Index of every page on this site: llms.txt
Skip to content

Query and export search data

A focused query lets you check a Report Section against the underlying Search Analytics rows. Keep the Site, search type, and date window fixed while comparing results.

Choose live or stored rows

Live rows

gscdump query --live --site example.com --type web \
  --dimensions page,query --limit 100 --format json

Live reads end on the newest final Google date by default, three days ago in its Pacific reporting calendar. Check meta.source in JSON output. Google may omit low-volume query detail and caps its API rows.

Store rows

gscdump query --site example.com --type web \
  --dimensions page,query --limit 100 --format json

The CLI chooses the Store when its required dates are complete. If coverage is partial, it stops and prints a sync command. The query reference explains routing.

Filter a result

gscdump query --site example.com --type web \
  --start 2026-08-01 --end 2026-08-28 \
  --page '~/blog/' --dimensions page,query --format table

A bare filter matches exactly. ~text contains text; !~text excludes it. A page path can also be a full URL. Use --query, --device, and --country for narrower reads. On the Store, a dimension and filter combination needs a matching stored table; the CLI names --live if none exists.

Export the query

gscdump query --site example.com --type web \
  --dimensions page,query --format csv --output rows.csv

query supports table, JSON, and CSV. Use --output - to write to stdout in a script. The query reference describes the JSON metadata and date semantics.

Export the Store

gscdump dump --site example.com --format parquet --out ./search-export

dump exports stored datasets. It supports Parquet, CSV, JSON, NDJSON, SQLite, and DuckDB. Analytics rows include site and search_type. See the Export reference before joining tables or sharing files.

Read a terminal chart

gscdump query --site example.com --dimensions date --format table
gscdump analyze trends --site example.com

Table output places exact numbers beside bars or sparklines. A missing-series dot means missing data, not zero. Each sparkline scales independently, so compare its shape and read the totals for magnitude. Lower average position means improvement. Use JSON or CSV when another tool needs exact values. TERM=dumb gives ASCII charts.

If data is partial

Check meta.source, selected dates, and Store coverage. A live row budget warning means the displayed rows are incomplete. Raise --fetch-budget for supported Analyzer and Report runs, or narrow the question. Next, review weekly traffic. More data tasks: Use data.