PPCleaner

Phrase Match Negative Keyword Examples You Can Copy Safely

Published November 5, 202612 min read
An ecommerce owner and marketer reviewing search-query cards beside products and customer orders.
The short answer

A phrase match negative blocks any search that contains its words, in that order, with anything before or after them. That makes a two-word phrase negative fairly narrow. A one-word phrase negative behaves exactly like a one-word broad negative. Either way, the phrase will also match where it sits inside a longer phrase you actually sell: "dog food" blocks "stainless steel dog food bowl".

Before you add a phrase negative, search your search terms report for it and read every match, converting terms first. If a match converted, make the negative longer and more specific, or switch to exact. The examples below show what each negative blocks, what it kills by accident, and a safer version.

What a phrase match negative actually blocks

A phrase match negative blocks a search when the search contains the negative's words, in the same order, next to each other. Extra words before or after don't save it. The negative "plumber salary" blocks "plumber salary texas" and "average plumber salary". It does not block "salary for a plumber", because the words are no longer side by side in that order.

That's the version every guide gives you, and it's accurate. What those guides leave out is how it plays out in a real account. They show phrase match as the careful middle option: tighter than broad, looser than exact. That only holds when the negative has two or more words. And even then it says nothing about where in the search the phrase lands.

Three things the usual explanation leaves out:

  • A one-word phrase negative is a broad negative. With one word there's no order to respect. "free" and free (broad) block exactly the same searches: anything containing the word "free". Picking phrase match for a single word changes nothing.
  • The phrase doesn't care what role it plays in the search. "coffee beans" blocks "buy coffee beans", which you wanted. It also blocks "burr grinder for coffee beans", where "coffee beans" is just describing what the grinder is for.
  • Word order protects some searches you wanted to block. "plumber salary" lets "salary of a plumber" through. So the same negative can block too much and too little at once.

Google's documentation also says negative keywords don't match close variants. Your positive keywords match plurals and misspellings. Your negatives mostly don't. "plumber salary" won't reliably block "plumbers salaries". This is the reverse of how your positive keywords behave, which we cover in close variants and why exact isn't exact.

Example 1: the product inside the phrase

The most common overblock we see is a phrase negative that names a product the advertiser doesn't sell. That same product is part of how buyers describe what the advertiser does sell.

Take a store that sells pet bowls and feeders, not pet food. The search terms report is full of food searches, so someone adds the phrase negative "dog food". Here are 30 days of search terms that contained that phrase. The figures are made up but typical:

Search termClicksCostConversions
dog food brands40$360
best dry dog food25$220
stainless steel dog food bowl18$273
raised dog food bowl large12$192
Total95$1045

After "dog food": all four terms are blocked. The account saves $104 a month and loses 5 conversions. Only $58 of that $104 was waste. The other $46 was buying the store's best customers.

Safer version: block the food intent, not the noun phrase.

  • "dog food brands"
  • "dry dog food"
  • "wet dog food"
  • "dog food recall"

After the safer set: "dog food brands" and "best dry dog food" are blocked, saving $58. Both bowl searches keep running. You'll need to add new food searches as they show up, such as "grain free dog food". That's the price of not blocking buyers, and it's worth paying.

The same pattern shows up anywhere a product is described by what it holds, fits or goes with:

  • Phone case seller: "iphone 15" blocks "iphone 15 case". Block "iphone 15 price" or "iphone 15 trade in" instead.
  • Grinder seller: "coffee beans" blocks "grinder for coffee beans". Block "coffee beans online" or "coffee beans bulk".
  • Mattress protector brand: "memory foam mattress" blocks "memory foam mattress protector". Block "memory foam mattress sale", but only after checking it.
  • Car seat cover shop: "car seat" blocks "car seat covers for leather seats". Block "infant car seat" and "booster car seat".

The test is simple. Ask whether a buyer of your product would ever type this phrase as part of a longer search. If yes, the phrase is too short to be a negative on its own.

Example 2: one-word negatives that look harmless

One-word negatives cause the widest damage, because they match anywhere in the search. Choosing phrase match doesn't narrow them at all.

A project management software company with a free trial adds "free" because the report shows "free project management app" and "free gantt chart template". Here's what the negative catches. Again, the figures are illustrative:

Search termClicksCostConversions
free gantt chart template60$900
free project management app35$701
project management software free trial20$604
free trial asana alternative10$352
Total125$2557

After "free": $255 a month saved, 7 conversions lost. The two trial searches, which cost $95 and brought 6 trial sign-ups, are gone. Those are the searches the trial exists for.

Safer version:

  • "free template"
  • "gantt chart template"
  • "free app", but only if the free plan really isn't what you sell
  • "open source"

After: the template search is blocked ($90 saved). "free project management app" stays live, because it brought a conversion and needs more data before you decide. Both trial searches keep running.

