Which Search Terms Waste My Google Ads Budget, and What Should I Block?

The search terms you can see are the easy part of wasted Google Ads spend. Your search terms report is full of clicks you never meant to buy: job hunters, freebie seekers, people looking for a competitor's login. The report's rows also don't add up to what the campaign spent. Adding negatives one at a time rarely ends the problem, and a careless negative can quietly switch off searches that bring customers.

The harder waste sits in two places: queries the report hides, and buyers your own negatives block. So block irrelevant intent on sight, and judge relevant terms against your cost per conversion. Keep every negative as narrow as the problem it solves. Over-blocking costs more than under-blocking, and it never shows up in a report.

How much spend can the report actually show?

Less than the campaign total, and Google won't say how much less. Since September 2020, the Google Ads search terms report has listed only queries that "a significant number of users searched for." That wording comes from the account alert Search Engine Journal reported at the time, and Google never defined "significant." You still pay for every click behind the hidden queries. Their cost sits in the campaign totals and in no row of the report.

You'll find percentages online claiming to size the gap. None of the ones we've reviewed names its study, sample or period. A single average would mislead even if one existed, because the hidden share depends on how your account is built. Three things push it up:

• Broader match types, which reach phrasings too varied to repeat often.
• Long, specific queries, which are rarely typed twice.
• Low volume, which leaves too few repeats to clear the threshold.

Two campaigns in the same account can sit far apart.

So measure your own. Take one campaign over a full month and subtract the summed cost of the report's rows from the campaign's cost. The difference is your hidden spend. Nothing Google shows you splits that gap by cause, so treat it as an upper bound on threshold-hidden spend.

Then do the step most audits skip. Divide hidden spend by hidden conversions, and compare that hidden CPA with the CPA of the visible terms:

• If the two are close, the gap is a reporting cost, not a performance problem.
• If hidden conversions are near zero, you're funding a blind spot that doesn't pay back.
• If the hidden CPA is better, the invisible traffic may include your most precise searches. This happens more than people expect in B2B accounts, and tightening match types would throw those searches away.

That result shapes everything that follows. Every term-by-term method works on the visible rows only, so the gap marks what term-by-term work can never reach. You can't exclude a query you never see. Your defense against hidden waste is a negative written as a pattern, which catches low-volume variants whether or not they surface, plus match-type changes judged against the hidden CPA.

When you report waste upward, say which denominator you used. A fifth of the visible rows and a fifth of the campaign are very different claims. The full method, including a clicks-only workaround, is in how much of your Google Ads spend is hidden from the search terms report.

When has a search term spent enough to block?

The first question isn't about spend at all. Some searchers will never buy from the ad they clicked:

• job seekers typing "careers" or "salary"
• people looking for a free version or a DIY method
• someone looking for a competitor's login page
• an existing customer searching for support documentation
• anyone searching for a product or service you don't offer

Exclude these the moment you see them, even at a single click. Waiting for them to cross a threshold just pays for proof you already had.

Most advice on how to find negative keywords stops there. The harder calls sit in the grey zone, where intent can't be read from the words. "How to" queries come from people doing it themselves and from buyers researching a purchase. A competitor's brand can overlap with a generic word for what you sell.

These terms earn a spend test, and a fixed dollar figure is the wrong test. Take $60 with no conversion. In an ad group that converts at $25, that's more than two fair chances, missed. Where a qualified lead costs $400, it's a handful of clicks that prove nothing.

The line that works is relative. Start from the ad group's typical cost per conversion, not the account average, and apply three bands:

• Below one CPA with no conversions: the term has told you nothing yet. Leave it alone.
• Between one and two CPAs: put it on a watch list, and check whether the landing page matches the query.
• Past two CPAs, with no conversion in its full history: it's a strong candidate for a negative.

Widen the bands if your conversion rate is low, or if one conversion is worth far more than average. Judge only spend old enough to have converted, because long sales cycles lag.

