There's a report inside every Google Ads account that most people have never opened. It's not buried or hidden behind some advanced setting. It's right there in the interface, one click away from the keywords tab. It's called the search terms report, and it shows you the actual words people typed into Google before they clicked your ad and spent your money.
Most advertisers look at keywords. Keywords are the words you told Google you want to show up for. Search terms are what people actually typed. Those two things are not the same, and in most accounts we audit, they’re not even close. You might be bidding on "emergency plumber" but paying for clicks from people who searched "how to become a plumber" or "plumber salary in Texas." The keywords look fine. The search terms tell a different story.
We check this on every audit. It’s step 3 in our 12-point Google Ads audit because it’s where you find out what you’re actually paying for. Not what you intended to pay for. What Google decided you should pay for.
Why Your Keywords Aren’t What People Typed
Google Ads hasn’t worked on exact matches in years. Even if you use what Google calls "exact match" (the one with the brackets around it), Google will still show your ads for searches it considers to have the "same meaning or same intent" as your keyword. So your exact match keyword [roof repair] might trigger on "fix my roof," "roofing contractor," or "roof leak repair" without you ever telling it to. That’s exact match in 2026.
Phrase match is worse. Google’s official definition says ads may show on searches that "include the meaning of your keyword" and that "the meaning of the keyword can be implied." In practice, phrase match in 2026 behaves almost identically to what broad match used to be five or six years ago. Your phrase match keyword "emergency plumber" might trigger on "plumbing companies near me" or "24 hour pipe repair" because Google decided those searches share the same intent. Sometimes they do. Sometimes they don’t. You won’t know unless you look.
Broad match is the loosest of all, and it’s now the default when you add a keyword without specifying a match type. Broad match considers the user’s recent search history, the content of your landing pages, and the other keywords in your ad group to decide what’s "related" to your keyword. It will match on synonyms, semantic neighbors, and sometimes things that feel only tangentially connected to what you’re actually selling. Broad match for "roof repair" might show your ads when someone searches "home renovation contractors" or "how much does a new roof cost." That could be fine, or it could be a complete waste, depending on whether that person actually needs a repair or is just pricing out a full replacement they won’t do for three years.
What Google Doesn’t Show You
Google doesn’t even show you all of your search terms. Since September 2020, the search terms report only includes queries that were "searched by a significant number of users." Everything else gets lumped into a category called "Other search terms," which is Google’s way of saying "we spent your money on these but we’re not going to tell you what they were."
How much is hidden varies by account. Multiple analyses have found that anywhere from 30% to 50% of clicks and spend now come from queries you can’t see. Some accounts report up to 80% of their search term data being hidden. Google’s stated reason is privacy. They say low-volume queries might identify individual users. The practical effect is that you’re managing your account with incomplete information. You can only add negative keywords for queries you can actually see, which means there’s a chunk of your spend that’s essentially unauditable.
Google made a small improvement in 2024 by grouping misspellings under their correctly-spelled variant, which they estimated would increase visibility by about 9% in the average account. That helps, but it doesn’t change the fundamental problem. You’re still making decisions with incomplete data. Which is why the data you can see matters more than ever. If your visible search terms show patterns of waste, the hidden ones almost certainly have the same patterns at similar or worse rates.
What We Actually Find When We Pull the Report
We pull search terms two ways on every audit: sorted by cost (highest spend first) and sorted by impressions (highest volume first). Then we merge and deduplicate the two lists. The cost sort catches the expensive waste, queries that are burning real budget. The impression sort catches the silent damage, queries that aren’t spending much individually but are showing your ads thousands of times without clicks, dragging down your click-through rate and hurting your Quality Score over time. Most agencies only pull one or the other. Both matter.
On a recent multi-location account, we pulled 500 terms by cost and 500 by impressions. After deduplication, we had about 680 unique search terms to analyze across a 90-day window. The account was spending roughly $28,000 per month on search campaigns. The raw number: $9,800 in spend going to search terms that had zero conversions over 90 days. That’s the number most agencies would put in a report and call "waste." They’d say the account is wasting $9,800 a month, recommend adding 200 negative keywords, and position themselves as the hero who just saved you $10K.
That’s not how we do it. Because $9,800 in zero-conversion spend is not $9,800 in waste. Most of those terms are perfectly legitimate searches from people who just didn’t convert yet. At a typical 5-10% conversion rate, 90% or more of your clicks aren’t going to convert. That’s not a targeting problem. That’s a funnel. Someone searching "emergency plumber near me" who clicks your ad and doesn’t call isn’t waste. They’re a qualified prospect who didn’t convert on that visit. Maybe they’ll come back. Maybe they won’t. But the targeting was correct.
