If you have inherited a Google Ads account and asked what are negative keywords, the short answer is that they are the instructions telling Google which searches should never trigger your ads. The longer answer is that they are the difference between an account that scales and one that gets more expensive the more you spend on it.
Negative keywords are defined as: terms you add to a Google Ads campaign or account to prevent your ads from showing on searches containing those terms. Where a keyword tells Google what to bid on, a negative keyword tells it what to ignore.
Most B2B SaaS teams know the concept. Far fewer have a working list, and a surprising number have one that actively costs them money. Here is how they work and the 6 places I most often find them going wrong.
How negative keywords actually work
Three match types, and choosing between them is most of the skill.
Broad negative. Blocks any search containing all of your words in any order. A broad negative on jobs blocks “security analyst jobs,” “jobs near me,” and “how to get jobs in cybersecurity.” This is the blunt instrument and it is usually correct for whole categories of junk.
Phrase negative. Blocks searches containing your exact phrase in that order. A phrase negative on free tool blocks “best free tool for scanning” but not “tool for free scanning.” Use this when a word is dangerous on its own but safe in most contexts.
Exact negative. Blocks only that precise search and nothing else. Narrow enough that it is mostly useful for specific brand collisions.
One critical asymmetry: negative keywords do not have the loose matching that regular keywords do. A broad match keyword will happily match synonyms and related concepts. A negative keyword will not. If you negate certification, searches for certifications may still come through depending on how Google handles the variant, so plural and singular forms both belong on the list. Google’s documentation on negative keywords covers the current matching rules, which are worth re-reading because they have changed.
The 6 mistakes B2B SaaS accounts make
1. Blanket-negating comparison terms
The most expensive one, and I find it constantly.
Somebody notices that a lot of informational traffic contains the word “vs,” so they negate vs account-wide. That single negative blocks every query of the form “[competitor] vs [your product],” which is among the highest-intent searches anyone types about your category. A person comparing two named vendors is at the end of a research process, not the beginning.
The fix is to negate vs inside informational campaigns only, then run a dedicated comparison campaign that bids on those terms deliberately.
2. Treating it as a one-time cleanup
An account gets audited, someone spends an afternoon adding 300 negatives, and the task gets marked done.
Search behavior changes every month. Competitors launch and rebrand. A new certification gets popular and floods your category with learners. A funding announcement puts a new company name into circulation. The list you built in March is meaningfully out of date by September.
The accounts that stay clean have a recurring 20-minute habit rather than an annual purge.
3. Adding negatives at the campaign level only
This one is quiet and compounds. Campaign-level negatives do not travel. Six months later somebody launches a new campaign, it inherits nothing, and the junk you already paid to identify starts costing you again.
Build negatives as shared lists applied at the account level, and reserve campaign-level negatives for the genuine exceptions, like protecting a free-trial campaign from a broad free negative.
4. Importing a giant scraped list with no protect list
There are published negative keyword lists with thousands of entries. Pasting one in feels productive and is one of the faster ways to damage an account.
Those lists are built for general B2B or ecommerce and contain terms that are buyer signals in your category. pricing, cost, enterprise, platform, vendor and demo all appear on lists I have seen, and every one of them is a qualifying word that indicates a business buyer.
Any list worth using comes with a do-not-negate list attached. Mine is in the cybersecurity negative keyword list, and it is the section I would read first.
5. Negating “free” while running a free trial
Self-inflicted and common. free looks like obvious junk intent, and for free tool and free download it is. But if your primary offer is a free trial, a free demo or a free assessment, a broad negative on free blocks your own campaign from serving on its own offer.
Negate the specific junk phrases rather than the word.
6. Assuming Performance Max obeys them the same way
Negative keyword support in Performance Max has changed several times, and what is true this quarter may not have been true last year. Check the current documentation rather than trusting a blog post, including this one.
The practical point that has held steady: PMax matches queries more loosely than a tightly-run Search campaign, so a negative list gives you less protection there than you would expect. If a meaningful share of your budget sits in PMax and your lead quality dropped after you turned it on, the negative list is not the whole fix.
What this actually costs
Concretely, in a B2B SaaS or cybersecurity account with no governed negative list, expect somewhere between 40% and 60% of search spend to reach people who could never buy. I break the categories down in what counts as wasted ad spend.
The reason it goes unnoticed is that none of it looks like a problem in a weekly report. Junk traffic has a healthy click-through rate, because a curious person clicks readily. Some of it converts, because that same person will trade an email address for a whitepaper. Cost per lead holds steady or improves.
So the dashboard says the account is working while sales quietly stops picking up the leads.
You can size your own number in about 20 minutes, using the one report that shows raw queries. Open your search terms report, set it to the last 90 days, sort by cost, and read the top 50 queries out loud. Mark each one plausible or not plausible for someone who would eventually pay you. The cost of the not-plausible rows, divided by the total, is your waste rate.
Most people are surprised, and the number is usually worse than the sample suggests, since Google hides a share of low-volume queries.
The short version
Negative keywords tell Google what to ignore. Getting them right means using the correct match type, applying them as shared account-level lists, maintaining them monthly rather than annually, and keeping a protect list of the buyer-intent terms that look like junk but are not.
Get them wrong and you either keep paying for students and job seekers, or you block the comparison queries that were about to become customers. Both happen. The second one is worse and harder to notice, because the account just quietly gets smaller.
Most of the 6 mistakes above are invisible from inside a dashboard, which is why they survive so long. If you want to know which of them your account is currently making, book a BADASS Discovery Call at bad2badass.com and we will look together.