What is the difference between a broad match and a phrase match negative keyword?

A negative broad match keyword blocks a search only when the query contains all of your negative terms, in any order, with other words allowed around them. A negative phrase match keyword blocks a search only when it contains your exact terms in the exact order you wrote them, again with other words allowed around the string. The key practical difference is order and completeness: a broad negative of "cheap shoes" blocks "shoes that are cheap" because both words are present; a phrase negative of "cheap shoes" does not, because the words are not adjacent and in order. Note that unlike positive keywords, negative keywords have no true broad-match expansion — they never match synonyms, close variants, plurals, or misspellings, so both negative types are more literal than their positive namesakes.

When should I use a broad match negative keyword?

Use a broad match negative when you want to block an entire concept regardless of how the searcher phrases it. If you never sell used goods, a broad negative of "used" blocks "used bikes", "bikes used", and "buy used mountain bike" in one entry, now and for any future query containing the word. Broad negatives are the right tool for universal disqualifiers — words like free, cheap, jobs, salary, DIY, or a product category you do not carry — because the concept is unwanted no matter what surrounds it. The risk is that a single-word broad negative also blocks any legitimate query that happens to contain that word, so reserve broad negatives for tokens that cannot appear in a search you want to keep.

When should I use a phrase match negative keyword?

Use a phrase match negative when the words you are blocking could legitimately appear in queries you want to keep, so you need to block only the specific ordered string. If you sell some products from a brand but not one particular model, a phrase negative of "intex 2000" blocks searches for that model while leaving other brand searches eligible; a broad negative of the two words would over-block. Phrase negatives are the surgical option: they respect word order and adjacency, so they remove a precise phrase without disqualifying the surrounding traffic. The trade-off is coverage — a phrase negative will not catch the same idea written in a different word order, so you accept that a reworded version of the query can still slip through until you add it.

Why does my broad match negative keyword block too much traffic?

Because a broad negative fires whenever all its words appear anywhere in the query, so a multi-word broad negative blocks every search containing that combination of words in any arrangement — including searches you want. A broad negative of "running shoes" will also block "running shoes for flat feet" and "best trail running shoes", which may be exactly the converting traffic you meant to keep. The usual cause of silent over-blocking is treating a broad negative as if it targeted a specific phrase. If the words in your negative can plausibly appear in a good query, switch it to phrase match so order and adjacency are required, or narrow the negative to the single disqualifying token rather than the whole phrase.

Should account-level negative lists use broad or phrase match?

Mostly broad, but only for tokens that are universally disqualifying. Account-level and shared negative lists exist to block concepts that are wrong across every campaign — other languages, plainly irrelevant categories, free and job-seeking intent — and broad match is the efficient way to catch those regardless of phrasing. The danger is scope: a broad negative on a shared list applies everywhere, so an over-broad token can quietly suppress good traffic across the whole account with no obvious symptom. Keep shared-list broad negatives to words that could never appear in a query you want, and push anything phrase-specific or campaign-specific down to the campaign level where its blast radius is contained.

Broad vs Phrase Negative Keywords: When Each Fits

Every time you add a negative keyword you make a match-type decision, and the two everyday choices — broad and phrase — block traffic in genuinely different shapes. A negative broad match keyword blocks a search when the query contains all of your terms in any order; a negative phrase match keyword blocks only when the query contains your exact terms in the exact order you wrote them. Get that choice right and one entry cleanly removes the waste you meant to remove. Get it wrong and you either leave the same junk running under a slightly different phrasing, or you silently disqualify the converting traffic sitting next to it.

This is a per-term decision, not an account-wide setting, and it is worth making deliberately rather than defaulting to whichever match type your tool pre-selects. The rule that resolves almost every case is a single question about the words in your negative: could they appear in a search you actually want? This post works through how each negative match type fires, when each is correct, the over-blocking failure mode that makes broad negatives dangerous, and how the choice changes when the negative lives on a shared list versus a single campaign. Google's own reference on how negative keyword match types work is the anchor throughout.

How broad and phrase negatives actually fire

A negative broad match keyword blocks a query when every word in the negative appears somewhere in the search, regardless of order and regardless of what else is present. A broad negative of cheap shoes blocks "cheap shoes", "shoes that are cheap", and "cheap leather shoes" alike, because in each case both words are present. A negative phrase match keyword is stricter: it blocks only when the query contains your exact terms in the exact order, as a contiguous string. A phrase negative of cheap shoes blocks "buy cheap shoes online" but not "shoes that are cheap", because the two words are neither adjacent nor in order there. Order and adjacency are the whole difference: broad ignores both, phrase requires both.

One property trips up practitioners who reason from positive keywords: negative keywords have no true broad expansion. Unlike a positive broad match keyword, a negative never matches synonyms, close variants, plurals, or misspellings — if you block "shoe" you do not automatically block "shoes", and if you block "sneaker" you do not block "trainers". Both negative match types are more literal than their positive namesakes, which is why plurals and misspellings need their own handling; that specific behaviour is covered in negative keyword match types, plurals, and close variants. For the broad-versus-phrase decision, the thing to hold onto is simpler: broad is about presence, phrase is about the exact ordered string, and neither reaches beyond the literal words you typed.

When broad is right: blocking a whole concept

Reach for a broad negative when you want to disqualify an entire concept no matter how it is phrased. The cleanest examples are universal intent modifiers: if you never sell used goods, a broad negative of used blocks "used bikes", "bikes used", and "buy a used mountain bike" in a single entry, and it keeps blocking any future query that contains the word. The same logic applies to free, cheap, jobs, salary, DIY, and to a product category you simply do not carry. The value of broad here is coverage over time: you are not chasing every wording of an unwanted idea, you are blocking the idea itself, once.

