Ask to see an account’s negative keywords and you’ll usually get a list. Somewhere between forty and four hundred terms, added in bursts, most of them dating from whenever someone last got annoyed at a search terms report. It worked when it was built. It hasn’t been built since. That’s the difference between a list and a system, and it’s the difference between an account that leaks a little more every month and one that doesn’t.
Why a One-Time List Decays
Search behavior moves. New products launch with names that collide with yours. A competitor rebrands. A phrase that meant one thing in January means something else by June because a TV show used it. Google’s matching gets looser with every update, and the broad match drift that comes with it means the set of queries you’re being matched to is never the set you approved.
A list built once is a snapshot of the problems you had on that day. Every week after, new irrelevant queries arrive that the list has never seen, and unless someone is looking, they spend.
The Three Layers
The structure that holds up has three layers, each with a different job.
Account-level negatives are the terms that should never trigger an ad anywhere: “free,” “jobs,” “careers,” “salary,” “definition,” “reddit,” “login,” competitor names you’ve decided not to bid on. These live in a shared list applied to every campaign, and they change rarely.
Campaign-level negatives keep campaigns from stealing from each other. If one campaign targets “seo consultant” and another targets “seo tools,” each needs the other’s core terms as negatives, or Google will route queries to whichever ad group it prefers, which is rarely the one you’d pick. This layer is about routing, not waste.
Ad group-level negatives are the sculpting layer, and the one that gets overused. They exist to keep a specific query going to a specific ad. If you find yourself adding dozens here, the ad group structure is probably the real problem.
The Cadence Is the System
A negative keyword system is mostly a calendar entry. Weekly for accounts spending more than a few thousand a month, every two weeks below that. The review itself is fifteen minutes:
- Open the search terms report for the period, sorted by spend. Not by impressions, not by clicks. The queries that cost the most and converted least are the only ones that matter this week.
- Sort each bad query into a layer. “Never, anywhere” goes to the account list. “Wrong campaign” gets a campaign negative in the campaign that caught it. Everything else is a judgment call, and the default judgment should be to leave it alone.
- Look at the good queries too. A converting query you’re not explicitly bidding on is a keyword you should add, and the report is the cheapest keyword research you’ll ever get. Mining and blocking are the same activity looked at from two sides.
- Log what you added and why. A negative list without notes turns into a mystery within a year. When someone asks in 2027 why “template” is blocked, the answer should be findable.
When to Stop Adding
The failure mode on the other side is an account so heavily negated that it can’t find new demand. Every negative narrows the funnel a little, and enough of them starve campaigns that were meant to discover queries you hadn’t thought of. Two rules keep this in check. First, only negate on evidence: a query needs real spend and no conversions before it’s blocked, not just a feeling that it looks off. Second, audit the negative list itself twice a year. Terms that made sense for a product you no longer sell, or a market you’ve since entered, should come out.
What This Looks Like When It’s Working
Search term waste settles in the single digits as a share of spend and stays there. Campaigns stop overlapping. The account list grows slowly and the ad group lists barely grow at all. Nobody has a heroic cleanup session, because there’s never enough mess for one.
It isn’t an exciting piece of paid search work, which is exactly why it gets skipped. If your search terms report hasn’t been opened in a month, we can take a look.