N-Gram Analysis: How We Find the Real Patterns
Reading through 680 individual search terms one at a time is not how you find waste. You’d spend hours staring at individual queries and miss the patterns that connect them. The methodology that actually works is called N-gram analysis. You break every search term into its component fragments (one-word, two-word, and three-word chunks) and then aggregate the spend and conversion data for each fragment across all the search terms that contain it.
Say you have 40 different search terms that all contain the word "salary." Individually, each one might have $12 or $30 in spend. Nothing that jumps out when you’re scrolling through a list. But when you aggregate, you find that the fragment "salary" accounts for $840 in total spend across those 40 terms with zero conversions. Now you have a pattern. One phrase match negative keyword, "salary," blocks all 40 of those queries with a single addition. That’s what N-gram analysis does. It surfaces the patterns that are invisible when you look at individual terms.
We run this at three levels. Single words (unigrams) catch broad categories: "jobs," "salary," "career," "DIY," "free." Two-word combinations (bigrams) catch more specific patterns: "how to," "near me" for the wrong city, a competitor's name. Three-word combinations (trigrams) catch the really specific stuff: "HVAC technician school," "plumber license requirements," "best deodorant for men." Each level tells you something different about what's going wrong with the targeting.
On that same multi-location account, the N-gram analysis surfaced these patterns in the first five minutes: "jobs" and "career" fragments totaling $1,100 in spend (people looking for employment, not services), a competitor’s brand name showing up in 23 different search terms totaling $640, and a set of city names outside the service area accounting for another $1,060. None of those individual terms would have stood out on their own. The fragments made the pattern obvious.
Geographic Waste Is Bigger Than You Think
Geographic waste is the one that surprises people the most because they assume their geo-targeting is handling it. You set a 15-mile radius around your office, so your ads should only show to people in that area, right? Not exactly. Google’s geo-targeting uses the setting "Presence or interest" by default, which means your ads can show to someone anywhere who Google thinks is interested in your target area. Even if you’ve changed it to "Presence only" (which you should), Google still interprets location loosely. Match type expansion means someone searching "plumber in Springfield" might trigger your ad even though there are 34 Springfields in the United States and yours is the one in Missouri.
We don’t estimate geographic waste by eyeballing city names. We calculate it. For every search term that contains a geographic reference (a city name, a neighborhood, a zip code), we identify which campaign it came from, find the coordinates of that campaign’s target location, find the coordinates of the mentioned city, and calculate the actual distance between them using the haversine formula. That’s the formula that accounts for the curvature of the Earth when measuring distance between two latitude/longitude points. It sounds academic until you realize it’s the difference between "this city is 8 miles away and probably fine" and "this city is 45 miles away and definitely not your customer."
On the account I mentioned, brand campaign search terms get excluded from this particular analysis. Not because geography doesn’t matter on brand (we still geo-target brand campaigns), but because branded search terms containing city names behave differently. Someone searching "Smith Plumbing Dallas" from a suburb 15 miles away is expected behavior, and brand campaigns have their own impression share targets that handle geographic coverage. The geo waste calculation is specifically for non-brand campaigns where someone searching a service plus a far-away city is genuinely wasted spend. For those non-brand campaigns, we flagged every search term containing a city that was outside the campaign’s targeting radius with zero conversions. The result: $1,060 in spend going to searches mentioning cities 20-60 miles away from any of the business’s locations. Not ambiguous "maybe they’d drive that far" distances. Genuinely out-of-area searches that had no realistic chance of becoming customers.
The Difference Between Flagged Waste and Defensible Waste
Most audit reports don’t make this distinction, and it’s the reason their "waste" numbers are inflated. On that $9,800 in zero-conversion spend, the breakdown after proper analysis looked like this:
Clearly wrong intent (defensible waste): $2,800. This includes competitor name searches ($640), verified out-of-area geographic searches ($1,060), job/career/employment searches ($1,100), and navigational queries for other websites. These are searches where the person was definitively not looking for what the business sells, or was looking in a place the business can’t serve. You can confidently call this waste and build negative keywords to block it.
Qualified traffic that didn’t convert (not waste): $5,200. Searches for services the business actually offers, in cities within the service area, from people who appear to have the right intent but just didn’t pick up the phone or fill out a form. "Emergency plumber near me" with zero conversions after 6 clicks isn’t waste. It’s a conversion rate problem, or just normal funnel behavior. You don’t negate "emergency plumber" because six people didn’t call.
