Where the time goes
Every chart is a query. A page with fifteen charts across three data sources fires a lot of requests, and the report is only as fast as the slowest one.
First diagnostic: reduce the page to one chart and see whether it is still slow. If it is, the problem is the source or the query. If it is not, the problem is chart volume.
The levers
Fewer charts per page. Split into several pages; charts on unviewed pages are not queried.
Data freshness / caching. Longer cache windows serve repeated views instantly and reduce API quota consumption.
Extracts. Snapshot a slow source into Looker Studio storage. The biggest single win for API-backed connectors.
Reduce fields. Fewer dimensions per chart means less data returned. Tables with dozens of columns are the usual culprit.
Fix it upstream. Aggregate in BigQuery into a summary table, partition and cluster it, and connect to that. A report on a pre-aggregated view will always beat one computing over raw events.
Blends and calculated fields
Blends are computed at query time and are frequently the slowest element on a page. If a blend is unavoidable, keep it narrow and consider materialising the joined result upstream instead.
Complex calculated fields — particularly regex over large text columns — add cost per row. Move stable derivations into the source table where you can.
Common errors
Configuration incomplete means a chart is missing a required dimension or metric. You do not have access usually means data source credentials rather than report sharing. Data set configuration error often follows a schema change upstream — reconnect the source and refresh fields.
After any upstream schema change, open the data source and click Refresh fields; Looker Studio does not detect new or renamed columns automatically.