Google Ads Negative Keywords: The Complete Guide to Eliminating Wasted Spend

Google Ads negative keywords are terms that stop your ads from showing for irrelevant searches. You add them at the campaign, ad group, or account level to exclude queries, wrong audiences, or unrelated products. In most accounts Opascope audits, unmanaged negative keyword lists drive a material share of wasted spend, often 15 to 30% of non-brand Search and Shopping budget. The fastest reclaim lever in any paid account is rarely a new bid strategy or a creative refresh. It is the negative keyword list nobody has touched in 90 days.

This guide covers the three match types and where each one fails, the difference between campaign, ad group, and account-level negatives, the Search Terms report as the only trustworthy source for new exclusions, the March 2025 Performance Max (PMax) expansion to 10,000 negatives per campaign, and a 30-day system that holds up across portfolio accounts.

Google Ads negative keywords are exclusion terms that prevent your ads from appearing on searches you do not want to pay for. Across Opascope’s managed accounts, 50%+ of paid accounts arrive on audit with underreported conversion data and material wasted spend, and negative keyword hygiene is the fastest reclaim lever the diagnostic team finds. It is the cheapest hour of work in paid media and the most consistently neglected.

Negatives matter more in 2026 than they did three years ago because the query surface has changed. Broad match is the default in Smart Bidding campaigns, close variant matching now sweeps in synonyms Google considers similar, and 25.11% of Google searches now trigger AI Overviews [1]. Each of these three changes adds queries to the auctions you participate in, including queries that were never relevant. Irrelevant queries that earn no clicks also drag down Quality Score on the serving ad group. A wasted click also costs more. Cost per click on competitive non-brand queries has tracked the 30 to 40% year-over-year cost per mille (CPM) inflation paid social has seen, so the same junk click costs roughly a third more than two years ago.

Why negatives matter more in 2026:

• Negative keywords work like inverted keywords: instead of telling Google what to match, they tell Google what to skip.
• Broad match expansion, close variants, and AI Overviews all enlarge the query pool you pay against.
• Wasted clicks cost roughly a third more than they did two years ago. The reclaim lever has gotten more valuable, not less.

For how query control fits into broader account health, see paid media management and how AI search optimization is changing the query mix paid accounts sit inside.

Google Ads offers three negative match types: broad, phrase, and exact. Most wasted spend comes from using the wrong match type, not from missing negatives entirely. The same word added at the wrong match level can either overblock useful traffic or quietly let unwanted queries slip through.

The behavior is opposite to positive keywords, which trips up most teams the first time:

• A broad-match negative blocks searches that contain all your terms, in any variation or order. Use it to exclude an entire theme regardless of word order. The term “jobs careers” as a broad negative blocks any query containing both words. It fits blanket categories like employment, free intent, or DIY content. It overblocks when the same words have legitimate uses.
• A phrase-match negative blocks searches that contain the phrase, including when surrounding words appear on either side and when the phrase is inverted by Google’s close variant matching. Use it to block a specific phrase but tolerate flexibility in how the rest of the query reads. “Best free crm” as a phrase negative blocks “best free crm software” but leaves “best paid crm” alone. Synonyms Google treats as close variants can still slip through.
• An exact-match negative blocks only the precise query in the order specified. Use it for surgical removal of one query without disturbing close variants. “Free trial” as an exact negative blocks only that two-word query, not “free trial software” or “crm free trial.” The trade-off is that it misses every variation of the same intent, so it is slow for theme cleanup.

The most common error is using broad-match for terms with mixed intent. “Free” as a broad negative looks reasonable until it blocks “free trial,” “free consultation,” and “toll-free.” For terms with both legitimate and unwanted contexts, default to phrase or exact.

Match-type defaults:

• Broad: theme-level. Phrase: contextual. Exact: surgical.
• Default to phrase for mixed intent. Default to broad for whole categories you never want to appear for. Default to exact for one-query surgical removal.

Test every new broad-match negative against the Search Terms report for a full week before locking it in. During that window, watch the impression and click counts on the queries the negative was meant to block, and watch conversion volume on the queries you still want to serve. A clean signal to keep the negative looks like a clear drop in impressions and clicks on the irrelevant queries with no drop in conversions on the legitimate ones. A signal to revert looks like a conversion-volume dip on the campaign as a whole, or relevant queries quietly disappearing from the Search Terms report, which usually means the negative is overreaching into close variants. Revert at the first sign of either, and demote the negative from broad to phrase or exact before re-testing.

Negative keywords apply at three levels: ad group (narrowest), campaign (broader), and account via shared lists (broadest). In audits, most wasted spend ties to a failure to use shared lists, which forces manual reapplication on every new campaign and lets drift accumulate.

