Google Ads negative keyword lists that actually work
What separates a working Google Ads negative keyword list from one that quietly drains good traffic. The patterns that hold up over time.
A working negative keyword list does not look like a great list when you scroll through it. It looks small, organized, and a little boring. The dramatic-looking lists, the ones with two thousand entries and elaborate category prefixes, are usually the ones doing the most damage.
This piece is the patterns that hold up after looking at a lot of accounts. None of them are clever. The good lists win on consistency and restraint.
Pattern one: small lists with clear names beat one giant list
The single biggest predictor of whether a negative keyword setup is working is whether the lists have meaningful names and a consistent scope per name. "Account-wide DIY blocks" is a name that tells you what is in the list and when to update it. "Negatives 2024" is a name that means almost nothing six months later.
Three or four named lists, each with a clear single purpose, is the right scale for most accounts. Examples that recur across well-organized accounts:
- Account-wide info-seeker filter. How to, what is, tutorial, guide, fix, replace, repair (when you sell service rather than parts).
- Account-wide job and category filter. Salary, jobs, schools, training, certification.
- Out-of-area geo blocks. Cities and regions you do not serve. Grows over time.
- Brand and competitor adjacents. Specific competitor names you have decided not to bid against, plus phrases that come up in comparison searches.
These four cover most of the universal junk. Industry-specific blocks live on the campaigns themselves.
takeaway
Lists with meaningful names get audited. Lists with throwaway names get ignored, then grow into chaos. The name is half the discipline.
Pattern two: prefer phrase blocks for ambiguous words
There is a temptation, especially after the first sweep, to broad-block every word that ever appeared in a junk query. Resist this. Single-word broad blocks are powerful and dangerous; they catch every variant of the word, including the ones you have not seen yet.
The rule of thumb is: if the word has any reasonable use in a buyer query, phrase-block instead of broad-block.
"Free" is a clean broad block. Buyers do not use the word "free" to describe paid services.
"Cheap" is a borderline case. Some price-sensitive buyers do use the word cheaply, but most who type "cheap [service]" are tire-kickers. Most accounts can safely broad-block "cheap" with minimal collateral damage. A few cannot.
"Used" is a phrase block in almost every account. "Used" appears in valid queries (used car parts near me in a parts retailer, used as bait in a fishing supply store) often enough that broad-blocking it sweeps up real buyers along with the junk.
"Wholesale," "commercial," and "residential" are phrase blocks for businesses that serve only one of those categories. Broad-blocking the wrong one of these will break the campaign in subtle ways.
- −free → broad (almost always)
- −cheap → broad (most accounts)
- −salary → broad (always)
- −diy → broad (always)
- −jobs → broad (always)
- −wholesale → phrase ("wholesale supply")
- −used → phrase ("used parts")
- −training → phrase ("training certification")
- −commercial → phrase if you serve residential only
- −parts only → phrase (the word "parts" alone is fine)
Pattern three: delete duplicates before they multiply
Most accounts that have been running for a year or more have duplicates: the same negative on three campaigns, plus once in a shared list. The system handles duplicates fine; the human reviewing the lists does not.
Once a quarter, dump every negative in the account into a spreadsheet via Editor's export. Sort by keyword. Anything appearing more than once is a candidate for promotion to a shared list and removal from the per-campaign locations.
Duplicates also accumulate as a side effect of legitimate activity. You audit campaign A and add a negative. You audit campaign B and add the same negative because you forgot it was on A. Six months later both have it, plus the shared list, plus a third copy from the colleague who didn't know either of you had added it.
The collapse step is mechanical. The audit step that produces the data takes ten minutes. The collapse takes another fifteen. Doing it once per quarter keeps the lists clean enough that the next audit goes faster instead of slower.
Pattern four: do not block what you have not paid for
The most common over-blocking pattern: somebody reads a "best practices" article, sees a list of "top 100 negative keywords for [industry]," and bulk-imports the list into a campaign. Two weeks later the campaign is starved of impressions for reasons nobody can pin down.
The negatives in those generic lists are not wrong. They are just not necessarily applicable to your account. A residential plumbing campaign in Phoenix has a different junk pattern than a residential plumbing campaign in Boise, even though the industry is identical.
The right pattern: block what you can see in your own search terms report. Generic guides and templated lists are suggestions, never the source of truth. Your account is the data; everything else is hint.
