A negative keyword added after the search term report is a receipt for money already spent. Catalog stores already know whole families of queries that will never become an order: free, DIY, jobs, repair manuals, used, wholesale logins, and the brand name sitting inside a non-brand ad group. Those terms should be blocked before the campaign is allowed to spend.

Public accounts of Google Ads automation keep returning to the same failure. A recommendation or a broad setting turns on, negatives come off or never get loaded, and spend moves to queries the store did not choose. The fix is not a longer weekly review. The fix is a list that exists on day zero.

What belongs on the list before launch

Start from the store, not from a generic "negative keyword" download. A 4,000 SKU home catalog and a 4,000 SKU parts catalog do not share the same junk.

Keep four groups separate so you can see what each one is protecting:

  • Research noise: free, DIY, how to, jobs, salary, course, PDF, manual, used, second hand, rental.
  • Wrong commercial intent: wholesale login, distributor application, replacement parts you do not sell, or a size and voltage you cannot ship.
  • Brand leakage: your own brand terms, misspellings, and brand-plus-product queries, applied only to non-brand and test ad groups.
  • Support intent: shipping, returns, tracking, coupon codes for a past order, and careers. Pair this with URL exclusions so the click never lands on a policy page either. The URL side of that problem is covered in how to stop paid clicks from landing on blog and policy pages.

A term that is junk in one country can be a product name in another. Do not paste an English list onto a German or French campaign and call it localized.

Match type is part of the negative

"Free" as a broad negative and "free" as an exact negative do different jobs. Exact blocks that query. Phrase and broad negatives block queries that contain the idea, and they can also block a product you still want, such as a "free-standing" cooker or a "second-skin" base layer.

Write the match type next to the term before anyone uploads the list. Shared lists are useful when the same junk applies to every non-brand campaign. They are dangerous when a clearance campaign and an acquisition campaign need different exceptions. Name the list after the job it does, not after the date someone exported it.

Do not wait for the first invoice to discover the gaps

After launch, the search term report is how you extend the list. It is not how you create it. If the first week of a new campaign is the moment you learn that "jobs" and "manual" are spending, the pre-launch list was incomplete.

Review converting queries before you add a negative. A query with orders and a query that merely looks ugly are not the same decision. Block the one that cannot become a profitable order. Leave the one that already has, and decide whether it deserves its own ad group instead.

How the draft should carry the list

In Kopa, Smart Campaigns attaches negative keyword lists in the chosen match types as part of the launch draft, including brand terms kept out of non-brand ad groups. The list stays in draft with the keywords and ads until your team approves publishing.

What good looks like

  • Every non-brand campaign has a named negative list attached before the first spend.
  • Brand terms are negatives on non-brand and test ad groups, and they are keywords only in the brand campaign.
  • Match type is written on every negative, and product phrases that contain a junk word are checked before a broad negative goes live.
  • Support and informational queries are blocked in keywords and in URL exclusions together.
  • The search term report extends the list. It is not the first time the list exists.