Set the platform and let the model know the campaign type
Open the Marketplace Keyword Extractor from the homepage. The brief has slots for product name, target platform, region/language, and a one-sentence product summary. Add a sixth slot of intent: are you building a Google Ads search campaign, an Amazon PPC campaign, or a generic SEO keyword list? The output adapts to the campaign type.
A Google Ads search campaign wants exact-match and phrase-match candidates. An Amazon PPC campaign wants phrase-level keywords that match real shopper queries. A generic SEO list wants broader themes. Tell the model which one and the output is usable on the first pass.
Feeding the right brief
Write a one-sentence product summary focused on what the product is, not why someone should buy it. 'A 22 oz double-wall stainless steel water bottle with leak-proof lock cap' is a good summary. 'The best bottle for active people' is a bad summary. The first gives the model facts; the second gives the model adjectives.
Add your existing keywords and the keywords of two or three competitors. The model uses these as anchors. The output clusters will be more relevant to your actual market than a generic run.
Mapping clusters to match types
Use core terms for exact match. Use long-tail variations for phrase match. Use synonyms and intent-grouped terms for broad match. The keyword extractor's output is grouped roughly along these lines; the operator's job is to slot the right terms into the right match type.
Export the clusters into a single CSV with three columns: keyword, match type, suggested bid. Upload the CSV into Google Ads Editor or Amazon PPC bulk upload. The whole loop from brief to live campaign should take less than an hour per ad group.