The exception: starter lists for brand-new accounts that have no data yet. In the first thirty days, before you have a real search terms report, applying a known set of universal junk-words (free, cheap, diy, salary, jobs) is fine. Just plan to audit it after the first month and remove anything that turns out to be over-aggressive.
- 200 entries copy-pasted from an article
- Some apply, some block real buyers
- Hard to audit because you did not pick them
- Removing any feels risky
- 40 entries each backed by a search-terms-report row
- Every block is justified
- Easy to audit, easy to remove if circumstances change
- Each block keeps saving the same dollars per month
Pattern five: keep a journal
Two paragraphs from now you will forget half of what you blocked. Six months from now, somebody will ask why a particular query is not showing up, and you will spend twenty minutes finding out it was a negative you added without noting why.
A journal does not need to be elaborate. A shared doc with three columns: date, term, reason. One line per non-obvious block. Single-word blocks for "free" and "salary" do not need journal entries because the reason is obvious from the term. Phrase blocks for "wholesale supply" or "best free trial" do need entries because in six months you will not remember why you decided phrase over broad, or why you scoped that phrase that way.
The journal pays back the most when you are debugging an under-performing campaign. The first thing you check is "did we block something we should not have," and the journal turns that question from a half-hour search into a thirty-second scan.
Pattern six: review old negatives once a year
Most negatives age well. Add "salary" once and it stays useful forever. Add "diy" and it stays useful forever. The exceptions are the ones tied to a specific business state at the time of the block.
Examples that go stale:
Geo blocks for cities you now serve. You blocked Tucson when the business only served Phoenix. Two years later, the business expanded to Tucson, but the geo block is still in place. Result: the new campaigns silently exclude Tucson searches.
Brand competitor blocks for competitors who pivoted. You blocked a competitor's brand name. The competitor is now a different business or no longer exists, but the block is still in place. Usually harmless, occasionally not.
Service-line blocks for services you now offer. You blocked "drain cleaning" when the business only did emergency repairs. Two years later, drain cleaning is half the revenue, but it cannot get an ad served because of the block.
The fix is a once-a-year audit of negatives older than twelve months. Most of them stay. A handful, every year, need to come out because the business shifted under them.
Pattern seven: write down the policy, even if it is for one person
The smallest accounts can get away without written policy. A single account owner who does the work themselves will remember what they meant. Past two campaigns or two contributors, written policy starts to matter.
A negative-list policy fits on a single page. It covers four questions:
- Which lists exist, and what is each one's scope.
- Which match type to default to in which situation.
- Where new negatives go (shared list versus campaign versus ad group).
- How often each list is audited and by whom.
That document, read once and updated quarterly, prevents most of the drift that turns clean accounts into messy ones. A team of two people without the document will diverge inside a month. A team of two with the document will stay aligned for years.
What does not matter as much as people think
Two things people optimize for that, in our experience, do not move the needle.
Perfect categorization of negative reasons. Some accounts use elaborate category labels in their negative-list naming: "DIY-tier-1," "DIY-tier-2," "Information-Surface," "Information-Deep." This is over-engineering. Two or three flat categories per account are enough. The categorization tax climbs faster than the audit benefit.
Match type optimization for every entry. It is fine to default to broad for unambiguous single words and phrase for everything else, and only optimize the few entries where the choice matters. Tuning every match type carefully on every entry produces marginal gains over the simpler default.
The point of the list is to block junk so the rest of the account can focus on real buyers. The list is infrastructure. Boring lists work; clever lists usually drift.
Three or four small named lists. Phrase blocks for ambiguous words, broad blocks for unambiguous ones. No duplicates. Block what your data shows. Generic templates do not replace your own search terms report. Keep a journal. Audit old negatives yearly. Write the policy down. The boring lists are the ones that work.
Reader questions
- How big is too big for a single list?
- There is no hard rule, but past about two hundred entries the list becomes unauditable. Once a list crosses that line, the right move is splitting it into named sub-lists by category rather than continuing to add to one big bucket.
- Should I clean out old negatives?
- Yes, but rarely. About once a year, audit any negatives older than twelve months and ask whether they are still blocking what you intended. Most stay; a few have outlived their usefulness or were over-aggressive in retrospect.