Other one-word negatives that do more damage than people expect:

  • "cheap" for a mid-priced service. "cheap" is sometimes just how people say "fair price". "cheap emergency plumber" often converts at full price, because it's still an emergency.
  • "used" for a new car dealer. This also blocks "trade in used car for new", which is a buyer.
  • "repair" for an appliance retailer. This also blocks "repair or replace water heater", where half the people are leaning toward replacing.
  • "kids" for an adult orthodontist. This also blocks "orthodontist for adults and kids", which may be a parent booking for the whole family.
  • "jobs" for a recruiting agency's employer campaign. This also blocks "post jobs", "fill warehouse jobs fast" and "jobs board for employers".

"jobs" is a good word to negate for a plumber and a bad one for a staffing firm. The word isn't safe or unsafe on its own. It depends on what you sell. That's why copying someone else's list of one-word negatives is dangerous. Service businesses run into this all the time, and we break it down by intent in negative keywords for service businesses.

Example 3: "how to" and other research phrases

Research phrases are the classic phrase match negative. They're also where intent varies most within a single phrase. "how to" blocks people trying to do the job themselves. It also blocks people working out how to hire someone.

A roofing company adds "how to". Searches that contain it:

  • "how to fix a leaking roof": DIY, block it
  • "how to patch roof shingles": DIY, block it
  • "how to choose a roofing contractor": a buyer, comparing
  • "how to get roof replaced by insurance": a buyer with a claim, often a big job
  • "how to find a roofer near me": a buyer

After "how to": the two DIY searches are gone, and so are three of the most commercial searches in the account.

Safer version: pair the research phrase with the DIY verb, not with the topic.

  • "how to fix"
  • "how to patch"
  • "how to repair"
  • "how to replace", after checking it. "how to replace roof cost" can be a buyer.
  • "diy" (single word, and fine here, because "diy" never means "hire someone")

The same applies to "what is" (blocks "what is the best roofer in Austin"), "tutorial" (fine for most), "course" (bad for anyone selling training), and "pdf" (usually safe, check anyway).

Example 4: when word order lets waste through

Word order works in the other direction too. A two-word phrase negative only blocks searches where the words appear together and in that order. So the waste often comes back in a different order.

An HVAC company blocks "technician salary". A month later the report shows:

  • "hvac technician salary": blocked
  • "salary for hvac technician": not blocked
  • "how much does an hvac technician make": not blocked
  • "hvac tech salary": not blocked (different word)
  • "technician salaries hvac": not blocked (plural, and negatives don't reliably match variants)

There are two ways to fix it, and each has a cost:

  1. Add more phrase negatives: "salary", "how much does"… wait. "how much does" also blocks "how much does ac repair cost", which is a buyer. Stick with "salary", "salaries", "tech make", "technician make". This takes more upkeep, but you know exactly what each one blocks.
  2. Use a broad negative: technician salary (broad) blocks any search containing both words in any order. It's more compact, but broad negatives are harder to predict across your keyword set, and harder to audit later. Someone reading the list six months from now can't easily tell what it covers.

For a word like "salary", which a buyer of HVAC repair never types, the single word "salary" is the simplest answer. It's a one-word negative, but a safe one, because no version of the word means "I want to hire you". That's the real rule for one-word negatives: the word has to be wrong in every search it could ever appear in.

A copy list, with the check each one needs

None of these is safe everywhere. Each one comes with what it blocks, what it can also block, and a narrower fallback. Read the "also blocks" column against your own offer before you paste.

Phrase negativeBlocksCan also blockNarrower fallback
"salary"plumber salary, hvac salarynothing, for most service businessesnone needed
"jobs"plumbing jobs, roofer jobs"post jobs", employer searches"jobs near me", "jobs hiring"
"free"free template, free download"free trial", "free shipping", "free estimate", "gluten free""free template", "free download"
"how to"how to fix a leakhow to choose a contractor"how to fix", "how to repair"
"diy"diy roof patchrarely anythingnone needed
"cheap"cheap knockoffcheap emergency locksmith (buyers)"cheapest", check first
"used"used laptops (if you sell new)trade in used, most used"used for sale", "refurbished"
"near me"out-of-area intent? No.most local buyersdon't use as a negative
"reddit"reddit best crmrarely anythingnone needed
"login"brand loginrarely anything, unless you sell login softwarenone needed
"[competitor] careers"competitor careersnothingnone needed
"[competitor]"competitor brand searches"[competitor] alternative", a buyer"[competitor] login", "[competitor] support"

The "near me" row is on the list because it gets copied around as a "low intent" negative. It isn't. For local services it's usually the highest-intent wording in the account. If you're seeing location waste, the fix is in your location settings and the place names in your search terms, not in the words "near me". We go through that in wrong location clicks.

How to test a phrase negative before it goes live

Search your own search terms for the exact phrase, over the last 90 days, and read every match before you add the negative. It takes two minutes and catches almost every case above.

