Google Ads Negative Keywords: Match Types and Workflow

Most Google Ads accounts need far more negative keywords than they actually have. The result is budget flowing into search queries that never lead to a paying customer, regardless of whether the account spends $300 or $3,000 per month.

Why Negative Keywords Matter More Than New Keywords

According to a WordStream study of more than 15,000 Google Ads accounts, the average account wastes around $1,127 per month. 25% of advertisers in that analysis had not added a single negative keyword. And: accounts with at least one negative keyword had on average a 3x higher conversion rate than accounts without.

New keywords give you reach. Negative keywords determine if the traffic you’re paying for is actually relevant to your business. I consider them the most underrated optimization lever in any Google Ads account. Not because the topic is complicated, but because it is boring. No one does it voluntarily.

The Three Levels and When Each Applies

Shared lists are centralized negative keyword lists you can manage in one place and apply to multiple campaigns at the same time. They are the most important foundation, before the question of match type even comes up: which negative belongs at which level?

My decision tree for every new negative-keyword candidate:

Term ALWAYS irrelevant, regardless of campaign?
  → SHARED LIST (account-wide)
Term irrelevant in only ONE campaign, relevant in others?
  → CAMPAIGN LEVEL
Term is a competitor name?
  → SHARED LIST "Competitors" (Exact Match)
Cross-negative inside a campaign (SKAG/STAG structure)?
  → AD GROUP LEVEL (rarely needed)

Level	When	Typical example
Shared List “General”	For jobs, free, DIY queries, anything wrong in every campaign	[free], [training], [jobs]
Shared List “Competitors”	Competitor names as Exact Match, taken 1:1 from search terms	[dr max mustermann]
Campaign Level	Term hurts one campaign, helps another	[online appointment] in a phone-appointment campaign
Ad Group Level	Cross-negative in granular SKAG structure	Rare, only with tightly themed ad groups

Choosing the wrong level creates work. Shared List for everything = unmaintainable over time, because the list will block campaigns that actually need the term. Campaign Level for everything = you maintain the same 50 generic terms across ten campaigns separately.

Match Type: Exact Is the Standard, Phrase Is the Exception, Broad Is Off Limits

Match type defines how exactly a search query must match your negative keyword to be blocked.

Match type	Format	Effect	Typical example
Exact	[keyword]	Only on exact match	[dr max mustermann]
Phrase	"keyword"	When the phrase appears in the search term	"reviews forum"
Broad	keyword	On any word combination	NEVER

Exact is the default for over 90% of negatives I add. The logic: I have seen a specific search query that does not fit, and I want to block exactly that one. Nothing more, nothing less.

Phrase Match is the exception, not the rule. It makes sense when a phrase appears in many variants. A SaaS company that does not want free-trial requests adds "free" as Phrase. That blocks free version, free access, download software free. But: Phrase Match can sweep up wanted queries, so use it only where the pattern is genuinely clear.

I never use Broad Match for negatives. It blocks too widely and too unpredictably. A single misplaced Broad negative can hide relevant queries across multiple campaigns. In my opinion, there isn’t enough documentation on how Smart Bidding affects matching behavior to build a strategy around it. When in doubt, stay with Exact or Phrase.

Quick Distinction: Broad Match as a Positive Keyword

This is not to be confused with Broad Match as a positive keyword strategy. In mature accounts with clean conversion tracking and enough historical data, Broad Match in combination with Smart Bidding can make sense. Not as a replacement for Exact and Phrase, but as an additional layer to discover new queries that tightly worded keywords miss.

The emphasis is on “can”. The approach only works when the Smart Bidding model has enough qualitative data: enough conversions, clean value tracking, clear contexts. Without that foundation, Broad Match is an expensive bet. That is its own topic with its own conditions and will be covered in a later article.

Hypothetical Examples Across Industries

• E-commerce: [brand-x product] Exact, when the shop does not carry that brand but ranked for the brand name anyway.
• SaaS: "free" Phrase, when you do not run a free-plan business.
• Trades and contractors: [tutorial] Exact as a DIY signal.
• Local service: [neighbor city] Exact as a location negative, when targeting does not include that city. Not Phrase. Phrase would block too widely.

The Most Common Mistakes From Real Audits

Four patterns I keep finding in account takeovers, all anonymized and described generically.

1. Truncating the Search Terms Manually

The search-terms report shows exactly what was searched, for example max mustermann doctor.

Mistake: you add [max mustermann] as a negative. The logic (“then I catch all variants”) sounds plausible. It is wrong.

Exact Match matches that exact term. [max mustermann] blocks max mustermann only, not max mustermann doctor. The shortcut goes nowhere and the original query keeps firing.

Rule of thumb: Copy the search term exactly as it appears in the report and use square brackets. No simplifications, no summaries, no “clever” rewordings. I made this mistake myself early on, before it became a hard rule.

If you want to block both max mustermann AND max mustermann doctor: add both as separate Exact Match negatives. Or use Phrase Match on "max mustermann", with full awareness that variants like max mustermann appointment will also be blocked.

2. Excluding a Geo That Is Actively Targeted

A ZIP code or city name as a negative, while the campaign targets that exact area. Result: the campaign keeps running in the region. The negative only triggers when the search query contains the place name, paradoxically blocking the right queries from exactly that region. External providers also rarely catch this proactively, because it requires looking at two settings simultaneously: targeting and the negative list.

