Keyword tools are generous. They will suggest every phrase that shares a noun with your product, including the people who want a job using it, a free version of it, or a Saturday tutorial that ends with no cart. Those queries can have real volume. Volume is not intent.

For a store spending on Search, the filter is commercial: would a person who types this, and who lands on a product or collection page, have a reason to buy on this visit or the next one? If the honest answer is that they want instructions, employment, or a PDF, the phrase does not belong in the campaign. It can still belong on the site as an article. It should not be a keyword you pay for. That split is the same one as keeping paid clicks off the blog.

The queries that usually fail the filter

Run every survivor of the volume check through a short reject list, then through a human pass for the category:

  • Price-avoiding: free, cheap patterns, download, template, PDF, crack, serial.
  • Learning: how to, DIY, tutorial, ideas, meaning, vs as a pure definition, history of.
  • Employment: jobs, salary, career, hiring, course, certification, when you do not sell a course.
  • Wrong ownership: used, rental, repair, parts, manual, when you sell the finished product and not the service.
  • Entertainment and research: lyrics, memes, reddit, "near me" when you do not have local stock.

"Near me" is not automatically junk. A store with real local pickup can justify it. A warehouse that ships in four days cannot, unless the page says otherwise. The filter is about your fulfillment, not about a universal banned-word file.

Borderline queries need a page, not a hunch

"Best waterproof trail shoes" is commercial for a trail-shoe collection and informational if your page is a thin blog roundup of other brands. "Trail shoe size chart" might be a support query or the exact hesitation that blocks a size-8 purchase. Keep it only when the landing page answers it and offers the product in the same view. If you would be embarrassed to pay $3 for that click and land the person on a grid of shoes, drop it.

Problem queries can stay when they describe a use you sell into: "shoes for wet rock," "sheet for hot sleepers." Those are not DIY queries. They are a different bucket in the four-way campaign split, and they still need a page that sells the answer.

Put the rejects into negatives, not into a spreadsheet graveyard

A filtered keyword that remains available to Broad match, Dynamic Search Ads, or a Performance Max URL expansion will come back as a search term. When you drop "jobs" from the keyword list, add it to the negative list in the match type that matches the risk. Filtering and negatives are one decision recorded in two places. Building that list before launch is the work in the pre-launch negative keyword list.

How Kopa applies the filter

In Kopa, Smart Campaigns keeps phrases with buyer intent and drops informational research queries such as free, DIY, jobs, salaries, and how-to while the keyword list is still a draft. You can put a borderline term back if the page truly sells to it.

What good looks like

  • The keyword list has a written reason each term is commercial for this store.
  • Jobs, DIY, free, and how-to families are absent unless you sell that thing.
  • Borderline queries have a landing page that can take the click.
  • Rejected themes are added as negatives so another campaign type cannot buy them quietly.
  • Blog content can still target the informational query in organic search.