Connector types
Google connectors are free and first-party: Analytics, Ads, Search Console, Sheets, BigQuery, Cloud SQL, YouTube Analytics.
Partner connectors are third-party, mostly paid, covering platforms like Facebook Ads, Shopify and Salesforce.
Community connectors are built with Apps Script, including ones you write yourself against any API.
For anything not covered, the pragmatic route is landing data in BigQuery or Sheets first and connecting to that.
Reusable vs. embedded
A data source can be reusable — a standalone object you connect to from many reports — or embedded inside one report.
Reusable is almost always right. Field renames, calculated fields and formatting live in one place, so a change propagates everywhere. Embedded sources are convenient for a throwaway report and a maintenance problem for anything else.
This mirrors the published-vs-embedded dataset decision in Power BI, and the reasoning is identical.
Extract data
The Extract Data connector snapshots up to a configured row limit from another source into Looker Studio's own storage, refreshed on a schedule.
It is the main performance lever available. A report on a slow API connector becomes fast when it queries an extract instead. The trade-offs are freshness and the row cap.
Alongside that, data freshness settings control cache duration per source — from 1 minute to 12 hours depending on connector. Longer caching means faster reports and fewer API quota hits.
BigQuery specifics
BigQuery is the connector to know. Cost is per byte scanned, so a badly built report can be expensive.
Mitigate it: query a partitioned and clustered table, connect to a view or a summary table rather than raw events, avoid SELECT * in custom queries, and enable BI Engine where budget allows for sub-second responses.
A custom SQL query as the source gives you control, but it defeats some of Looker Studio's own optimisation — prefer a well-designed view in BigQuery.