3. Half-Excluding a Service

Blocking [repair tutorial] as a DIY signal is clean. Blocking [repair] outright at a contractor that offers repairs costs you paying customers. The difference between “request a repair” and “do the repair yourself” lives in the qualifier. The qualifier belongs in the negative, not the base term.

4. Copy-Pasting Logic From One Account to Another

What is clearly junk for one shop can be relevant search volume for a service business. A list that works for Account A must not be blindly copied into Account B. Every list belongs to its account and industry, not to your routine.

Universal Negatives: A Base for Most SMB Accounts

There are terms that are rightfully blocked in most SMB accounts. My working starter list looks like this:

free | gratis | cheap
jobs | salary | training | careers
wikipedia | youtube | forum | blog
tutorial | how-to | diy
review | test | comparison

This list is a starting point, not dogma. I still review it per account.

A classic case for case-by-case review: reviews. In a review-heavy niche (software comparison, financial advice, practices using patient feedback as a buying argument), it is highly relevant traffic. For a contractor it is more of a DIY query. Do not blanket-add it.

Same with test and comparison: for a SaaS provider often desired traffic, for an online shop more informational searching without buying intent.

The Workflow: Time and Efficiency

There’s no one-size-fits-all time commitment for negative keyword maintenance. How long the routine takes per account depends on multiple factors:

• Search query volume: an account with 2,000 new search queries per week takes longer than one with 50.
• Fluency: if you are doing this for the first time, decisions are slower. After a few months you spot patterns at a glance.
• Filter setup: with saved filters in the Google Ads UI (Cost greater than X, Conversions equals 0, Time range 7 days) you save time on every pass.
• Scripts and automation: the PPC community publishes free Google Ads scripts that pre-sort negative candidates. For example the scripts by Nils Rooijmans, an established resource for SMB accounts.

My base flow per account:

1. Open the Search Terms Report for the last 7 days.
2. Sort by cost descending. Most expensive search queries first.
3. Review your top spenders that haven’t converted:
   • For accounts with clean conversion tracking: threshold = 2x your target CPA without conversion. If your target CPA is $20, review everything starting at $40 in cost without conversion.
   • For accounts without a clear CPA: a fixed dollar threshold. For example $20 per search query without conversion. If you don’t have conversion tracking set up, start there. That is the more important groundwork than any negative-keyword maintenance. (Why your Google Ads conversion tracking is probably wrong.)
4. Decide per candidate:
   • Level. Shared List, Campaign, rarely Ad Group.
   • Match type. Exact as standard, Phrase only with a clear pattern.
   • Add to the right list.
5. Next account.

For high-volume accounts, an additional step pays off: the n-gram analysis. Instead of looking at single queries, you analyze which word fragments (“tutorial”, “diy”, “free”) show up disproportionately often in your costly searches. This makes negative work more efficient at a higher level, because you no longer decide click by click but block patterns. I’ll cover the full n-gram workflow in its own post (linked here when it goes live).

When the Topic Is Less Critical

Budget size is not the factor. What counts:

1. Number of cost-triggering search queries. 500 clicked queries per month need maintenance, 30 do not. This is independent of $300 or $3,000 in budget.
2. Single queries with high cost. A $20 click without conversion is a problem in any account and should be fixed immediately.
3. Account maturity. Newly launched accounts with low volume are still gathering data. Weekly maintenance is overkill there. Once your Search Terms Report starts hitting double-digit entries every week, it’s time to take it seriously.

Rule of thumb: if you see more than 50 new cost-triggering search queries weekly, negatives save you real money, regardless of total budget.

A clean negative keyword setup is the core discipline that keeps an account healthy independent of agency tool stacks. The lever still works five years from now, because it has nothing to do with a specific tool or a specific optimization trend.

FAQ

How often and how long should I maintain negative keywords?

How often: weekly for active accounts, monthly for small or stable ones. The driver is search query volume, not budget. How long: no fixed time. With filter setup and fluency, very fast. For large accounts without a routine, considerably longer. Free scripts from the PPC community can speed up the process significantly.

Can negative keywords hurt performance?

Yes. Phrase Match negatives that are too broad also block relevant queries. I do not use Broad Match negatives for that reason. Whether Smart Bidding changes the matching behavior here is not documented well enough to build a strategy on. The standard is Exact, Phrase only with clear word-group patterns. When in doubt: Exact Match, search term copied 1:1 from the report.

What is the difference between Shared List and Campaign Level?

A Shared List applies account-wide, provided you link it to the relevant campaigns. Campaign Level applies to a single campaign. Anything irrelevant in every campaign belongs in the Shared List. Anything that hurts only one campaign stays at Campaign Level. Important: Shared Lists you create but do not link to campaigns have no effect.

I do not have conversion tracking, is it still worth maintaining negatives?

Only partially. Without tracking, the most important basis for decisions is missing, which search terms actually brought paying customers. You can still block obvious junk queries, but prioritization becomes guesswork. The right order: get tracking clean first, then negatives. Why your Google Ads conversion tracking is probably wrong.

Your Next Step

If you think your account hasn’t had a negative keyword scrub in months, an audit makes that visible. How much per campaign flows into irrelevant searches, which exclusion candidates bring the largest savings, in which order to work. With me this is part of the 90-day project work, with clean documentation so you can take over the maintenance afterwards. No agency lock-in. → Free initial consultation