A negative keyword strategy is a set of rules for which searches your ads should skip, where each exclusion lives, and how you check that it still makes sense. The goal is not the longest list. The goal is a list that cuts waste while leaving every converting query alone.
Most damage from negatives is quiet. A bad keyword shows up in a report. A bad negative just removes traffic, and nobody sees the conversions that never came. This post covers how to structure lists, how match types behave for negatives, the mistakes that block good traffic, and a review routine you can start tomorrow.
Start with the structure: three layers
Think of negatives as three layers. Each layer has a different job and a different risk.
Account level
Account-level negatives apply everywhere. Use them for terms that are wrong for the business no matter the campaign. Think of job-seeker terms, free-download intent for a paid product, or a product line you do not sell.
The risk is reach. A negative here can block a query in a campaign where it was a winner. Keep this layer short and boring. If you have to think about whether a term belongs here, it probably belongs one layer down.
Shared lists
Shared lists are reusable sets you attach to many campaigns. They work well for themes. A "research and learning" list, a "jobs and careers" list and a "competitor and brand confusion" list are common. You build the list once and attach it where it fits.
The upside is consistency. The downside is that one edit changes every campaign the list touches. Name lists by purpose, not by date or by person, so the next buyer knows what they are looking at.
Campaign and ad group level
This is where most of the precision lives. A term that is waste in one ad group may be the best query in the next. Ad group negatives are also how you steer traffic. If two ad groups could both match a query, a negative in one sends the query to the other.
Say a campaign has one ad group for "emergency" service terms and another for "scheduled" service terms. Adding "emergency" as a negative to the scheduled ad group keeps the right ads on the right queries. That is routing, not waste removal, and it is worth treating as its own habit.
Match types behave differently for negatives
This is where good intentions go wrong. Negative match types do not work like positive ones.
Negatives do not match close variants. A positive keyword can match plurals, misspellings and close forms. A negative generally will not. If you block "free" and the query is "freebie," the negative does nothing. You have to add the variants you care about.
Broad match negative. The ad is blocked if every word in the negative appears in the query, in any order. A broad negative of "cheap leather boots" will not block "cheap boots." It only blocks queries that contain all three words. People often expect it to behave like a wide net. It is narrower than that.
Phrase match negative. The ad is blocked if the query contains the words in that order. Use this when a concept is clearly wrong, like "how to" or "salary" for a service you sell to businesses.
Exact match negative. The ad is blocked only when the query matches exactly, with no extra words. Use this when you want to remove one query and protect its neighbors.
A simple working rule:
- Use exact match when the term is wrong only in that exact form.
- Use phrase match when the concept is wrong wherever it appears.
- Use broad match negatives sparingly, and only for multi-word negatives where you want all words to be present.
Always check the plural, the misspelling and the word-order variants of any phrase you block. You are building the net by hand.
The classic mistakes
Blocking terms that convert
This is the big one. Someone sees a few clicks with no conversions on a query, adds a negative, and moves on. Two clicks is not evidence. Before you add a negative, look at the query across a decent window, and check whether it has assisted conversions or leads that closed later.
Here is a quick check, as an example. Say a campaign has a $50 target CPA, and a query has spent $40 with no conversion. That is under one target CPA of spend. It has not had a fair shot. A common approach is to wait until a query has spent roughly two to three times the target CPA before calling it waste. Pick your own threshold, write it down, and apply it the same way every time.
Over-broad negatives
Single words are dangerous. "Free," "cheap," "used" and "jobs" all look like clean exclusions, until your offer includes a free trial, a budget tier, a certified pre-owned line or a staffing product. Before blocking a single word, search your own converting terms for it.
A useful test: take the proposed negative and run it against your last 90 days of converting queries. If anything that converted would have been blocked, narrow the negative or drop it.
Negatives that block your own ad group
It happens more than you would think. A negative in a shared list gets attached to a campaign that targets the very term. The ad group goes quiet, and the cause is hard to trace because nothing is "paused." Check for conflicts after any list change. Google flags some conflicts in the interface, but do not rely on that alone.
Copying a generic list
Public "master" negative lists are a starting point at best. They were built for someone else's offer. Use them as a prompt for ideas, then test each term against your own data.
Never revisiting old negatives
A negative that made sense two years ago may block a product you launched last spring. Lists age. Without a review, they slowly shrink your reach.
A build process you can run this week
Here is a repeatable way to build or clean up a negative keyword strategy for one account.
- Pull the search terms report for the last 30 to 90 days, depending on volume. If you want a routine for reading it quickly, this walkthrough on reading a Google Ads search terms report in 15 minutes a week covers it.
- Sort by spend, not by clicks. The queries that cost the most with nothing to show are your first candidates.
