Negative Keywords Playbook for Cutting Wasted Spend

A single negative keyword you forgot to add last Tuesday is probably still burning budget right now. Not a lot, maybe a few dollars a day, but multiply that by dozens of irrelevant queries across every campaign in your account and you’ve got a line item your CFO would call “waste” and you’d call “I meant to get to that.”

Bid strategies decide how much you pay. Targeting decides who sees your ads. But negative keywords decide who doesn’t see them, and most PPC managers underinvest in that decision by a wide margin.

You’ll learn match-type mechanics most guides get wrong, a level-by-level decision framework, a discovery workflow with a real cadence, and the one thing no competitor covers: when a “negative” would actually be a mistake.

A PPC manager's workspace — dual monitors showing Google Ads interface with search terms report visible

TABLE OF CONTENTS:

What are negative keywords?

A negative keyword tells Google Ads: don’t show my ad when a search query contains or matches this term.

You add “free” as a negative, and searches containing “free” stop triggering your ads. That’s the glossary definition.

Here’s why it’s a budget problem.

Every irrelevant click costs you twice. You pay for the click itself, and you dilute the conversion data your bidding algorithm relies on. Smart Bidding learns from what converts. Feed it junk clicks and it learns the wrong lessons, which means worse bids on the queries that actually matter.

Cleaning up negatives is the foundation your entire bidding strategy sits on.

If you’re already running expert-level bid strategies, bad negative hygiene undermines every one of them.

Negative keyword match types: what actually gets blocked

This is where most articles, including Google’s own help doc, either oversimplify or get it wrong.

Negative match types do not behave the same way as regular keyword match types.

The single most important difference: negative keywords do not match close variants.

Close variants don’t apply to negatives

When you add a regular broad match keyword, Google happily matches it to synonyms and misspellings.

Negatives don’t get that treatment.

If you negate “running shoe,” Google will still serve your ad for “running shoes” (plural), “runing shoe” (misspelling), and “jogging shoe” (synonym). You must add each variation separately.

Casing, however, is handled automatically, so “Running Shoe” and “running shoe” are treated identically.

This means your negative keyword lists need to be more explicit than your targeting lists. It’s counterintuitive, and it’s the single mechanic most PPC managers miss.

How each match type filters queries

Let’s use the negative keyword “leather jacket” and see what each match type blocks and allows.

Search Query Negative Broad Negative Phrase Negative Exact
leather jacket men Blocked Blocked Not blocked
buy leather jacket Blocked Blocked Not blocked
jacket leather vintage Blocked Not blocked Not blocked
leather jacket Blocked Blocked Blocked
faux leather coat Not blocked Not blocked Not blocked
leather jackets Not blocked* Not blocked* Not blocked*

*Remember: “jackets” (plural) is a close variant that negatives won’t catch.

You’d need to add “leather jackets” as a separate negative.

Negative broad blocks any query containing all the words in your negative, in any order.

Negative phrase blocks queries containing the exact phrase in that word order.

Negative exact blocks only that precise query, nothing more.

Our default recommendation: use negative phrase for most exclusions.

It’s precise enough to avoid overblocking and broad enough to catch variations you’d otherwise miss. Reserve negative exact for surgical, single-query blocks where phrase would be too aggressive. Use negative broad sparingly and only for universally irrelevant terms like “free” or “DIY.”

Where to place negatives: a decision framework

Google Ads gives you three levels for negative keywords, plus shared lists.

Most guides describe the hierarchy and stop there. What you actually need is a framework for deciding which level gets which negatives.

Account-level negatives

Account-level negatives apply across every campaign in the account, including Search and Shopping.

Use this level for terms that are never relevant to your business, regardless of campaign type or objective.

Good candidates: your industry’s classic junk terms. If you sell B2B software, “jobs,” “salary,” “internship,” and “careers” belong at the account level. So do competitor brand misspellings you’ll never bid on and terms from entirely wrong verticals.

Be conservative here. An account-level negative cannot be overridden anywhere below it. If even one campaign could plausibly target that term, push it down a level.

Campaign-level negatives

Campaign-level negatives let you shape traffic routing between campaigns.

This is where you enforce brand vs. non-brand separation: negate your brand terms in non-brand campaigns and negate generic terms in brand campaigns.

They’re also critical for Performance Max. Google now supports campaign-level negatives in Performance Max, with a capacity of up to 10,000 terms per campaign. That capacity matters because PMax’s automated targeting is aggressive, and negatives are one of the few manual controls you have.

Ad group-level negatives

Ad group-level negatives are your scalpel.

Use them to prevent cannibalization between ad groups within the same campaign.

Example: you have ad groups for “running shoes” and “trail running shoes.” Add “trail” as a negative to the general “running shoes” ad group so trail-specific queries route to the right ad copy and landing page. This improves Quality Score and conversion rates without touching bids.

How the levels actually interact

Here is the mechanic most guides state backwards, and getting it wrong will cost you traffic you meant to keep.