  1. Open the search terms report for the campaigns the negative will apply to. Set the date range to 90 days.
  2. Filter search term contains the phrase. Contains filters on the characters you type, so it will over-match a little. That's fine. "free" will also show "freelance", and you want to see that.
  3. Sort by conversions first, then by cost. Any match with a conversion is a reason to make the negative longer. If it's a single conversion on a small search, that's a judgment call. If it's several, don't block.
  4. Read the zero-click, low-volume matches too. They show you what the phrase will catch next month. We explain why they matter in low volume terms add up.
  5. Check the negative against your own keywords. A phrase negative that contains a keyword you bid on will quietly stop that keyword from serving. Google Ads won't warn you about it everywhere. See negative keyword conflicts Google Ads won't flag. If a keyword has stopped getting impressions since you added negatives, start with this check.

There's a limit to this test. You can only check against searches you can see. Some searches are hidden from the report for privacy reasons, and your negative will still apply to them. A phrase that looks clean in the visible terms may block a converting search you never saw. That's one more reason to prefer longer phrases: the longer the negative, the less it can catch that you couldn't see. More on that in search terms hidden by privacy.

Where phrase negatives apply, and where the list lives

Phrase match negatives work in Search, Standard Shopping and Performance Max. Where you can put them differs, and that changes how careful you need to be.

  • Search and Standard Shopping accept negatives at campaign and ad group level, plus shared negative keyword lists. A shared list applies everywhere it's attached. One overblocking phrase in a shared list hurts every campaign at once. Look at a shared list's phrases more strictly than a single campaign's.
  • Performance Max accepts up to 10,000 campaign-level negative keywords. It doesn't accept shared negative keyword lists, so the same phrase has to be added campaign by campaign. Teams often bulk-paste a Search list into PMax without re-checking it against PMax's own search terms, which can look quite different. Why PMax negatives sometimes seem not to work is covered in Performance Max negative keywords not working. How to decide what goes at account level and what goes per campaign is in account-level vs campaign-level lists.
  • Display, Video and App don't report search terms, so you can't run the test above for them at all.

Search, Shopping and Performance Max search terms are all reported in campaign_search_term_view. That's what makes the 90-day check possible across all three, whether you do it by hand or with a tool.

What PPCleaner does here, and what it doesn't

PPCleaner reads the search terms your account paid for each night, across Search, Shopping and Performance Max. It judges which ones were never going to convert and adds them as negatives. It never blocks a term that converted in the last 90 days. The "stainless steel dog food bowl" search from Example 1 would stay untouched while it keeps converting. Every block can be undone in one click for 90 days.

What it can't do is make your intent decisions for you. Whether "free project management app" is waste depends on whether your free plan brings in paying customers later, and that may not show up as a conversion in Google Ads. If your conversion tracking misses real sales, any system that protects converting terms is working from incomplete facts. That includes PPCleaner and that includes you. Fix the tracking first. And PPCleaner never touches budgets, bids, ads or keywords, so if the real problem is a keyword pulling in the wrong searches, the negative list is only a patch.

If you'd rather build the list by hand, the method that avoids guessing is in add negative keywords from search terms without guessing. For a faster weekly pass through the report, see how to analyze a search terms report in 20 minutes.

negative keywordsphrase matchsearch termsmatch types

Frequently asked

Is a one-word phrase match negative different from a one-word broad negative?

No. With a single word there's no word order to respect, so `"free"` and `free` block the same searches: any search containing that word. Match type only changes what a negative blocks once it has two or more words.

Does a phrase match negative block plurals and misspellings?

Not reliably. Google's documentation says negative keywords don't match close variants the way positive keywords do. If "plumber salary" and "plumbers salaries" both show up in your report, add both, or use a single word like `"salary"` if it's wrong in every search it could appear in.

Should I use phrase or exact match for negatives?

Use exact when one specific search is waste and similar searches might convert, for example blocking "free project management app" while keeping "project management software free trial". Use phrase when every search containing the words is waste, like `"dog food brands"` for a bowl seller. If you aren't sure, make the phrase longer before you fall back to exact.

Can I copy a negative keyword list from another account?

You can copy the format, not the words. `"jobs"` is a good negative for a plumber and a harmful one for a staffing agency. Before adding any copied phrase, search your own last 90 days of search terms for it and check whether any match converted.

Do phrase match negatives work in Performance Max?

Yes. Performance Max accepts up to 10,000 campaign-level negative keywords, but not shared negative keyword lists. You add them per campaign, and you should check them against that campaign's own search terms, because they often look different from Search.

About the author

Alain Abulafya

Founder of PPCleaner · Google Ads consultant & media buyer

Alain has worked in digital marketing for more than 15 years, over 10 of them managing and optimizing Google Ads campaigns. He has worked with more than 100 advertisers in e-commerce and lead generation, and he is a Google Partner. He built PPCleaner to automate the search term reviews he used to do by hand for his clients.

Stop the leak

Clean your search terms. Save time and money.

Every night, PPCleaner reviews the searches your campaigns actually paid for, groups them by intent, and excludes the ones that never convert. No spend threshold to wait for, nothing touched but negatives — and one click undoes any of it.

Start 7 days free

Search, Shopping and Performance Max · Cancel in one click

Related articles

Phrase Match Negative Keyword Examples, Before and After — PPCleaner