Negative keywords and the money they save

Google's job is to spend your budget. Yours is to decide what it may spend it on. Those two goals are close but not identical, and the gap between them is where your money goes.

Modern match types are loose. You bid on "drain cleaning" and Google decides that "how to clean a drain with vinegar" is close enough. It is close in language and nowhere near in intent. That person is not hiring anyone. They are standing in their kitchen with a bottle.

What do negative keywords actually stop?

Three categories of search, and they show up in every local account I have opened.

Research searches. Anything with how, why, DIY, tutorial, video, or diagram. These people are solving it themselves. Some of them will fail and call somebody, but you are paying for the whole population to catch the few.

Job seeker searches. Jobs, hiring, careers, salary, apprenticeship, near me hiring. In the trades this is a meaningful share of volume, and every one of those clicks is a person who will never buy from you.

Wrong service and wrong buyer searches. Free, cheap, parts, supply, wholesale, used, rental, and the name of a service you do not offer. Also the brands and products you do not carry. If you install one manufacturer and not another, the other one belongs on the list.

Add a fourth for anyone with a service area. City and neighborhood names outside your territory. Location targeting alone does not catch a person in your city searching for a plumber in a city two hours away, and that click is worthless to you. That failure has a bigger cousin worth reading about, since geographic targeting mistakes waste a large share of local budgets before negative keywords ever come into it.

What is a wasted click actually worth?

Price it, because the total is usually larger than the effort it takes to stop it.

Price a click. LocaliQ, which publishes annual search advertising benchmarks under the WordStream name, puts the 2026 average cost per click at $8.33 for Home and Home Improvement, $8.00 for Dentists, $4.62 for Beauty and Personal Care and $5.42 across all industries. Those are medians rather than means, LocaliQ sells advertising management, and the 2026 edition does not disclose its sample size. The 2025 edition of the series disclosed 16,446 US campaigns. Vendor research, and a placeholder until you pull your own average from the account.

Count the clicks. A $2,000 monthly budget in home services at $8.33 buys 240 clicks (derived).

Find your share. Sort your search terms report by cost and add up everything in the three categories above. Say it comes to 18% of spend. That is $360 a month (derived), $4,320 a year (derived).

Now price what that money would have bought instead. $360 buys 43 more clicks (derived) from searches that look like customers. At LocaliQ's 8.05% conversion rate for the category, that is 3 more leads a month (derived), and at a one-in-three close rate, one more customer (derived).

The saving is real but the redeployment is the point. Negative keywords do not just stop a loss. They move the same budget onto searches that can pay you back, which is why this is worth doing before any bid change.

I should say plainly that no published study with a disclosed sample measures what share of a local ad budget goes to junk searches. The 18% above is an input to an example, not a benchmark. Your report has your number in it, and that is the only version worth acting on.

Where do I find the ones that apply to me?

Not from a downloaded list. From your own account.

Generic starter lists are fine as a floor. They are written for everybody, which means they miss the specific waste in your business, which is where most of your money actually goes. The real list comes from the report showing exactly what people typed before they saw your ad, and reading the search terms report properly is the single most useful hour in a small account.

Sort that report by cost, descending. Work down from the top. For each term ask one question. If a customer walked in and said that sentence out loud, would I want them.

Not "could I sell them something." Would I want them, at my prices, for my capacity. Most owners are far too generous the first time they do this.

Adding a negative keyword is a decision to buy fewer clicks on purpose. That is uncomfortable and it is correct. A smaller number of better clicks against the same budget is the entire point.

Where does this mechanism break down?

In one place that surprises almost everybody, and it is documented by Google.

Negative keywords do not match close variants. Google's own help documentation states it directly and gives the example: if you exclude the negative broad match keyword "flowers," your ads will not serve on "red flowers" but can still serve on "red flower." Singular and plural are different negatives. So are misspellings, so are common abbreviations.

That is the opposite of how positive keywords behave, and it is why so many negative lists look thorough and leak anyway. Somebody adds "job" and never adds "jobs." Somebody blocks "repair manual" and pays for "repair manuals" all year.

Two practical consequences. Add the plural and the singular of every junk term. And check the report a month later specifically for the near-misses of things you already blocked, because those are where the money quietly kept going.

The second limit is one you cannot fix. Google omits search terms without enough query activity from the report, to keep with its stated data privacy standards. Those searches still spent your money and you will never see them written out. Search terms insights groups some of them into themes without exposing the queries. Work the terms you can see and accept that a share of the account is not visible.

How aggressive should I be?

More aggressive than feels comfortable when the budget is small, less when it is large.

A thin budget cannot afford to educate itself across a wide surface. Cutting hard concentrates the money on searches that look like customers. You will block a few searches that might have converted. At small volume that is a good trade, because the alternative is spreading thin enough that nothing gets enough exposure to work.

A large budget can tolerate looser matching because it has the volume to sort winners from losers over time. Even then, the research and job seeker categories should be cut without hesitation. They never convert.

One thing to be careful about is your own brand name and your competitors' names. Blocking competitor names is often right, since those searchers want that company. Sometimes it is wrong, because that is exactly the traffic you want to intercept. That decision deserves its own thinking, and bidding on brand names cuts both ways depending on whether you are the one being targeted.

How often does this list need attention?

Monthly at minimum, weekly when a campaign is new.

New campaigns drift fastest. Google is exploring, matching broadly, and finding new ways to interpret your keywords. The first month produces the largest crop of negatives you will ever add. After that it settles into maintenance.

A negative list that has not changed in ninety days is the clearest evidence an account is unmanaged. If you pay someone to run your ads, that modification date is a fair thing to check.

Structure matters too, not just the words. Google supports account-level negative keyword lists that apply automatically across eligible search and shopping inventory, which is the right home for terms that are junk everywhere, like jobs and salary. Campaign-level negatives are for terms that are junk in one campaign and valuable in another, like a service one campaign sells and another does not. Mixing those two up is how people accidentally block their own best keyword.

What to do this week

Open your search terms report, set the date range to the last thirty days, and sort by cost. Read the top fifty terms. That is twenty minutes.

Mark every one where the searcher was not a buyer. Add them as negatives, using phrase match for multi-word ideas and broad for single junk words like jobs or salary. Add the singular and the plural of each, because Google will not do that for you.

Add up the cost of everything you just blocked and write the annual figure on a sticky note. That is the number that justifies putting this on the calendar every month.

Then look at the terms you kept and check that the page they land on actually addresses them. If somebody searched for a service and the landing page opens with your company history, the negative keyword saved you money and the page is still losing it. What a first-time visitor needs to see in five seconds is the other half of the same job.

If a lot of your kept terms are city names you serve, that is a signal your site is thin on geography. Service area pages that work, versus the ones that get ignored is worth reading before you build twelve of them.

Questions about where your budget is going?

Email me at eric@seod.com with a screenshot or export of your last thirty days of search terms, sorted by cost. I will send back a starter negative list built from your actual data, marked up with why each one is on it, and the annual dollar figure it would have saved you.

That takes me about fifteen minutes and it is usually the cheapest improvement available in a small account. No pitch attached, and I will tell you if the list is already clean.

Or keep reading more on paid search for local businesses.