, not the wording: use phrase. "Insulated water bottle" in phrase finds the sizes, colors and audiences you hadn't listed.
You're exploring a new idea or category: use broad or auto with a capped budget and a low bid.
The shopper is on a product page, not searching: use product targeting. Expanded to explore, individual exact to scale a proven ASIN.
The search can never convert: negative exact for one search, negative phrase for a concept.
Brand terms are the exception to all of this. We always run them in exact, often in their own campaign, so competitors' conquesting never decides what we pay for our own name.
Mistakes We See in Real Accounts
One ad group with every match type. The highest bid wins overlaps, so broad quietly takes exact's traffic. Separate them.
Same bid on broad and exact. Broad converts lower, so an equal bid overpays on every broad click.
Harvesting without negating at the source. The term then serves from two campaigns and your exact bid stops setting the price.
One default bid in auto. Complements and close match rarely convert alike. Bid per targeting group.
Leaving expanded product targets unreviewed. They update dynamically, so a target that converted in March can drift by June.
Judging broad by its keyword ACOS. A broad keyword at a fine ACOS can hide two great search terms and ten bad ones. Read the search terms.
Ignoring off-Amazon traffic. Inferred search terms from creator and social placements now show up in reports, at your max bid.
A full Amazon PPC audit checks for all of these, alongside budgets and structure.
How We Manage Match Types in Adbrew
Everything above works in the Amazon Ads console. The manual cost is the weekly loop: read search terms, harvest into exact, negate at the source, rebalance bids per match type. Across a portfolio of accounts it's the first job that slips.
Adbrew is an agentic retail media management platform for Amazon and Walmart, and that loop runs inside it on rules you write and can read:
Harvesting across match types. One rule moves a converting search term from auto, broad or phrase into exact and adds the negative exact at the source in the same step. Our keyword harvesting funnel guide shows the setup.
Bid rules by match type. Set different target ACOS, lookbacks and step sizes for exact, phrase, broad and each auto targeting group, so a broad keyword is never bid like an exact one.
Search terms and n-grams in one view. Review terms across every campaign in the Sponsored Ads management workspace and act in bulk.
Adbrew Intelligence. Ask Adbrew Intelligence for "broad keywords whose search terms spent over $50 with no orders this month", or launch exact campaigns for every harvested term from one prompt. Through our MCP server the same works from Claude or ChatGPT, and nothing applies until you approve it.
Every change lands in Change History with the rule that made it. Brands on Adbrew cut ACOS by 46% on average, and Seller Presto saves more than 250 hours of manual optimization a month.
Match Types Are How You Keep Control
Amazon now does more of the matching for you, through semantic broad match, inferred search terms, keyword groups and prompts. Match types, negatives and per-group bids are still where you decide what each search is worth.
Want harvesting and bid rules by match type running on your own account? Book a demo with our team and we'll walk through your search terms with you.
Frequently Asked Questions
What are the match types in Amazon PPC?
Amazon PPC has three keyword match types for Sponsored Products and Sponsored Brands: broad, phrase and exact. There are also two negative match types, negative exact and negative phrase. Automatic campaigns use four targeting groups instead (close match, loose match, substitutes, complements), and product targeting offers individual ASIN exact or expanded targets.
What is the difference between broad, phrase and exact match on Amazon?
Exact match serves on the keyword and close variants only. Phrase match serves on any search containing the keyword in the same order, with words before or after it. Broad match serves on searches in any order, including synonyms and related terms, and Amazon says the keyword itself may not appear in the search.
Which Amazon match type is best?
No single Amazon match type is best. Exact usually has the highest conversion rate and suits proven searches and brand terms. Phrase suits proving themes and long-tail terms. Broad and auto suit discovery. Most mature accounts put most spend in exact and product targeting and keep broad and auto running at lower budgets.
Does Amazon still have broad match modifier?
Yes, but only in Sponsored Brands. Adding "+" before a word in a Sponsored Brands broad keyword, such as "+kids shoes", forces that word to appear in the shopper's search. Sponsored Products does not support broad match modifiers, so use phrase match there when a word must be present.
What are close match, loose match, substitutes and complements?
They are the four Amazon auto targeting groups in Sponsored Products. Close match and loose match serve on searches closely or loosely related to your product. Substitutes serve on detail pages of similar products, and complements on detail pages of products used alongside yours. Each group can be switched off and bid separately.
What is expanded product targeting on Amazon?
Expanded product targeting is a Sponsored Products option that targets a chosen ASIN plus related products Amazon selects, including substitutes and complements, updated dynamically. Individual exact product targeting serves only on the chosen ASIN. Use expanded to discover converting product pages and exact to scale the proven ones.
Can you change the match type of a keyword on Amazon?
No. Amazon does not let you change a keyword's match type after the campaign is created. Add the same keyword again with the new match type, then pause or archive the old one. Negative keywords work the same way: archive the negative and add it again with the other negative match type.
How should I bid by match type on Amazon?
Bid highest on exact, lower on phrase and lowest on broad, as Amazon's own targeting guide recommends. Set each bid from target ACOS times average order value times that match type's conversion rate. At a 30% target, $30 order value and 12% conversion rate, an exact bid of $1.08 hits the target.


