Amazon PPC match types set how closely a shopper's search must resemble your keyword for a Sponsored Products ad to show. Broad match allows any word order plus synonyms and related terms. Phrase match needs your words in order, with extra words around them. Exact match needs the same words in the same order, plus plurals. Broad and phrase discover new search terms; exact runs the proven ones.
Match types are one of the first things I read in a new account, because they decide which searches it pays for before any bid is argued. The fault is rarely the wrong type. It is three types with no assigned jobs, chasing the same searches at different prices. This guide covers what each type catches, per Amazon's own definitions, when I use each, whether they need separate campaigns, how they feed the harvest loop, a worked example on real numbers, and the overlap that undoes all of it.
Broad vs phrase vs exact: how Amazon's match types compare
The three Sponsored Products match types trade reach for control. Broad reaches the most searches and controls the least, exact reaches the fewest and controls the most, and phrase sits between them. That trade decides each type's job: broad and phrase find searches you did not know about, and exact runs the searches you have already proven.
| Match type | What it matches | Job in the account | Typical bid posture |
|---|---|---|---|
| Broad | Your words in any order, plus plurals, synonyms, and related terms; can match searches that do not contain your keyword | Discovery: wording and adjacent searches you would not have guessed | Lowest. Down-only bidding on a small research budget |
| Phrase | All your words in your order, with extra words before or after, plus the plural | Controlled discovery: the long tail around a term you trust | Middle. Down only, in the research campaign |
| Exact | The same words in the same order, plus the plural | Performance: proven search terms at a precise bid | Highest: the term's revenue per click × target ACOS, with up-and-down bidding only once it has orders |
Amazon's targeting guide recommends the same ladder: highest bid on exact, lower on phrase, lowest on broad. It also calls exact the most restrictive type and the likeliest to convert. Each step toward broad adds searches further from your listing, and a click that converts less is worth less.
What each match type catches, on one keyword
The fastest way to learn match types is to run one keyword through all three. Take insulated lunch bag. Exact catches that search and its plural. Phrase adds searches containing those words in that order. Broad adds reordered versions and can reach related searches that never contain the keyword.
| Shopper's search | Broad | Phrase | Exact |
|---|---|---|---|
| insulated lunch bag | Yes | Yes | Yes |
| insulated lunch bags | Yes | Yes | Yes (plural) |
| large insulated lunch bag for men | Yes | Yes | No |
| lunch bag insulated | Yes | No: word order | No |
| insulated bag for lunch | Yes | No: word order | No |
| cooler lunch box | Possibly: Amazon's call | No | No |
The last row is where broad earns its budget and where it wastes it: Amazon judges relatedness from your keyword's meaning and the advertised product, so the call is Amazon's, not yours. The table follows Amazon's published definitions, but the final word is your search term report, where the keyword column and the search term column part ways and you see what each type actually bought.
When to use broad, phrase, and exact match
Use broad match to discover how shoppers actually describe your product, phrase match to mine variations around a term you already trust, and exact match to run search terms that have proven they convert. Most products need all three running at once, each in its own lane, rather than a choice between them.
When broad match earns its budget
Broad match earns its budget when you do not yet know your buyers' language: a new product, a new category, or a seed list built on guesses. It finds wording you would never have typed, and because it can also wander, it gets the lowest bid and the tightest leash of the three.
The wandering is literal: broad is the one type that can leave your keyword behind. Picture a lunch bag keyword drifting to lunch boxes, then bento boxes, then kids' backpacks, one plausible hop at a time. I find that drift wherever broad goes unread. So broad runs on down-only bidding at the bottom of the ladder, and its search terms get read on a schedule. N-gram analysis catches words failing across dozens of searches at once, and the free N-Gram Analyzer runs it on your own file.
Amazon phrase match: when it is the better tool
Phrase match is the better tool once a head term converts and you want the long tail around it without broad's wandering. Every word must appear in your order, so it holds relevance on multi-word keywords while still catching the modifiers shoppers add: sizes, materials, audiences, use cases.
Phrase is only as tight as the words you hand it: a one-word phrase keyword qualifies for every search containing that word. Word order is the other trap. Shoppers do type lunch bag insulated, and a phrase keyword on insulated lunch bag will never see it. If a reordered version converts in broad, give it a keyword of its own.
When exact match is the right home
Exact match is the right home for a search term that has already proven itself with orders. It is the tightest control Amazon offers, one search, one bid, one clean row of data, which lets the bid be priced from the term's own economics instead of an average across hundreds of searches.
Price it from its own revenue per click times target ACOS, and save up-and-down dynamic bidding for terms with an order history. Two cautions. Exact is not perfectly literal: Amazon's definition includes the plural. And launching on exact alone bets that you already know how buyers search. For your brand terms you do, so they start in an exact match brand campaign. For everything else, pair exact with discovery from day one; my keyword research guide covers building the seed list.
Should broad, phrase, and exact be in separate campaigns?
Exact match belongs in its own campaign, apart from broad and phrase, because budgets, bidding strategy, and placement adjustments are set at the campaign level, so match types sharing a campaign share those settings. Broad wants a small budget and down-only bidding; proven exact terms can justify a bigger budget and up-and-down bidding.
The second reason is legibility. A mixed campaign reports one blended ACOS, and a blend cannot say which lane is losing money; in audits I often find exact's efficiency quietly paying for broad's exploration, with nobody able to say by how much. My working minimum is exact in its own performance campaign and broad plus phrase sharing a research campaign, split further when broad needs a tighter budget. The wider discovery-to-performance campaign structure has its own guide, and the free Campaign Builder generates the split as an uploadable bulk file, ownership negatives included.
