How to Triage Titles You Only Sampled
A confidence-based triage for deciding whether a brief playback was a sample, a real start, or an unresolved entry.
In short: Do not classify an entry as a sample from its progress length alone. Confirm that you only intended to preview it, identify the exact title or version, verify the active profile, and check whether anyone else plans to continue. Clear one confirmed sample, then reopen Continue Watching to verify that only the intended state changed.
Brief playback creates an ambiguous signal: it may be a rejected preview, the first minutes of a real watch, or another viewer’s session. A good triage resolves that ambiguity without turning cleanup into guesswork.
Ask four questions before acting
Use this card for each candidate:
| Question | Yes | No | Unknown |
|---|---|---|---|
| Did I explicitly intend only to preview it? | |||
| Can I identify the exact title and version? | |||
| Is the active viewer or profile confirmed? | |||
| Has every relevant viewer finished deciding? |
A safe sample has four clear answers: preview intent is yes, identity and profile are confirmed, and no relevant viewer needs it retained. Any unknown moves the entry to investigate, not remove.
Progress percentage is supporting context, not proof. A very short session can still be a genuine start, while a longer session can be an abandoned sample after comparison.
Reconstruct the moment of playback
Recall what you were trying to do. Were you checking tone, picture, language availability, a particular edition, or whether the title matched its description? Did you stop because you rejected it, because the session was interrupted, or because you changed screens?
Write one neutral sentence: “Previewed to check language; no plan to resume,” or “Interrupted during opening; resume planned.” Avoid recording sensitive subject matter. This separates viewing intent from the software state.
If sampling is frequent, use a low-clutter exploration workflow so each preview ends with an explicit keep, reject, or investigate decision.
Confirm exact identity
Similar artwork can hide different episodes, cuts, languages, or source versions. Read the title, season and episode where relevant, date or edition clues, and available format or language labels. DCMI’s metadata vocabulary distinguishes identifiers, relations, formats, and languages; those fields help prevent a cleanup from targeting the wrong entity.
When two cards might represent the same work, follow the multiple-version review before changing either. Do not infer sameness from a poster.
Prove viewer scope with one item
Verify the visible profile or viewer label. If a screen is shared, ask whether anyone else started the title. Then choose a non-sensitive, clearly sampled entry and apply the available removal or dismissal action in the current interface. Reopen the row and confirm:
- the intended card changed;
- another viewer’s active entries remain intact;
- history, favorites, and progress are not assumed to share the same action;
- the state is consistent where the current product says it should be.
Norva can retain progress and preferences across supported devices under the same account. Exact profile scope and the available row controls must be verified in the current build before publication.
Choose keep, clear, or investigate
Keep when there is a real return intention, even if progress is short. Clear only when sample intent, identity, and viewer scope are confirmed. Investigate when the card could be another version, another profile’s activity, or a session whose purpose is forgotten.
For forgotten intent, use the paused-or-abandoned decision rather than inventing a reason. A review date can preserve an uncertain entry without letting it remain indefinitely.
Original evidence: the sample-or-start card
Test the four-question card on three entries: one definite sample, one genuine start, and one uncertain card. Record the classification before changing anything. After one safe action, add an outcome column showing whether the expected card alone changed.
This small exercise measures decision quality, not product performance. Its value is that another reviewer can follow the same questions and see exactly where uncertainty remains.
Common mistakes and limitations
- Using a fixed minute or percentage threshold to define every sample.
- Assuming a card belongs to the person holding the remote.
- Clearing both duplicate-looking versions at once.
- Treating an interrupted start as a rejected title.
- Keeping a private log of unnecessary viewing detail.
- Believing row cleanup changes source availability.
The method cannot recover intent that no one remembers. In that case, preserve the minimum identity details, set a review date, and make the least destructive choice.
Frequently asked questions
Can I clear every entry below a certain percentage?
No. Percentage does not reveal intent, viewer, or version. Review those entries as candidates, not automatic removals.
What if two people sampled the same title?
Confirm whether either person now wants to continue. Shared curiosity and personal intent are separate decisions.
Should I repeat the test on every device?
Only when cross-screen state is uncertain. Start with one controlled action and inspect supported screens relevant to your workflow.
Your next step
Review Norva's Product Overview