That last point is where careful practice collides with a common shortcut. Some automated rules negate terms that have clicks but no conversions after a week. In a B2B account, those rules will block queries whose pipeline simply hasn't matured. A CPA-based line already allows for this, and a calendar-based one doesn't.

Some terms cross the line and should still stay live:

• Several high-intent terms fail at once. Suspect your conversion tracking or landing page before the queries.
• A relevant term converts, but too expensively. Isolate it as its own exact-match keyword at a lower bid.
• You genuinely can't tell. Leave the term running. One wasted term costs a few clicks, while a badly scoped negative can block every converting variant that shares a word with it.

The bands and the exceptions are worked through in how much a search term should spend before you block it.

Which match type should a negative use?

Phrase, most of the time. But first understand that negative match types are narrower than the keyword match types they share names with. None of the three negative types matches close variants, so a broad match negative doesn't reach synonyms, plurals or misspellings. Here is what each one does:

• A broad negative blocks queries containing the exact spellings of all its words, in any order.
• A phrase negative blocks queries containing its words in that order, with anything around them.
• An exact negative blocks one query and nothing else.

So "used laptop" walks past a negative for "used laptops," whatever its match type. You have to add plurals and misspellings yourself.

Phrase wins as a default because most waste is a concept buried inside varied queries: "free," "jobs," a city you don't serve. A phrase negative on that fragment blocks it wherever it turns up, including the low-volume variants the report hides. That's the pattern defense the hidden-spend problem calls for. For single-word negatives, broad and phrase block exactly the same queries, because there's no order to respect.

The other two types have narrower jobs. Exact negatives handle two things:

• Routing a query. An exact negative for [business laptops] in your broader ad groups sends that query to the ad group built for it. The broader groups keep longer variants like "business laptops with long battery life."
• Cutting one loser. An exact negative removes "laptop deals" without touching "gaming laptop deals."

Broad negatives suit word combinations that are wrong in any order but fine apart. "Laptop repair" is a good example for a retailer that doesn't offer repairs.

There's a real tension here, and it's worth stating plainly. The case for phrase is coverage. The case against short phrase negatives is collateral damage. A commercial cleaner adding "home" to stop residential enquiries also blocks "care home cleaning contract," which may be the most valuable search the account ever sees.

Both points are true. The way to reconcile them is through the word you choose, not the match type. Phrase is right when the fragment is wrong in every form a customer could type it. When a word's neighbors might convert, use a longer phrase or go exact. One-word negatives carry the most risk, so before saving one, write down three searches it would block.

Then run every candidate against your recent converting search terms. If it would have blocked one of them, tighten it. The full decision sequence, from isolating the fragment to the collateral-damage check, is in choosing between broad, phrase and exact negative match types.

Where should each negative keyword live?

As wide as the query is wrong, and no wider. Each placement reaches a different set of campaigns:

• A campaign or ad group negative affects only the campaign or ad group it sits in.
• A shared negative keyword list applies to every campaign attached to it, and does nothing until it's attached.
• The account-level list needs no attachment. Google's help documentation on account-level negative keywords says it applies across Search, Performance Max, App, Shopping, Smart and Local campaigns, and new campaigns inherit it automatically. At the time of writing, it's capped at 1,000 negatives per account.

That reach is why the account list deserves a high bar. Only terms that are wrong for any campaign you could plausibly launch belong there. "Jobs," "careers" and "salary" usually qualify.

"Free" often doesn't, because it may describe your best prospect if you run a free trial. "Cheap" is wrong for a luxury line and fine for next season's outlet campaign. Put either at account level and you've made that decision for every campaign, Performance Max included. The cap argues for restraint too, since plurals and misspellings each need their own entry.

Shared lists should carry most of your exclusions, and most accounts underuse them. They hold anything wrong for a definable group of campaigns. Competitor names might be excluded everywhere except a conquest campaign. "Student" might be excluded from prospecting but not from education pricing.

