The Negative Keyword List Is a Living Document: How to Maintain It Weekly

Your negative keyword list is probably three years old

Most Google Ads accounts we audit have a negative keyword list that was built once, during setup, and then forgotten. Someone pasted in a generic list of junk terms — "free," "jobs," "cheap," "download," maybe "reddit" — and moved on.

That list is a snapshot. Search behavior isn't. New queries hit your account every week that no generic list could have predicted, and every one you don't block is a bid you're placing on a search you don't want.

This post is the weekly maintenance routine we use on our own account at Plyto and on customer accounts. It takes about 20 minutes if you do it properly, and it compounds.

Why static lists fail

A few reasons the paste-once approach breaks down:

• Your own ad copy teaches Google new matches. Change a headline, and the broad-match algorithm starts finding new neighborhoods of queries. Some of them are garbage.
• Seasonality shifts intent. "Tax CRM" means something different in March than in September.
• Competitors launch products with names that sound like categories. Suddenly you're bidding on a competitor's SKU because it contains a common word.
• Performance Max and broad match keep widening the net. Google's own recommendations push you toward looser matching every quarter.

A static negative list cannot keep up with a dynamic matching engine. The list has to breathe.

The weekly 20-minute routine

Block out the same time every week. Fridays work well because you're reviewing the week that just happened while it's fresh.

Step 1: Pull the search terms report for the last 7 days. Filter to terms with at least one click. You don't care about impressions-only terms yet — those are a separate exercise.

Step 2: Sort by cost, descending. The expensive mistakes are the ones that matter. A $0.42 wasted click is annoying. A $14 wasted click eats a day of budget.

Step 3: Go line by line through the top 50. For each term, ask one question: *would I have written this ad for this search?* If the answer is no, it's a negative.

Step 4: Decide the match type for each negative.

• Exact negative when the term is specific and might have close variants you *do* want. Example: block `[crm free]` but keep `crm pricing` live.
• Phrase negative when a word combination is reliably bad. Example: `"crm jobs"` kills every job-seeker query containing those two words in that order.
• Broad negative when a single word is poison regardless of context. Example: `tutorial`, `course`, `salary`.

Broad negatives are the most powerful and the most dangerous. One wrong broad negative can shut off a whole segment. Add them deliberately.

Step 5: Push the list to a shared negative keyword list, not the individual campaign. Shared lists apply across campaigns and are the only sane way to manage this at scale. If you're still adding negatives one campaign at a time, stop.

Real examples from our own account

Some negatives we've added over recent weeks, with the reasoning:

• `salesforce` (phrase). We don't want to show against a competitor's brand name. Different problem than a generic searcher.
• `free crm` (phrase). Anyone leading with "free" isn't going to pay us. We let "free trial" through separately as a positive.
• `crm meaning` (exact). Someone looking up a definition isn't buying software today. Educational intent, not commercial.
• `hubspot alternative` (phrase). Counterintuitive — we are an alternative. But in practice these searchers want a feature-by-feature HubSpot clone and we're a different product shape. The clicks converted at roughly a third of our account average.
• `course` (broad). People wanting to learn marketing, not buy marketing software.
• `for nonprofits` (phrase). We don't have nonprofit pricing. Blocking this is honest.

The `hubspot alternative` one is the interesting case. Nothing on a generic negative list would have caught it. You only find it by looking at your own data.

The quarterly deep clean

Every 90 days, do a bigger pass.

• Pull search terms for the full quarter, not just the week.
• Look at terms with high impressions but low CTR. Those are telling you the matching is wrong even if you're not getting billed for clicks — they drag your Quality Score down.
• Review your *existing* negatives for anything you've outgrown. Launched a feature that changes who you want to reach? Some old negatives may be blocking valid traffic now.
• Check for overlap between your shared negative lists. Duplicates aren't harmful but they make audits slower.

What AI changes about this

The weekly routine above is doable by hand. It's also the kind of work that gets skipped the week you're busy, which is most weeks.

An agent inside your ad account can do it continuously: read the search terms report every day, score each term against your actual conversion data and your positioning, propose negatives with a reason attached, and apply them to a shared list once you approve the batch. That's the loop we run inside Plyto, and it's why our wasted-spend line shrinks instead of drifting up over time.

The principle holds either way. The negative list is a living document. Treat it like one and your cost per conversion drops without touching bids, budgets, or creative.