A movie room should not become a profile of its participants.

What a room stores

A room stores its short code, the nickname each participant supplies, the host's constraints and consensus mode, participant presence, private votes, candidate titles, matches, shortlist choices, the final selected title, and decision events needed to keep everyone's state consistent.

Individual votes are not returned to other participants. Room members can see nicknames, participation state, aggregate consensus outcomes, shortlisted titles, and the final selection. Shared fit reasons describe group-level constraint satisfaction; they do not identify who disliked a title.

Browser storage and room access

The solo discovery experience keeps a random anonymous device identifier in local storage. Joining a room sets a Secure, HttpOnly, SameSite=Lax capability cookie for that room. JavaScript cannot read that cookie, and the server stores only its SHA-256 hash. Clearing site data removes the browser copy.

Solo discovery also stores progress separately for each feed: up to the 500 most recently seen title IDs, the current pagination cursor, and a saved timestamp. This keeps already-viewed titles out of the next local session. Creating a room sets a separate Secure, HttpOnly, SameSite=Lax returning-host cookie for 90 days so a later room creation can be counted as a repeat; the cookie is a random value and is not an account or advertising identifier.

An invite contains a room code, not a participant identifier or vote history. Anyone who receives a valid code can ask to join, so treat an invite like a private link and share it only with the intended group.

Measurement and abuse prevention

TikFlick records a limited decision funnel: create, invite, join, first swipe, match, shortlist, selected title, abandon, and repeat. Separate engagement events can record a random device identifier, title identifier, vote or player action, and dwell time. These records are used to operate and evaluate the decision flow, not to build advertising profiles.

Vercel Web Analytics supplies aggregate page and device statistics without third-party analytics cookies. Vercel documents that this data is anonymized and that its visitor hash resets daily. TikFlick also derives short-lived, one-way rate-limit keys from request network information to slow abusive room operations; it does not place the raw address in those rate-limit records.

Retention and deletion

• A room accepts activity for up to 24 hours after creation.
• The host can delete the active room immediately. This removes the room, participant, vote, candidate, and decision-event records; a minimal funnel record containing the room code, event category, and optional title ID can remain within the 90-day ceiling, but it contains no nickname, capability, or individual vote.
• Room, participant, vote, candidate, and decision-event records are scheduled for automatic database deletion no later than seven days after creation. Database TTL cleanup can occur shortly after the deadline rather than at the exact second.
• New engagement and aggregate funnel events have a retention ceiling of 90 days.
• Rate-limit records expire with their short abuse-prevention window. Cached movie metadata is not participant data.

Services involved

Vercel hosts the application and provides Web Analytics; MongoDB stores room and event records; TMDB supplies movie and series metadata; JustWatch supplies regional availability data through TMDB; and YouTube serves a trailer only when its embedded player is opened. Following an external link or loading an embed is also subject to that service's own privacy terms.