The level you choose changes how the exclusion behaves and how much maintenance it creates:

• Ad group level suits negatives that protect a single ad group from internal cannibalization. If one ad group bids on crm software and another bids on crm software free, the first should hold a negative for the free variant so the right ad group serves the right query.
• Campaign level suits negatives that apply to every ad group in a campaign but not to other campaigns. Brand campaigns block competitor brand names. Generic non-brand campaigns block your own brand terms so brand campaigns earn the brand click instead.
• Account level via shared lists is where most accounts underinvest. A shared list covering job-seeker terms, free intent, DIY and education, and competitor names should attach to every relevant Search and Shopping campaign. The list lives in Tools, Shared Library, Negative keyword lists, and updates propagate everywhere it is attached. New campaigns inherit the list at launch instead of leaking spend until someone notices.

The gap is sometimes a resource problem rather than a knowledge gap. In-house teams often know shared lists exist but do not have the hours to maintain them across multiple accounts and campaigns. The right move at that point is outside help, not another internal checklist.

Where each level fits:

• Ad group negatives stop internal cannibalization. Campaign negatives shape one campaign. Shared lists scale across the account.
• New accounts should ship with a baseline shared list before campaigns go live.
• A team that knows the work but cannot get to it is a candidate for Google Ads management, not another spreadsheet.

The Search Terms report shows the search queries that actually triggered your ads, not just the keywords you bid on. Since 2020, Google has limited search term visibility, which means 20 to 30% of queries aggregate under “Other,” but the visible queries still carry most of the spend signal. The report is the only place a negative keyword decision can be grounded, because every other source (keyword planner, instinct, last quarter’s list) is a guess about queries the account may or may not actually serve.

How to pull the report. Open the campaign or ad group, click Insights and reports, then Search Terms.

The view that helps managers spot wasted spend the fastest is the one that sorts by cost descending and filters to zero conversions. That filter shows the queries spending money without producing anything. A small subset will be queries with intent that should not be blocked yet; the rest are exclusion candidates.

A standard cleanup workflow:

• Sort by cost, filter to zero conversions over the last 30 to 90 days.
• Mark queries unrelated to the offer. Choose match type by intent (broad for theme, phrase for context, exact for one-off).
• Add at the right level. Single-query goes ad group or campaign. Theme exclusion goes to the shared list.
• Re-pull one week later. Negative keyword work is a continuous practice, not a one-time cleanup. Query patterns shift as Smart Bidding finds new auctions, as competitors enter and exit, and as seasonal terms move in and out of intent. The list is never finished, and an account that runs the same shared list for a year without revisiting it is leaking spend by the end of that year.

The “Other” aggregation is the part most teams ignore. Google rolls up low-volume queries to protect user privacy, so a portion of every account’s spend lands in queries you cannot read individually. Visible queries still carry most of the spend signal, and Other usually moves in proportion to the visible side once obvious negatives go in. The pattern: 20 to 30% of queries land in Other on accounts with mature query control, and a higher share on accounts with no discipline. If you clean the visible queries, you can improve Other.

How to work the Search Terms report:

• Sort by cost, filter to zero conversions, and review weekly. The work compounds when it is recurring.
• Only the Search Terms report tells you what queries actually served your ads. Every other source is a guess.
• “Other” hides 20 to 30% of queries on a healthy account. If you clean the visible queries, you can improve Other.

As of March 2025, Google raised the PMax negative keyword limit from 100 to 10,000 per campaign, bringing PMax in line with Search [2]. Performance Max can now be managed with the same query discipline as Search, if advertisers actually use the new limit. Most PMax campaigns Opascope audits still run with the old 100-keyword cap as a working assumption, which leaves a meaningful spend leak open.

The 2025 sequence matters. In January 2025, Google rolled out campaign-level negatives for PMax, replacing a clunky account-level support form. In March 2025, the limit moved from 100 to 10,000 per campaign, matching Search. Together, those two updates closed the gap on negative keyword discipline. Most PMax accounts launched before January 2025 have never been touched.

What still does not get exposed: auto-generated asset variations and the search queries that triggered them. PMax bundles Search, Shopping, Display, YouTube, and Discovery placements behind a single asset group, and its Search Terms report is thinner than the Search version. A portion of PMax spend always flows through placements where keyword negatives do not apply. Brand exclusion lists handle one specific case: keeping PMax from absorbing your branded queries, which inflates platform-reported return on ad spend (ROAS, the cost-of-ad-spend efficiency metric) without producing incremental revenue.

What changed in PMax 2025:

• PMax negatives went from 100 to 10,000 per campaign in March 2025. The gap with Search is closed.
• Most accounts launched before January 2025 run as if the 100-keyword limit were still in place.
• Brand exclusion lists are the single most important PMax negative to set.

For the structural side of getting PMax under control, see Performance Max management and how it pairs with the broader account.

