Plain-language policy
What is recorded
While reader data is enabled
The reader records episode opens, coarse scroll milestones, completion, next-episode clicks, returns, follows, bookmarks, comments, and reader-pulse responses.
How an episode is read
About every five seconds while the page is visible, the reader can record a 5% scroll-depth bucket, scroll direction and speed bucket, visible, focused, active, and idle time, and the stable paragraph or illustrated-panel ID in view. Collection pauses in hidden tabs and after five inactive minutes.
Where a reading session began
A fixed category such as search, social, internal, direct, or external is derived in the browser, together with a broad landing surface such as story or reader. The referring URL, hostname, search terms, and campaign text are discarded rather than sent or stored.
What is never collected
Detailed reading records contain no raw story text, exact scroll pixels, pointer coordinates, IP address, user agent, device fingerprint, or raw referrer. Hosting infrastructure may keep separate short-lived security logs.
Retention and analysis
Raw five-second samples are deleted after 30 days. Pseudonymous session totals are deleted 13 months after their last update, giving editors a bounded window for comparing reading depth, dwell time, and continuation without exposing reader IDs in reports.
Automatic privacy signals
Global Privacy Control, Do Not Track, the Fictures opt-out, and automated QA mode suppress passive collection. Deleting or disabling reader data clears the local IDs and entry category and cascades deletion through stored reader data.
Comments and pulse notes
Free-text comments become public only after review. Optional pulse notes stay in the editorial database and are not returned by aggregate analytics.
What not to share
Do not include names, contact details, or other personal information in comments or notes.