Negatives are cumulative. A negative applied at one level applies to everything beneath it, and nothing below can cancel it.

If you negate “wholesale” at the campaign level, every ad group in that campaign stops serving for wholesale queries. Adding “wholesale pricing” as a positive keyword inside one ad group will not bring that traffic back. The positive keyword and the negative are not in competition; the negative simply wins, because exclusions are evaluated as a union across all three levels.

There are only two real fixes. Remove the negative from the campaign level and re-apply it to each ad group that genuinely needs it, or split the traffic you want to keep into its own campaign without that negative.

This is why placement matters more than it looks. Set a negative too high in the hierarchy and you lose coverage silently: nothing breaks, no warning appears, and impressions just quietly stop. Check your keyword diagnostics after any change at the campaign or account level.

Two PPC specialists reviewing campaign structure on a large monitor, sticky notes with keyword categories on the desk

A repeatable discovery workflow

Finding negative keywords is a recurring process, and the cadence matters as much as the sources.

Search terms report mining (weekly)

Pull the search terms report every week.

Sort by spend descending and look for queries that ate budget without converting. But don’t stop at zero-conversion terms.

Also flag queries where the cost per conversion is more than double your target CPA. Those are slow leaks that won’t show up in a simple “zero conversions” filter.

Pay attention to patterns as well as individual queries. If you see five queries containing “template,” that’s a modifier to negate. One entry does the work of five individual negatives.

Internal site search and CRM data (monthly)

Your site’s internal search bar captures what visitors look for after they land. If people keep searching for things you don’t offer, those terms are negative keyword candidates.

Sales and support call notes are even more revealing. When your sales team keeps hearing “do you offer a free version?” or “is this for personal use?” those are intent signals your search terms report won’t surface until after you’ve already paid for the clicks.

Low-converting modifiers and intent clustering (biweekly)

Group your search terms by modifier patterns.

Words like “tutorial,” “example,” and “how to” often signal research intent rather than purchase intent. Words like “cheap,” “free,” and “coupon” signal bargain-hunting.

Don’t negate these reflexively. Cluster them first, check conversion data for the cluster, then make a call. Some “how to” queries convert well for service businesses. We’ll cover when not to negate in the next section.

Competitor brand terms (quarterly review)

Unless you’re running a deliberate competitor conquest campaign, competitor brand names should be negated. They burn budget on users who’ve already chosen someone else.

Review quarterly because new competitors enter your space and brand names change.

Here’s the cadence summary: weekly for search terms reports, biweekly for modifier analysis, monthly for site search and CRM data, and quarterly for competitor reviews.

Stick to this rhythm and you’ll catch waste before it compounds.

Proper PPC audit practices should fold negative keyword reviews into every cycle.

Negative keyword categories by intent, and when not to use them

This is the section no competitor publishes.

Everyone gives you lists of negatives to add. Nobody tells you when a “standard” negative would be a mistake for your specific business.

Intent Category Example Negatives When to Exclude When Excluding Would Be a Mistake
Job seekers jobs, careers, salary, hiring, internship Almost always, unless you’re a recruiting firm Staffing agencies and HR platforms: these are your buyers
Students / researchers what is, definition, example, PDF, thesis E-commerce and direct-response campaigns Educational SaaS, course sellers, or top-of-funnel content campaigns where research intent is the entry point
Bargain hunters cheap, free, discount, coupon, deal Premium and luxury brands, high-AOV B2B E-commerce with active promotions, freemium SaaS, or lead-gen offers where “free trial” is the conversion event
DIY intent DIY, how to build, homemade, tutorial Service businesses where DIY = lost sale Tool and supply companies whose customers ARE the DIYers, or brands using content marketing to capture early intent
Free-seekers free, open source, no cost, gratis Paid software, premium services Freemium models where “free” traffic converts to paid plans downstream
Wrong vertical Varies: “industrial” for consumer brands, “residential” for commercial When the vertical is truly unrelated Companies expanding into new verticals, or verticals that share decision-makers with your core audience

The “when excluding would be a mistake” column should be your sanity check before adding any category-wide negative.

Never negate an entire modifier category before checking whether it actually converts for your account. If “free” queries drive even a small number of conversions at an acceptable CPA, you’re better off keeping them and managing bids than cutting them entirely.

Mistakes that cost more than the waste they prevent

Negatives can damage performance just as easily as they protect it. Here are the mistakes we see most often in account audits.

Overblocking high-intent research queries

Adding “review” as a negative sounds smart until you realize that people searching “CRM software reviews” are deep in the buying process. They’re comparing options. That’s high-intent traffic.

Negate “review” and the people you block are buyers.

Always check the conversion data on a modifier before negating it. If it converts, it stays.

Conflicting negatives that suppress your own keywords

This one is surprisingly common, and it follows directly from the cumulative rule above.

You add “management” as a negative broad, forgetting that your own keyword “project management software” contains that term. Your ad stops serving entirely for your core keyword, and no adjustment at a lower level will rescue it.

