Seven common negative keyword mistakes in Google Ads

Negative keywords are forgiving in the sense that you can usually undo what you did. They are unforgiving in the sense that some of the most damaging mistakes are silent for weeks or months before anyone notices.

This piece is the seven mistakes we see most often in account audits, with the diagnostic for each and the working fix.

Mistake one: broad-blocking words that have legitimate uses

The mistake: someone adds the word parts as a broad-match negative because most "parts" queries are wholesale shoppers and the campaign is for service. Six months later the campaign is mysteriously narrow. The diagnostic is in the search terms report: queries that contain "parts" as a normal word ("replacement parts service," "parts pickup near me") are not appearing because the broad block sweeps them all.

The fix: change the broad block to a phrase block. Block "parts only" or "parts wholesale" instead of the bare word. The phrase block catches the actual junk pattern while letting "parts" survive in valid contexts.

The diagnostic: review every broad-match negative you have for words that have multiple meanings. Common offenders: parts, training, used, schools, training, repair (if you sell parts), sale, deal.

Mistake two: putting account-wide negatives at the campaign level

The mistake: you add the same fifteen negatives to every campaign in the account because it is easier to add as you go than to organize a shared list. Six months later, every campaign has fifty entries, half of which are duplicated across campaigns, and adding a new universal negative requires updating every campaign.

The fix: build three or four small named shared lists (or use the account-level negative list for true universals), and migrate the duplicates from per-campaign to shared. The migration is a one-time half-day job that pays back across the rest of the account's life.

The diagnostic: dump your negatives via Editor's export, sort by keyword, count duplicates. Above 30% duplication, it is time for a shared-list cleanup.

Mistake three: bulk-importing a generic "best negatives" list

The mistake: you copy two hundred entries from a "top 100 negative keywords for plumbers" article and bulk-apply them to your campaigns. Two weeks later, the campaign is starved of impressions and you are not sure why.

The fix: undo the bulk import. Use Editor to remove the imports en masse if you saved the source CSV; otherwise manually walk through the list and remove the entries that are blocking real buyers.

The diagnostic: any time impressions drop sharply within a week of a bulk import, the import is the most likely cause. The cure is removing it. Adding more negatives to compensate just buries the original mistake under a second one.

The right pattern: build the list from your own search terms report, one entry at a time, justified by a real row. The forty audited entries from your own data outperform the two hundred entries from a generic guide every time.

Mistake four: treating poor performance as a sign to add negatives

The mistake: a query did not convert this month and you add it as a negative. Six months later, you discover that the query was actually a buyer pattern that just had a slow month, and the negative has been blocking real buyers ever since.

The fix: change the criterion for adding a negative. The right test is intent. Recent performance alone is the wrong test. A buyer query that did not convert is not a negative; it is a query that needs better ad copy or a better landing page. A non-buyer query is a negative.

The diagnostic: look at your most recent twenty negatives. For each one, ask: would a reasonable person reading the term agree that the searcher was not going to buy? If three or more fail that test, your criterion has drifted toward "did not convert" and away from "wrong searcher."

Mistake five: forgetting the close-variant gap

The mistake: you add cheap as a broad-match negative and the same junk traffic keeps coming in. You assume the system is broken and add more negatives, none of which work.

The cause: the actual queries are using close variants of "cheap" that broad-match negatives do not catch. Queries like "cheaply done plumbing" or "cheaper plumbing service" share a stem with "cheap" but are different tokens. Broad-match negatives only catch the exact token.

The fix: enumerate the variants. Add cheap, cheaply, cheaper, cheapest, inexpensive, low cost, budget, and any other synonym you see in the report. Negative match does not expand to synonyms or close variants the way positive match does.

The diagnostic: any time a negative seems to "not be working," look at the actual query. Usually the query contains a variant of the blocked word rather than the literal token you typed.

Mistake six: not checking which level the negative was applied at

The mistake: you add a negative at the ad group level meaning to block account-wide. Six months later, the same junk is appearing because most of the campaigns are not seeing the block.

The fix: re-apply at the right level. Move the block from the ad group to a shared list (or the account-level negative list), confirm it now applies to every campaign, and remove the redundant ad-group entry.

The diagnostic: pick three negatives at random from your account. For each, check the level. If any are at the ad group level when they should be account-wide, your structure has drifted. The opposite is also possible: shared-list blocks that should have been ad-group-specific and are silently breaking related campaigns.

Mistake seven: never auditing old negatives

The mistake: a year ago you blocked Tucson because the business only served Phoenix. The business expanded to Tucson six months later. The Tucson block is still in place, silently excluding Tucson searches from the new service area's campaign.

The fix: a yearly audit of any negative older than twelve months. Most stay; a small number need to come out because the business shifted under them.

The diagnostic: the strongest signal is when impressions in a recently-launched campaign or service area are lower than the launch plan expected. Before assuming the bid is too low or the audience is wrong, check whether an old negative is silently blocking the new traffic.

The audit itself is straightforward. Sort negatives by date added (Editor exposes this). Anything older than twelve months gets a one-line review: is this still the right block. Most are. The few that have outlived their usefulness come out.

What good looks like, after avoiding all seven

Accounts that have avoided these mistakes share a recognizable shape. The negative lists are small, named clearly, and audited at least yearly. Universal blocks live at the account or shared-list level; campaign-level entries are the exception. Match types are mostly phrase for ambiguous words and broad for unambiguous ones, with very few negative-exact entries.

The most telling sign is that the operator can answer, for any given negative, both the reason it was added and the date. The journal exists. The discipline of writing it down keeps the structure honest.

If your account does not look like that today, the seven fixes above are the path to it. None of them requires more than an afternoon. The compounding benefits start the first week and last for the life of the account.