Watched status
Watched status records a user’s played classification for a media item separately from its resume position or a complete record of viewed content.
Watched status is a classification attached to a viewer and item: the system treats that item as played or unplayed. It drives choices such as unwatched filters and progress indicators.
It is separate from the current resume point and from a detailed record of which passages were actually viewed. Emby's public user-item model exposes Played alongside PlaybackPositionTicks, PlayCount and LastPlayedDate. Those fields describe different aspects of state.
| Information | Question it answers |
|---|---|
| Played flag | Is this item classified as played for this user? |
| Resume position | Where might this user continue? |
| Play count | What accumulated count does the system retain? |
| Playback event log, if present | Which events were recorded, and when? |
A catalogue item can be classified as seen because a viewer watched it elsewhere or explicitly marked it played. Emby's API includes a separate “mark played” operation; the classification need not originate from playing the local resource to its end.
Final position is not viewed coverage
Consider a fictional 60-minute programme. A viewer watches its first two minutes, seeks to minute 55, watches one minute and stops.
The final position is 56:00, or about 93.33% of the duration. The viewed material totals three minutes, only 5% of the programme. A policy based on reaching a position near the end could classify this differently from a policy based on actual coverage.
The example explains the difference; it is not Emby's completion threshold or a reproduction of its watched-state algorithm. A progress report describes a position and session state, not a proof that every intervening second was viewed.
Classification can be an intentional correction
Marking a film played can be useful when migrating an existing collection or recording that it was seen on another service. Marking it unplayed can express a new viewing intention. Neither action rewrites the movie's content.
A manual correction should therefore be understood as a change to viewer state. It does not supply a forensic account of a screening, or a reliable measure of viewing time.
A replay also introduces a new resume position without changing the historical fact that the person has seen the film before. How a product updates its played flag during replay depends on its state rules; the two concepts remain distinct.