In 50%+ of paid accounts Opascope audits, wasted spend traces to the same five negative keyword failures. Each is fixable in an afternoon. The pattern repeats across verticals, account sizes, and platform versions, which means the failure mode is structural, not specific.

The five patterns:

1. No shared account-level list for job-seeker, free-intent, and competitor queries. A new campaign launches without a baseline list, and the first 60 days leak spend on careers, salary, free trial (when the offer is paid), and direct competitor names. The fix is one shared list with employment, free intent, DIY, and competitor categories, attached at launch.
2. Wrong match type on high-frequency negatives. A broad negative for free blocks free trial, toll-free, and free consultation. The match-type error often costs more than the missing negative would have. When a campaign suddenly drops 20% in conversion volume or sees CTR collapse after a negative update, the new negative is almost always a broad-match overreach.
3. No negatives on brand campaigns against competitor product names. Without competitor negatives, broad match expansion pulls in queries that include both your brand and a competitor, and the brand campaign starts paying for queries it should not. The fix is a campaign-level negative list of competitor names on every brand campaign. When the goal is to target competitor queries on purpose, run a separate competitor-conquest campaign where those competitor names are not excluded, so the brand campaign stays clean and the conquest campaign carries that intent on its own.
4. Outdated negatives dragging down volume. A negative added in 2023 to fight a launch-period spike is still in place in 2026, blocking queries that would convert today. The fix is a quarterly match type sweep on the negatives serving against the most queries by impression count, which is where stale exclusions cause the most lost volume.
5. Performance Max running without negatives because the manager assumed they were unavailable. Pre-2025 PMax could not hold meaningful negative lists, and that assumption has not updated for most account managers. PMax then absorbs brand search volume and inflates platform-reported ROAS while real performance erodes. The fix is the campaign-level list now allowed at 10,000 entries.

The pattern across all five is that the team often knows the answer in principle and has not gotten to it in practice. Failing to optimize negative keyword management is a structural cause of platform-reported ROAS drifting far above P&L ROAS, because spend on low-intent queries earns clicks and reported conversions on the platform but never produces a paying customer the P&L recognizes. Across Opascope’s managed accounts, platform-reported ROAS has run 1,140% while the P&L came in at 77%.

The pattern across all five:

• The same five failures repeat across audited accounts. Each fix is a one-afternoon task.
• The most expensive error is often not a missing negative but a broad-match negative that overblocks legitimate traffic.
• Cleaning the query surface closes the gap between platform-reported ROAS and P&L ROAS.

For accounts where these patterns are likely live, the diagnostic is one paid media audit and a few hours with the Search Terms report.

A working negative keyword strategy has three recurring actions: a weekly Search Terms report review, a monthly shared list audit, and a quarterly match type sweep. On a $50,000 to $500,000 per month account, this cadence typically reclaims 10 to 20% of non-brand Search spend inside 90 days. The system holds up at scale because each pass catches a different class of failure, and skipping any one of them lets that class of waste accumulate.

The cadence:

• Weekly Search Terms review. Pull the report at the campaign level. Sort by cost, filter to zero conversions over the prior 7 to 14 days. Add the obvious negatives at the right level. 30 to 45 minutes per account.
• Monthly shared list audit. Open every shared list. Verify each is attached to the right campaigns. Add new categories from the weekly review. 60 minutes per account.
• Quarterly match type sweep. Pull the highest-impression negatives. Review whether each still blocks the right intent or whether the offer has shifted. Demote broad-match negatives to phrase where the data shows overblock. 2 to 3 hours per account.

For accounts above $500,000 per month, n-gram analysis adds a fourth layer. An n-gram is a sequence of N consecutive words pulled from a query: a 1-gram is one word (“free”), a 2-gram is two consecutive words (“free trial”), a 3-gram is three (“free trial software”). Tokenize 90 days of queries into 1-grams, 2-grams, and 3-grams, then rank each segment by spend on zero-conversion queries. The view catches recurring words that drive waste across otherwise-unrelated queries. For example, a 1-gram report might surface that the word “tutorial” appears in $14,000 of zero-conversion spend across 47 different queries, none of which would have stood out on a query-by-query review, and that one finding produces a single high-leverage broad-match negative.

On a lifestyle brand spending $1.4M a month on ads, this 30-day system was part of the structural fix that cut the ad-spend-to-revenue ratio from 70% to 35% in 12 months. Negative keyword hygiene was one lever among several (attribution and landing pages were the others), but the system above is what kept the gains compounding once the rebuild was done.

For a cloud platform Opascope rebuilt, cost per acquisition (CPA) dropped 93% and sign-ups grew 660% over four years once attribution, negative keyword hygiene, and bidding structure were aligned. Paid went from almost no paying customers to 30 to 40% of new customers. Without the negative keyword work, the bid strategy would have optimized against query data the team had not cleaned.

