How we count

The film page says ParentFilter never tracks what you watch. Here is the whole of what this site counts about a visit — every item, what each one carries, and how long it lives. Nothing in the first list is tied to a person; the second list says exactly where an account or an address does appear, and why.

What is counted

• Page views on ten pages — the homepage, /try, /film, /watch, /signin, /subscribe, /press, /methodology, /corrections and /analysis — through Cloudflare’s Web Analytics beacon: which page, the site that sent you, and the country and browser type as Cloudflare reports them. Cloudflare describes the beacon as cookieless and fingerprint-free; that description is theirs. The beacon is not on /help, /privacy, /terms, any film page, the page that opens when the extension is removed, or this page.
• Our own short links (/go/…) — when one is followed we count which link it was, which section of the site or the store it led to (the first part of the address, never a film), and the country Cloudflare reports.
• The demo on /try — we add one to the day’s count the first time in a visit that the sample clip starts playing, and one more the first time it plays to the end. A count of visits in which it played, not of people; it carries nothing else about you or your visit.
• Fresh installs — the browser extension opens our connect page once when it is first installed; we count those opens per day. Not who, and not whether it connected.
• Why an extension was removed — when the browser extension is removed, Chrome opens a page here that asks one question. If you tap an answer, we add one to the day’s count for that answer. Only the answer is sent — the page’s address carries nothing, it sets no cookie, and nothing else about you or your visit rides on it.
• Active extensions — each time a connected extension on the Full plan refreshes its guide list, about once a day, we add one to that day’s count. A count of refreshes, not of people.
• Extension check-ins — each time a connected extension on any plan, free included, checks its account — at most once an hour per connected browser (each sign-in holds its own credential) — and the moment it connects with one tap, whatever the hour — we add one to that day’s count under the plan (free, Lite or Full) and whether it was the connect, the extension’s later check, or the site asking. A count of check-ins, not of people; nothing about the account rides on it.
• Sign-ups by channel — when an account is first created we count one under a short channel name for the page or link the sign-in started from, if it carried one (otherwise “direct”), and whether the account came through the emailed link or through Google. The channel name rides on the one-time sign-in link, which expires in fifteen minutes, or — for Google — inside the ten-minute sign-in cookie that carries that one sign-in across the trip to Google and is cleared when the sign-in finishes: on the way back, or, for a new account or one that hasn’t yet agreed to the current terms, once the screen after Google is answered or cancelled, and at most ten minutes later. Until then that cookie also holds the address Google confirmed and Google’s identifier for the account, signed so it can’t be altered, and sent only to this site. It is never written to the account.
• Analysis cost per scan — when a picture analysis on the analyzer page ends (finished, stopped, or its tab closed), the page tells us what that scan cost us: how many requests it made, how many minutes of film it looked at, what those requests cost, and how much text the analysis read and wrote. We keep it under the day, which analysis model answered, whether it was a first scan or a retry of the parts that failed, and whether it finished — so we know what analysis really costs. It is stored with the time it reached us, which is when the scan ended. No account, film, title or file rides on it, and it shows you nothing new.
• Remote taps, per guide — when a phone is paired as a remote, we keep a count per guide of how often each listed moment (or, for a command outside any listed moment, which 30-second stretch of the film) was muted, blurred or skipped, with no account, session or time attached.

Where an account or an address does appear

• Abuse counters. To keep one source from flooding the site, we count attempts — sign-ins, ratings, flags, reports, uploads, guide requests, demo plays, answers to the removal question, and the like — per connection, per network operator, per hashed email address, per account, and in two cases site-wide; for flags there is also one marker per account per film that says the day’s first save was already counted. Each counter is a single number under such a key, lives for an hour, a day or a week depending on the action, and is then gone. They are the only counts on this page keyed to an account, and they hold nothing but the number.
• A guide request. If you ask to be told when a film is covered — from the “Is your film covered?” search on the phone guide, or from the demo on /try — we keep the email address you typed, the title you typed, which of those two pages you typed it on, and when. That is a request, not a record of anything you watched, and it is not linked to an account. It sends one email when that guide is published; the row is deleted thirty days after that email, or six months after the request if no guide came — or as soon as you use the link in the email, which deletes every request made with that address.

What is not counted

• No viewing history. Nothing here records what you watch. A moment you choose to flag is stored under your account, and the privacy page says exactly how.
• The covered-film search is answered and not kept. What you type into “Is your film covered?” is looked up and forgotten; nothing is stored unless you ask to be told.
• Cookies count nothing. The cookies this site sets sign you in and bring you back to the page you came from; none is read to make a count, with the one exception above — the ten-minute Google sign-in cookie, which carries the channel name for that single sign-in.
• No ad networks, no cross-site trackers. The human check on the sign-in page and on the guide-request form is a script from Cloudflare that decides whether a request came from a person; it counts nothing for us.
• No per-person analytics. Every count in the first list is one tally under a label — a page, a link, a day, a channel, a country as Cloudflare reports it — never under a person; a scan's cost carries its figures under the day and the model, with the time it reached us, and nothing else. Our code never asks the mail provider to track opens or clicks.

This page describes the code as deployed; when the code changes, this page changes with it. Patent pending.