Insufficient volume to judge: $1,800. Terms with 1-2 clicks and no conversion. There’s not enough data to know whether these are good or bad. They might convert next month. They might not. You can’t call them waste, and you can’t promote them to keywords. You leave them alone and check again in 30 days.
The agency that tells you they "found $9,800 in waste" is either exaggerating to close a sale or doesn’t understand conversion math. A 10% conversion rate means 90% of your clicks won’t convert, and that’s normal. The actual defensible waste on this account was $2,800 per month. Worth fixing, absolutely. But it’s a very different conversation than $9,800.
What Happens After You Negate the Wrong Stuff
Over-negation is a real problem and it’s almost as expensive as no negation at all. We’ve taken over accounts where the previous agency went on a negation spree and blocked terms that were actually bringing in qualified traffic. An accounting firm that adds "free" as a negative because most "free tax filing" queries are low-intent, except they actually offer a free initial consultation and people search "free tax consultation near me" before booking. A home services company that negates a nearby suburb because it’s "outside their radius," except it’s 4 miles away and people drive that far routinely.
This is why we crawl the business’s website before adding any negative keywords. If the business offers a service, we don’t negate searches for that service even if those searches have zero conversions. Zero conversions on a service you offer is a conversion problem. Maybe the landing page doesn’t mention that service, maybe the ad copy doesn’t match, maybe the scheduling widget is broken on mobile. The targeting was right. Something downstream is wrong.
Negating the keyword hides the problem without fixing it.
Same logic applies to geographic negation. We calculate actual distances, not approximate ones. A city that’s 6 miles from the office is not a negative keyword in most service businesses. People drive 6 miles. Zero conversions from a city 6 miles away tells you something is wrong with your ad or your landing page for that area, not that people from that city don’t want your service.
How to Check Yours
Go to your Google Ads account. In the left nav, click Campaigns, then Insights & Reports, then Search terms. You’re now looking at the actual queries people typed before clicking your ads. Sort by cost descending. Scroll through the first 100 terms. You’ll know within 30 seconds whether anyone has been managing this.
What to look for on your first pass:
- Competitor names. If you see your competitor’s business name showing up repeatedly, your ads are being triggered when people search for someone else. Unless you’re running a competitor campaign on purpose, this is waste.
- Job-related terms. "Careers," "salary," "hiring," "how to become a." People looking for employment, not services.
- Cities you don’t serve. If you’re a plumber in Austin and you see "plumber in San Antonio" burning budget, your geo-targeting or match types are too loose.
- Informational queries. "How to fix a leaky faucet" or "how long does a roof replacement take." People researching, not buying. These aren’t always bad, but they need to be in the right campaign with the right expectations.
- Other websites. "Yelp plumber reviews" or "angi.com HVAC." People navigating to other platforms, not looking for you.
If you see more than a handful of these in your top 100 terms by spend, nobody is managing your search terms. If you see zero negative keywords in the account, nobody ever has.
The real analysis (the N-gram work, the geographic calculations, the defensible waste distinction) takes tools and methodology that aren’t built into the Google Ads interface. But the first-pass eyeball test takes five minutes and tells you whether there’s a problem worth investigating.
Why This Is Step 3
In our audit framework, we look at search terms after conversion tracking (step 1) and campaign performance (step 2) because the search terms report is only useful if you trust the conversion data. If your conversion tracking is double-counting or missing actions entirely, your "zero conversion" search terms might actually have conversions you’re not seeing. And campaign performance context tells you which campaigns are actually spending money. There’s no point analyzing search terms on a campaign that spent $47 last month.
Search terms also feed directly into step 4, which is Quality Score. Every time your ad shows on an irrelevant query and nobody clicks, your Expected CTR takes a hit. That drags Quality Score down, which raises your cost per click on everything else in the account, including the searches that are relevant. So irrelevant search terms don’t just waste the budget they spend directly. They make the rest of your clicks more expensive too, and that cost accumulates quietly over months.
If your agency sends you reports that show keywords, CPAs, and click-through rates but never mention what people actually searched for, ask them to pull the search terms report. Ask them to show you the N-gram patterns. Ask them how they distinguish between real waste and normal funnel loss. If they can’t answer those questions, you’re paying for management that isn’t actually managing the one report that shows where your money goes.
This is one of the 12 steps in our full Google Ads audit process. If you want us to pull your search terms and show you exactly what you’re paying for, request a free audit and we’ll run the full analysis.