The condition that makes this safe is that the word cannot plausibly appear in a query you want to keep. As Search Engine Journal frames the same call, a broad negative is right when you want to completely block a concept — like a jewelry store that does not sell watches adding "watches" as a broad negative. That test does the work: "watches" for a store that will never sell one is pure disqualifier, so broad is correct and efficient. This is the tool you use when building the universal layer of a negative list — the recurring junk tokens rather than one-off terms — which is the same instinct behind grouping search-term patterns instead of negating them one by one. Block the token that recurs, and let broad match handle every phrasing it will ever appear in.

When phrase is right: the surgical block

Reach for a phrase negative when the words you want to block could legitimately show up in queries you want to keep, so you need to remove only the specific ordered string. The canonical case is a brand you partly stock: if you sell most of a brand's range but not one model, a phrase negative of that model — say intex 2000 — blocks searches for it while leaving the rest of the brand's queries eligible. A broad negative of those two words would over-reach, catching other searches that happen to contain both. Phrase respects word order and adjacency, so it excises a precise phrase without touching the traffic around it.

The trade-off you accept with phrase is coverage. Because it requires the exact order, a phrase negative will not catch the same idea written differently — block "cheap running shoes" as a phrase and "running shoes cheap" still gets through until you add it too. That is the deliberate cost of precision: you are choosing to under-block rather than risk suppressing good queries. Phrase is therefore the default whenever a word in your intended negative is ambiguous — a word that is disqualifying in one context and perfectly fine in another. When you find yourself hesitating because a term "usually" means junk but sometimes does not, that hesitation is the signal to use phrase and keep the block tight. It is also how you avoid the failure mode where a negative quietly blocks your own active keywords by overlapping with the very terms you bid on.

The over-blocking trap that makes broad dangerous

The single most expensive mistake with negative match types is using a multi-word broad negative as if it targeted a specific phrase. Because a broad negative fires whenever all its words appear in any order, a broad negative of running shoes blocks "running shoes for flat feet", "best trail running shoes", and "waterproof running shoes for women" — very likely the converting traffic you meant to protect. Nothing in the interface warns you; the impressions simply stop, and unless you are watching that segment you may never connect the drop to the negative you added weeks earlier. This is why broad negatives are the usual culprit behind over-negating that throttles good traffic.

The Store Growers guidance states the boundary well: choose too narrow and the same irrelevant search returns tomorrow with one extra word; choose too broad and you might accidentally block searches you actually want. The resolution is the per-term test again: if the individual words in the query could plausibly appear in searches you want to keep, a broad negative is the wrong tool — use phrase so order is required, or narrow the negative to the one disqualifying token rather than the whole phrase. In practice, most costly over-blocks come from broad negatives that are two or three words long; a broad negative is safest as a single unambiguous word, and grows riskier with every word you add to it. Before you save any broad negative, read it back and ask which good query it might also catch — and confirm it later against the search terms report to make sure the traffic that disappeared was the traffic you intended to remove.

Match type and scope: the shared-list multiplier

The broad-versus-phrase choice interacts with where the negative lives, and the interaction raises the stakes on broad. A negative on an account-level shared list applies to every campaign that uses the list, so an over-broad token there suppresses traffic across the whole account at once — the same over-blocking risk as above, multiplied by your campaign count and much harder to trace back to a cause. Shared lists are the right home for broad negatives only when the token is a universal disqualifier that could never appear in a query you want anywhere in the account: other languages, plainly irrelevant categories, free and job-seeking intent. Those earn broad because they are wrong everywhere by definition.

Anything narrower belongs lower down. A phrase negative that fences one campaign's specific terms, or a block that is only true for one product line, should sit at the campaign level where its blast radius is contained — putting it on a shared list applies it everywhere and can break deliberate structures like priority tiering or query fencing. This is the match-type dimension of the broader account, campaign, and shared-list structure decision: universal broad negatives high, specific phrase negatives low. The two decisions are really one — how wide the block should reach in the query, and how wide it should reach across the account — and answering them together is what keeps a negative list precise as it grows, rather than accumulating broad-match landmines you rediscover only when a campaign quietly stops converting.

A per-term workflow you can apply from the report

Turn all of this into a habit you run straight from the search terms report. When a wasteful query appears, ask first whether the concept is unwanted no matter how it is phrased. If yes, block the disqualifying token as a broad negative — and prefer a single word over a multi-word broad, because each extra word widens the blast radius. If the words could appear in a good query, use a phrase negative on the exact string instead, accepting that a reworded version may need adding later. When the term is genuinely one-of-a-kind and no variant is likely, exact match (the third negative type) is fine, but for the everyday broad-or-phrase call this two-step test settles almost everything.

Then confirm rather than assume. After adding a broad negative, watch the affected campaign's impressions and the search terms report over the next cycle: if volume you wanted disappeared, your negative was too broad and should be narrowed to phrase or to a single token. Fold the check into the cadence you already use to decide when a term has earned a negative in the first place, and into the broader search hygiene audit. The match type is not a detail you set and forget — it is the difference between a negative that removes exactly the waste you targeted and one that takes good traffic down with it.

For how plurals, misspellings, and close variants behave under negative matching, see negative keyword match types, plurals, and close variants. For the volume side of the same risk, see over-negating and throttling good traffic, and for where negatives belong across the account, see negative keyword list structure across account, campaign, and shared.