Split shared lists by theme: wrong intent, competitors, brand safety. A list attaches to a campaign as a whole, so a catch-all list forces every campaign to take all of it or none of it. Name each list and give it an owner. For an agency, a standard set of lists per client stops exclusions drifting between campaigns built months apart.

Campaign and ad group negatives handle routing: queries that aren't bad, just landing in the wrong place. Brand terms added as negatives in non-brand campaigns stop the two campaigns bidding against each other. They also stop cheap brand conversions flattering the non-brand numbers.

The same logic works one level down. "Blue" as a negative in a generic shoes ad group sends "blue shoes" to the group built for it. Put "blue" at account level instead, and you've blocked the traffic that ad group exists to win.

When a term comes out of the report, ask three questions in order:

1. Would any campaign ever want this click?
2. Is it wrong for a definable group of campaigns?
3. Is it fine but landing in the wrong place?

Defaulting to the narrowest placement that works is the safer way to be wrong.

One check before you rely on a negative for Performance Max. The account list is the placement Google documents as covering it, so confirm in the account that any other negative actually reaches it. The worked placement rules are in whether a negative belongs on the account, a campaign or a shared list.

Are your negatives blocking real customers?

Quite possibly, especially in an older account. A negative always wins. When a query matches one, your ad can't show, however well your keywords match and however high you bid. Nothing gets paused or disapproved. The keyword just goes quiet, which looks exactly like falling search volume.

The causes are usually reasonable decisions that stopped being reasonable:

• Inherited lists built for a product mix you no longer sell.
• Shared lists on the wrong campaigns. The efficiency that spreads one fix spreads one mistake just as far.
• Campaign negatives that collide with deliberate keywords elsewhere in the account.
• Over-eager cleanup, where one irrelevant query prompted a phrase negative that wiped out a converting theme.
• Automated no-conversion rules running in long sales cycles.
• A business that moved, so last year's irrelevant term is this year's service line.

Blocked searches never generate an impression, so the audit has to work backwards:

1. Export every negative, every active keyword and every shared list, with its campaign attachments.
2. Check each negative against the keywords it would collide with under its own match type.
3. Pull converting search terms from a window covering your full sales cycle, including the period before recent list changes. Test them against your current negatives. Any converting query your negatives would now block is a finding.

Treat that count as a floor. Because of the hidden share, you never saw every query you paid for.

Preventing new damage means measuring every change. Compare each campaign before and after a list change, over a full conversion cycle, rather than looking at the blended account. Track spend, impression share and qualified conversions, meaning the ones that become pipeline. Then read the result:

• Spend falls and qualified volume holds: the negative is working.
• Qualified volume falls while spend barely moves: you've cut demand, not waste.

In B2B, that judgment needs CRM data joined to the query, because the platform's count stops at the lead. SproutMe Companion reads Google Ads performance alongside HubSpot pipeline in one daily brief. A campaign that loses qualified volume after a list change surfaces there as something needing attention.

Two habits help whatever tooling you use. Have a second reviewer approve any shared or account-level negative. Keep a short log of what was added, where and why, so a volume drop can be traced in minutes rather than days. The full audit is in whether your negative keywords are blocking searches that bring customers.

Conclusion

Wasted spend in Google Ads search terms comes from three places, and only one of them appears in a report. The junk you can see is solved by blocking irrelevant intent on sight and giving relevant terms one to two CPAs to prove themselves.

The other two problems are invisible. One is the queries the privacy threshold hides. The other is the buyers your own negatives turn away. The same discipline handles both:

• Write negatives as patterns.
• Keep each one as narrow as the problem it solves.
• Place it no wider than the query is wrong.
• Test every change against what already converts.

To reduce wasted ad spend durably, measure both directions. A campaign getting cheaper and a campaign getting quieter can look the same until you check qualified volume.

See how Companion answers questions like which search terms have spent more than twice their ad group's CPA without converting, grounded in your live Google Ads spend.