Data sources & privacy
Know which observations you supplied, how they were normalized and what this release cannot see.
Manual and imported data
Every launched analytics tool uses your supplied counts or CSV. No social profile is fetched, no live counter is monitored and no private metric is inferred. Teaching examples are labeled illustrative.
Use the canonical CSV headers in the template. Unknown headers require an explicit mapping. Choose date order for non-ISO dates and a timezone for local timestamps. Ambiguous daylight saving times require an offset.
Your private workspace
Reports, imports and experiments are owned by this browser session. Preserve the browser cookie to reopen them. Another browser cannot access them even with the URL. This is local anonymous persistence, not a synchronized account.
The original CSV is not stored. Uncommitted normalized previews expire after a day and are removed by the worker. Committed observations and saved reports persist until you remove the Docker state volume; use only data you intend to keep locally. Jobs are removed after a week.
AI and unavailable integrations
DeepSeek receives a minimized computed fact summary only when you request it, plus your optional hypothesis. It does not retrieve social accounts. Generated prose is labeled, cited to computed facts and editable in the saved report.
Optional username retrieval, OAuth and real benchmark cohorts need separately verified data rights and configuration. The shared RapidAPI test credential is absent from normal runtime; no provider lookup is available in this release.