Google Ads won’t warn you about this conflict. You’ll only notice when impressions drop.

After adding negatives, check your keyword diagnostics within 48 hours to confirm your active keywords still serve. An enterprise-scale account should run periodic conflict audits as part of standard governance.

Ignoring brand vs. non-brand separation

If you don’t negate your brand terms in non-brand campaigns, your non-brand performance metrics are polluted. Brand traffic converts at a higher rate, so your non-brand campaigns look healthier than they are.

This leads to bad budget allocation decisions.

The fix is simple: negate your brand terms (and common misspellings) in every non-brand campaign. Do the reverse in brand campaigns.

The set-and-forget problem

A negative keyword list from six months ago reflects a six-month-old understanding of your query landscape.

Search behavior shifts. Your product mix changes. New competitors appear.

Negatives that made sense in Q1 might block profitable traffic in Q3. This is the most dangerous mistake because it’s invisible. Nothing breaks. You just slowly lose coverage on queries you should be winning.

A hand checking off items on a printed audit checklist, pen mid-mark

The negative keyword maintenance cadence

Discovery finds new negatives. Maintenance ensures existing ones aren’t causing harm.

Both need a schedule, and they’re different activities.

Weekly

  • Review search terms report for new irrelevant queries (sort by spend, filter by zero or below-threshold conversions)
  • Add new negatives at the appropriate level (ad group, campaign, or account)
  • Spot-check that top-spending keywords still serve impressions (catches conflicts early)

Monthly

  • Run a conflict check: compare your negative keyword lists against your active keywords and flag any overlap
  • Review ad group-level negatives for routing accuracy: are queries landing in the right ad groups?
  • Check overall PPC optimization metrics for campaigns where negatives were recently added, confirming you gained efficiency without losing volume

Quarterly

  • Audit your account-level negatives. Has anything changed in your product line or strategy that makes a previously universal negative now relevant?
  • Refresh competitor brand negatives (new entrants, rebrands, acquisitions)
  • Review the intent categories table above: are any “always negate” categories now conversion sources due to new offers or pricing changes?
  • Document your negative keyword governance: list naming conventions, who approves account-level changes, and how changes are logged

A naming convention for shared lists saves you future headaches. Something like [Level]_[Category]_[Date] (for example “Account_JobSeekers_2025Q2”) lets any team member understand what a list does without opening it.

Frequently asked questions

How do negative keywords work with Dynamic Search Ads (DSA) and other query-matching formats?

Negative keywords still filter queries in DSA, but you often need a tighter negative strategy because targeting is based on your site content rather than a fixed keyword list.

Pair negatives with page feed controls and regular search term reviews to prevent unexpected routing.

Should I use negative keyword lists (shared sets) or add negatives directly to campaigns and ad groups?

Shared lists are best for reusable exclusions you want consistent across multiple campaigns, like company-wide compliance or universal irrelevance.

Direct negatives are better when the exclusion is campaign-specific or used for routing between tightly themed ad groups.

How do I handle misspellings, plurals, and common abbreviations without bloating my negative lists?

Start by identifying the handful of variants that actually generate spend, then add only those that appear in your search terms data.

For scale, maintain a living “variant bank” for common spellings and abbreviations, then reuse it via shared lists where appropriate.

What is the safest way to test new negatives without risking a sudden drop in revenue?

Roll out changes in smaller batches, prioritize low-risk exclusions first, and document what you added so you can revert quickly.

After launch, monitor conversions and impression share for your top revenue drivers to confirm you removed waste without choking demand.

How do negatives impact reporting and attribution, especially when measuring incrementality?

Negatives can shift where conversions are credited by changing which campaigns and queries are allowed to trigger ads.

If you run incrementality tests, freeze negatives during the test window or log changes carefully so attribution shifts are not mistaken for performance changes.

How should international or multilingual accounts approach negative keywords?

Build language-specific negative lists because direct translations often miss local phrasing and slang.

Use country-level search term exports to identify region-specific irrelevant intent, then apply negatives at the geo-appropriate campaign level.

How do I decide whether to exclude a term or keep it and adjust bids, ads, or landing pages instead?

Exclude only when the intent is structurally wrong for your offer. Weak current performance on its own is a reason to keep investigating.

If the intent is close to your ICP, improving ad messaging or tightening landing page alignment is often more profitable than blocking it outright.

Stop the bleed, then build the system

Negative keywords are a living layer of your account that requires weekly attention, clear ownership, and honest evaluation of what should and shouldn’t be blocked.

The PPC managers who get the most out of negatives know when to add a term and when to leave it alone.

Start with your search terms report this week. Find the five highest-spend queries that never converted. Negate them at the right level with the right match type.

Then put a recurring calendar invite on your schedule, because the second round of waste is already accumulating.

If your negative keyword lists have gone stale, or your account structure makes governance difficult at scale, Single Grain’s paid media team can audit your setup and build a repeatable framework. Get a free consultation and stop paying for clicks that were never going to convert.