Cadence at scale:

• Weekly, monthly, quarterly negative keyword review. Each catches a different class of failure.
• A $50,000 to $500,000 per month account typically reclaims 10 to 20% of non-brand Search spend in 90 days on this cadence.
• Above $500,000 per month, n-gram analysis is the next layer. The eye misses what the tokenizer catches.

If the account has not had a real negative keyword review in 90 days, it almost certainly has wasted spend. The Search Terms report is the place to start, and a baseline shared list of employment, free intent, DIY, and competitor terms is the next move. Most accounts reclaim up to 20% of non-branded Search spend within 30 to 60 days of a real review. To quantify what an account is currently leaking, run a Google Ads audit or book an audit call.

Q: What are negative keywords in Google Ads?

Negative keywords are terms you add to a Google Ads campaign, ad group, or shared list to prevent your ads from showing for irrelevant searches. They work the opposite of regular keywords: instead of telling Google what queries to match, they tell Google what queries to exclude. Adding them is the single fastest way to cut wasted spend in most accounts.

Q: How do I add negative keywords to a Google Ads campaign?

In the Google Ads UI, open the campaign, click Search keywords in the left menu, then select Negative search keywords. Add your terms, choose a match type (broad, phrase, or exact), and save. You can also manage account-level negatives under Tools, Shared Library, Negative keyword lists. Adding at the shared-list level is preferred when the same exclusions apply across multiple campaigns.

Q: What is the difference between broad, phrase, and exact match negative keywords?

A broad-match negative blocks searches that contain all your terms in any variation, including different orderings. A phrase-match negative blocks searches that contain the phrase, including when the phrase is inverted or wrapped in surrounding words. An exact-match negative blocks only the precise query in the order specified, with no variation. Broad is the right call for whole themes you never want to appear for. Phrase fits when you want to block a specific unwanted context but not its close cousins. Exact is for surgical removal of one query without touching anything else.

Q: Do negative keywords work in Performance Max campaigns?

Performance Max supports up to 10,000 negative keywords per campaign as of March 2025, matching Search [2]. Before January 2025, PMax negatives were restricted to 100 per campaign, which made true query control nearly impossible. The expanded limit means PMax accounts can now be managed with Search-grade negative keyword discipline, though most accounts still run as if the old cap were in place.

Q: How many negative keywords should a Google Ads account have?

There is no universal number. Small accounts often run well with 50 to 200 negatives. Accounts with $100,000+ per month in non-brand Search spend typically need shared lists with 1,000 to 5,000 entries covering job-seeker, free-intent, competitor, and irrelevant-product queries, plus campaign-level additions. The right number is whatever it takes to keep the Search Terms report clean each week.

Q: What are common negative keywords every advertiser should add?

Common account-level negatives fall into four buckets: employment (jobs, careers, salary), price (free, cheap, discount for premium brands), DIY and education (how to, tutorial, DIY), and competitor brand terms you do not want to bid against. Industry layer matters: a law firm should also exclude pro bono and student; a B2B SaaS should exclude review and vs until brand search campaigns are warranted.

Q: How often should I review negative keywords?

On an active account, once a week. That is the cadence at which the Search Terms report surfaces new waste fast enough to matter. On lower-spend accounts under $10,000 per month, biweekly is defensible. Less frequent than monthly, and wasted spend compounds before anyone catches it.

Q: Can negative keywords hurt campaign performance?

The wrong match type can block relevant traffic. A common mistake is using broad-match negatives that overreach. For example, adding free as a broad negative will block free trial, free consultation, and toll-free, along with the unwanted free PDF download. Use phrase or exact match for terms that have both legitimate and unwanted contexts.

Q: Do negative keywords affect Quality Score?

Not directly. Negative keywords do not enter the Quality Score calculation. They affect Quality Score indirectly by improving Expected click-through rate (CTR, the share of impressions that produce a click): when ads stop showing for irrelevant queries, the ads that do show earn higher CTR, which raises Quality Score over time. Treat negatives as a structural lever, not a direct Quality Score tool.

Q: How do I find negative keywords to add?

The Search Terms report in Google Ads is the primary source. Sort by cost, then review every query with zero conversions or a CPA more than two times your account average. Add negatives at the narrowest level that prevents the waste without blocking related relevant traffic. For new campaigns, a pre-launch list covering obvious industry-level exclusions (jobs, DIY, free) is essential before the campaign enters the learning phase, and ideally locked in before launch. The learning phase is where Smart Bidding decides what queries are worth converting against; if you let early spend leak into junk queries during that window, the algorithm anchors to bad signal, and the cleanup later costs more than the prevention would have.

1. Conductor 2026 AI Overviews Research, 25.11% of Google searches trigger AI Overviews. https://www.conductor.com
2. Google Ads Help: Negative keywords in Performance Max campaigns. https://support.google.com/google-ads/answer/15726455