How match types feed the harvest loop
Match types feed each other through the search term report. Broad and phrase keywords surface real shopper searches; the ones that earn orders get promoted into exact match with their own bid; and the discovery campaign that found them receives a negative exact so it stops buying them. Discovery keeps exploring, exact keeps the proven terms.
My graduation bar has three parts: two or more orders on the term, orders credited to the advertised product rather than a sibling, and a look at the live results page to confirm the term describes what you sell. A term that clears it goes into exact with its own bid, and the source campaign gets the ownership negative. Which negative type, at which level, is covered in my negative keywords guide, and the free Negative Keyword Finder builds the upload file.
A graduate can also seed more discovery: if insulated lunch bag for men converts and heads a family of its own, run it in exact and mine its variants in phrase. Just give that phrase keyword a negative exact on the term exact owns, or you have built the overlap below on purpose.
A worked example: one search term from broad to exact, on real numbers
On the sample dataset I use across this hub, a search term found by broad match and converting at the account average needs about 22 clicks, roughly $17 of spend, to clear a two-order harvest bar, then enters exact at a bid near $0.73 on a 30% target ACOS. Here is where each of those numbers comes from.
The dataset: 45,672 clicks bought 4,140 orders and $111,058 in ad sales for $36,303 of spend. Divide it out and conversion runs at 9.1%, clicks cost $0.79 on average, ACOS sits at 32.7%, and each click returns $2.43 in revenue.
| Stage | The arithmetic | Result |
|---|---|---|
| Fair trial for one order | 1 ÷ 9.1% conversion rate | About 11 clicks |
| Two-order harvest bar | 2 orders × 11 clicks × $0.79 CPC | About 22 clicks, roughly $17 of broad match spend |
| Exact match bid at promotion | $2.43 revenue per click × 30% target ACOS | About $0.73 |
| Source campaign | Negative exact on the promoted term | Broad stops re-buying a term exact now owns |
Every graduate costs about $17 of discovery spend, which is why broad sits at the bottom of the ladder: a cent off each discovery click makes every future graduate cheaper. And a $0.79 average CPC against a fair price near $0.73 means clicks run about six cents above what they earn at a 30% target. With match types in separate campaigns, finding the lane that overpays is a sort; in a mixed campaign, those six cents are smeared across every keyword. The free Audit Dashboard breaks your search term report down by match type, and the demo shows that view on a real anonymized account.
The classic failure: several match types buying one search
The match type failure I find most often is not a wrong choice of type. It is one search term bought by broad, phrase, and exact keywords at the same time, with none of them named the owner. I call it search term bleed: the term's data splinters, and no row ever holds enough of it to act on.
Picture it. Exact insulated lunch bag sits in the performance campaign; phrase insulated lunch bag and broad lunch bag sit in research. The search insulated lunch bag qualifies for all three. Amazon enters only one of your bids in the auction, so you are not bidding against yourself, but the term's clicks and orders now land on up to three rows. Split the 22-click harvest bar three ways and each row gets about seven clicks, short of even the 11-click trial for one order. The term never earns a verdict, and no row looks bad enough to flag.
Bleed was on the diagnosis list for the health and personal care brand on my results page, next to keywords that looked busy but never converted and bids that rewarded impressions over orders. Over 90 days, with the budget unchanged, ACOS went from 39.66% to 27.02%; the fixes landed together, so I will not assign bleed its share. The cure is ownership, and my search term bleed guide walks the full diagnosis.
Frequently asked questions
How many keyword match types does Amazon PPC have?
Three, for Sponsored Products keywords: broad, phrase, and exact, from the most reach to the most control. Negatives are a separate system with two types, negative exact and negative phrase. Auto campaigns use no match types; they target through four groups, close match, loose match, substitutes, and complements, with no keywords written by you.
What is the difference between phrase match and broad match on Amazon?
Word order and reach. Phrase match requires every word of your keyword in your order, with extra words allowed before or after. Broad match accepts the words in any order and reaches synonyms and related searches, even ones without your keyword. Use phrase to mine variants of a term you trust, and broad to find buyer language you have not seen yet.
Amazon exact match vs broad match: which should get the higher bid?
Exact, almost always. Amazon's own targeting guide recommends the highest bid on exact, lower on phrase, and lowest on broad, and the economics agree: an exact keyword runs a search that has already proven it converts. I still price each bid from the keyword's own revenue per click times target ACOS, so the ladder is a sanity check, not the formula.
Does Amazon exact match include plurals?
Yes. Amazon defines exact match as the same words in the same order as your keyword, plus the plural form, so an exact keyword for insulated lunch bag can serve on insulated lunch bags. The search term report is the final word on what any keyword actually served on, so read it rather than trusting the label alone.
Should I start a new keyword in broad or exact match?
Start it where the evidence is. A keyword with no order history belongs in a research campaign as broad or phrase, bid low on down-only bidding, so it can show you which searches convert. A specific search term earns exact once it clears the harvest bar: two or more orders on the advertised product. Brand terms are the exception, exact from day one.
Is broad match worth it on Amazon?
Yes, when it is bid low and read on a schedule. Broad finds wording you would never have typed, and that discovery feeds every exact keyword you will own later. Unread, it turns into waste. My benchmarks tolerate wasted spend of up to 50–60% in the awareness phase but under 15% in profitability, so broad's leash should tighten as the account matures.
Sources
- A guide to targeting with Sponsored Products, Amazon Ads
See what your match types are actually buying
Your bulk file already knows which of these patterns you have. The free Account Health Snapshot scores it on eight metrics against your phase, wasted spend and auto campaign share among them. No email, no account, nothing stored. And if the file shows broad, phrase, and exact chasing the same searches, settle who owns what on a free 30-minute diagnosis call with me, keyword by keyword.