keyword-researchlisted
Install: claude install-skill visibian/seo-geo-skills
# Keyword research from Search Console
Use what Google already reports about the site: the queries it shows for, how often, at what position and with what click-through. It is the user's own data, free, and closer to their buyers than any keyword list.
## Get the data
Ask the user to export it (they need access to the property in Search Console):
1. Search Console, Performance, Search results. Set the date range (the last 3 months is a good start), and for a comparison, Compare to the previous period.
2. Export, Download CSV. The download holds several tables; use `Queries.csv`. For a comparison, export each period on its own.
3. For pages that compete for the same query, the query and page together are needed: as a CSV with `query` and `page` columns from Looker Studio or the BigQuery bulk export.
With a Search Console API token set in their environment, the `search-console` skill pulls the same data live, query and page together, without an export: `python3 gsc_api.py performance sc-domain:example.com --days 90` writes a CSV this skill reads as it is.
Never ask for the user's Google password, a token or access to their account; work from files they give you or from the token they set themselves.
## Steps
1. Run, from this skill's folder:
`python3 scripts/gsc_queries.py Queries.csv --brand "brand name,brand.com"`
Add `--compare Queries-before.csv` for declines. It lists quick wins, low click-through against the site's own median at each position, questions, top