- Group the candidates by reason. Wrong intent, wrong product, wrong location, wrong audience, or routing between ad groups. Reasons become list names.
- Choose the layer. Wrong everywhere goes in the account list. Wrong for a theme goes in a shared list. Wrong for one ad group goes at the ad group.
- Choose the match type. Exact for one query, phrase for a concept, broad only when you mean all words.
- Run the conflict check. Compare each proposed negative with your converting queries and your active keywords.
- Write the reason. More on that next.
- Add the negatives and set a date to look again.
A worked example (illustrative numbers)
Say a campaign sells a business service with a $100 target CPA. The search terms report shows the following, with round numbers made up for the example:
- "service jobs near me": $180 spent, 0 leads. Wrong intent. Add "jobs" as a phrase match negative in a shared "jobs and careers" list.
- "how to do service myself": $120 spent, 0 leads. Research intent. Add "how to" as a phrase match negative in a shared "research and learning" list.
- "cheap service": $90 spent, 1 lead at $90. Under target. Leave it alone, even though "cheap" looks bad.
- "service reviews": $60 spent, 0 leads. Too little spend to judge. Watch it.
Two negatives added, one query protected, one query deferred. That is a good day's work, and every decision has a reason.
Document why each negative exists
This is the habit that separates a list you can trust from a list nobody dares touch.
For each negative, keep a short note with four things: the reason, the date, the evidence, and who added it. A spreadsheet works. A label or list name that encodes the reason works too. The format matters less than the habit.
Why bother? Three reasons.
- Handoffs. When a buyer leaves or an account moves, the new person can tell a deliberate negative from a guess.
- Reversals. If performance drops, you can check recent negatives against the drop and undo the right one.
- Consistency. If the same suggestion comes up again, you already know what you decided and why.
When you reject a suggestion to add a negative, write that down too. "Reviewed, left in because it converts at target" saves the next person the same hour of analysis.
Set a review cadence
Negatives need two rhythms.
Weekly: new search terms. Scan the latest search terms for waste and for routing problems. Keep it short. The point is to catch new junk early, before it eats a week of budget.
Quarterly: the lists themselves. Read through each list. Remove anything that no longer fits. Re-run the conflict check against recent converting queries. Look for variants you missed.
Trigger an extra review when something upstream changes. A new offer, a new landing page, a switch from phrase to broad match on positives, or a budget increase all change which queries show up and which are worth having.
A note on other channels
The same logic shows up elsewhere. Marketplace ads use negative product targeting to keep ads off irrelevant listings, and the same questions apply: what is wrong, how sure are you, and what would you lose if you were wrong? In organic search, "negative keywords" is a looser idea, closer to deciding which terms a page should not try to rank for. The mechanics differ, but the habit of writing down your reasons carries over.
FAQ
What is a negative keyword strategy?
It is the set of rules you use to decide which search terms your ads should never show for, and where those exclusions live. A good one covers list structure, match types, a record of why each negative exists, and a review schedule.
Should negative keywords be phrase or exact match?
Use exact match when you want to block one specific query and nothing near it. Use phrase match when a whole concept is wrong for you, like a job-seeker term. Use broad match negatives rarely, because they block only when every word appears, in any order, and that can still be wider than you expect.
How many negative keywords should an account have?
There is no right number. The count should follow from your search terms and your offer, not from a target. A small account with a tight offer may need a few dozen, and a large account with broad products may need far more.
Can negative keywords hurt performance?
Yes. A negative that matches a converting query removes that traffic, and you may not notice for weeks. Broad, undocumented negatives added at the account level are the usual cause.
How often should negative keyword lists be reviewed?
Look at the search terms report weekly for new waste, and review the lists themselves every quarter. Also review after any change to your offer, landing pages or match types, since those change which queries are good.
What Ads Arsenal does for this today
Negatives are a judgment call made over and over, across many accounts. That is the part Ads Arsenal helps with, and it is built for teams like the ones in Ads Arsenal for agencies.
On Google, it reads the account through the way you run it. You write your goals and constraints in plain English, including what it must not touch, and it reads every account through that lens. Its Insights queue ranks what to change, and the types of insight relevant here include search terms slipping past an existing negative, low-intent terms, and landing pages that cover too little of an ad group's search intent. Every recommendation comes with the reasoning and the numbers behind it, in plain English.
Nothing changes in an account until someone approves it. If a suggestion does not fit how you run, dismiss it with a sentence on why, and that becomes a rule for the account. You can take it back later. Approved changes are written to the account, checked a few minutes later to confirm they landed, tracked to see whether they worked, and can be rolled back with one click.
Ads Arsenal is in a free design-partner beta for agencies and in-house teams running real spend on Google, Meta and TikTok. If that sounds like your setup, Become a design partner.