Negative keywords and the money they save.

Every Google Ads account is paying for searches that were never going to convert. Not through incompetence — through the ordinary drift of broad match doing what broad match does.

Negative keywords are how you stop it, and the work sits in a report most advertisers open once a quarter.

What a negative keyword is

A negative keyword prevents your ad appearing for searches containing that term.

Regular keywords tell Google what you want to match. Negative keywords tell it what to ignore. If you sell premium leather boots, “free” and “cheap” and “repair” are all searches you can safely never pay for again.

Where they hide: the search terms report

The distinction that matters most in paid search is between the keywords you bid on and the search terms people actually typed. They are not the same, and the gap between them is where money goes missing.

The search terms report shows the second. It sits under the Insights and reports menu, and it is the most under-read screen in the platform.

Sort by cost, highest first, and read down. You are looking for anything with the wrong intent:

• Free, cheap, DIY, how to make — people who have decided not to pay anyone
• Jobs, careers, salary — people who want to work for you
• Repair, replacement parts, manual — existing owners, not buyers
• Competitor brand names — sometimes worth bidding on deliberately, mostly not
• Wrong product entirely — “boot” pulling car boots, or Boot the software

That last category is the reason to read search terms rather than reason about them from an armchair. The genuinely expensive mismatches are almost always ones nobody would have predicted.

Match types, which behave differently here

Negative keywords use the same three match types as positive ones, and this is the part that trips people up.

Negative broad blocks a search only if it contains every one of your words, in any order. running shoes as a negative broad does not block “running” alone or “shoes” alone. It blocks “cheap running shoes for men”.

Negative phrase blocks searches containing your words in that order. "running shoes" blocks “buy running shoes” but not “shoes for running”.

Negative exact blocks only that precise search. [running shoes] blocks nothing else at all.

Then the rule that costs people real money:

Negative keywords do not match close variants. Positive keywords do. Add shoe as a negative and searches for shoes still run. Plurals, misspellings, and spacing variants each need adding separately.

Default to negative phrase for most exclusions. Broad is easy to over-apply and exact is usually too narrow to be useful.

Account level or campaign level

Some exclusions are universally wrong for the business. Free, jobs, DIY, salary. Those belong in a shared negative keyword list applied at account level, so you write them once.

Others are only wrong in context. A retailer selling both men’s and women’s ranges wants “women’s” excluded from the men’s campaign, and definitely not excluded everywhere.

The test: is this term wrong for the entire business, or only for this campaign? Getting that wrong in the generous direction is how people accidentally switch off working traffic.

How often

Weekly for the first month of any new campaign, when the search terms report is at its most revealing. Monthly after that, once patterns settle.

Half an hour a month is usually enough to keep an account clean, and it compounds — every negative you add keeps saving money for as long as the campaign runs.

Where it stops working

Negative keywords are a subtraction tool. They stop waste. They do not create demand, and they cannot rescue an account whose fundamentals are wrong.

If you are running at 1.5X ROAS, pruning search terms will not get you to 4X. That gap is structural — how the account is organised, and what happens after the click. Nocs Provisions had a 0.5X account; no amount of negative keyword work would have fixed it, because the site converted at 0.8%. Conversion had to move first.

There is also a real over-correction risk. Exclude too aggressively and you starve campaigns of the volume Smart Bidding needs to optimise, which makes performance worse in a way that is hard to attribute later.

The rule I would keep: if adding a negative would exclude a term that has ever produced a sale, do not add it. Cheap traffic that occasionally converts is still traffic that converts. Only exclude what has demonstrated it cannot.