Privacy
Privacy Policy
Decode My Feed (“we”, “us”) turns your YouTube watch history into something you can explore, and into a prompt you can take to your own AI assistant. It is free and there is no account. The whole product is built around a single fact: your watch history never reaches us. It is collected on your device, analysed in your browser, and stored in that browser so you can come back to it. There is no upload step and no server here that could receive it, so there is nothing for us to retain, sell, lose in a breach, or hand to anyone who asks. The one moment anything leaves your device is when you choose to paste our prompt into an AI assistant — that is a message from you to them (section 4). The only things we hold at all are fixed-shape error reports and anonymous usage counts (sections 7 and 8).
1. The browser extension
Decode My Feed offers an optional Chrome extension that collects the watch history used by the website. The extension is the only part of the product that reads your YouTube watch history page, and it does so entirely on your own device.
When a scan runs on your YouTube history page, the extension:
- scrolls the page to load your last 90 days of history (up to your 1,000 most recent videos);
- reads, for each video, the title, duration, how much you watched, the completion percentage, and the date watched; and
- hands that information to decodemyfeed.com in a new tab, so you do not have to export and re-upload a file. If that handoff cannot happen, it saves the information as a JSON file to your computer instead, and you can always ask for the file yourself.
That is all it does. The extension has no servers of its own and makes no network requests: it does not send your watch history to us, to anyone else, or anywhere at all — the handoff above passes it between two tabs inside your own browser, and it happens only for a scan you just asked for. It contains no analytics, advertising, or tracking code, and it reads only youtube.com pages (the permission it requests).
The extension also has an optional set of feed controls — marking bait-styled titles, hiding Shorts, the home feed, the sidebar, comments or end screens, and draining thumbnail colour. These change how YouTube looks in your own browser and nothing else. The title scoring runs entirely on your device and is the same code the report uses; no titles, counts, or anything about what you were shown is sent anywhere. Your on/off choices are the only thing kept — in the browser’s synced storage, so they follow your Chrome profile to your other machines. They are preferences, not data about you, and we never see them.
Two of those controls keep a little more, on your own computer, and neither can be switched on until the extension has shown you what it keeps and you have agreed. Track my feed over time stores one pair of numbers per day — how many titles you were shown and how many were bait-styled — with no titles, channels or links, capped at 35 days. Retrain mode stores the action list from your own report, which you send to the extension yourself from the report page, plus which items you have ticked off — and, sent in the same step, the channel names and completion figures from that report, so the blocker can suggest the channels you open and leave. Tell me when the feed has me and the daily time alert keep two durations per day: how long you browsed and how long you watched. Not what you watched, not which pages, not which videos — two numbers, capped at 35 days. Channels you block yourself are kept as a list of names.
All of it lives in your browser’s local storage on that machine, all of it is erasable with one button in the extension’s popup, and none of it is ever transmitted — not to us, not to anyone. We have no way to read any of it. The alerts show you a line and nothing more: the extension will never close a tab, block a video, or stop you watching something. The cooldown and one video at a time prompts keep nothing at all: they count the current sitting in the tab and forget it when you close it, and the answer to “why are you continuing?” is never recorded, never counted, and never sent — the question exists for you, not for us.
Between finishing a scan and handing it over, the extension holds your history in the browser’s session storage: kept in memory, never written to disk, deleted the moment the website collects it, discarded after 30 minutes if it is never collected, and gone entirely when you close your browser. Nothing else is stored. Your history stays on your device throughout: the handoff passes it to a page running in the same browser, which analyses it there. The rest of this policy describes what that page does with it.
For completeness, our upload page can also talk to the extension in two narrow ways: it can ask whether the extension is installed (so it can skip the install instructions if you already have it), and it can ask it to start a scan, which opens your YouTube history in a new tab. To the first question the extension answers with nothing but yes and its version number. Only decodemyfeed.com can ask, no other website can, and none of it leaves your browser.
Installing and using the extension is free, as is the analysis on the website.
2. What we process
For the analysis itself: nothing. Your watch history — handed over by the extension above, or opened from a file you choose (the extension’s youtube-watch-history.json, or a Google Takeout export) — is read, parsed and analysed entirely in your browser. It is not uploaded. There is no request that carries it, and no server here that could receive it.
This is a change from how the site used to work. Earlier versions sent a set of fields to an analysis service to generate a written report; that no longer happens, and the paid report that required it is not currently offered. If it ever returns, this policy will be updated before it does.
Before the analysis we also ask you two optional questions — what you feel you use YouTube for, and what you’re working toward — plus an optional free-text note. Answering is entirely voluntary and you can skip the whole step. Your answers stay in your browser alongside your history (section 9); they are never sent to us, never written to a database, and never recorded in our logs or usage analytics. We record only whether the questions were answered, never what was picked.
We analyse at most your 1,000 most recent videos from the last 90 days. We never ask for, receive or store your email address, and there is no account.
3. How the analysis works
On your device, in JavaScript, with no network calls. Topics are worked out by matching keywords against video titles and channel names; everything else — your totals, channels, sittings, the runs where you stayed on one subject, the hours you watch — is arithmetic over the same data. No artificial intelligence is involved at any point, and nothing about your history is sent anywhere to produce it.
4. The prompt you choose to send
The last step assembles a block of text — your figures, your stated goals if you gave them, and a description of what your export cannot show — designed to be pasted into an AI assistant such as ChatGPT or Claude. Assembling it happens in your browser like everything else.
Sending it is your decision, and it is the one moment data leaves your device. If you copy that prompt and paste it into an assistant, you are sending a message to that company under their privacy policy. We are not a party to it, we do not see it, and we have no relationship with them on your behalf. Two things are there to keep that decision informed: the full text is shown to you before you copy it, and a switch removes real video titles so you can send only the numbers.
5. Payment
None currently. The service is free to use today, we do not ask for card details, and we hold no payment records. If we introduce a paid feature later, this policy will be updated before it is offered.
6. Service providers
We rely on a small number of service providers to operate the website. They process data only to perform the function described, on our behalf. None of them receives your watch history, because it never leaves your browser:
- A website hosting provider serves these pages and collects the anonymous usage counts described in section 8.
- A cloud infrastructure provider receives the fixed-shape error reports described in section 7 and stores anonymous daily usage totals.
- The Chrome Web Store distributes the extension, under Google’s own terms.
The browser extension does not use any of these providers — it runs only on your device. These website providers process data in the United States and other countries. We do not sell your data or share it with anyone for advertising.
7. Logs
Our serverless functions may record basic operational and error information (for example, that a request failed) to keep the service running. These logs do not include your watch history content or the parsed video data.
The website also reports its own errors to us — if something breaks in your browser (an upload that won’t parse, a report that fails to generate), we record which step failed and why: a fixed error name, the step of the flow, an HTTP status code, and the page path without its query string. We deliberately do not send the browser’s error message text, because a failure while reading your file can quote part of that file. These reports contain no watch history, no email address, no report content, and nothing that identifies you or your device. They are written to our function logs and are not stored in any database.
8. Cookies and analytics
We do not set advertising or cross-site tracking cookies, we do not require an account or login, and we never sell or share data with advertisers.
We use privacy-friendly web analytics to count visits to public pages. It is cookieless and does not build a profile of you, follow you across other websites, or store a persistent identifier for your browser.
To see where a campaign gets stuck, the website sends our aggregate counter only two fixed labels: the campaign source and the funnel step reached. It stores no individual event. Instead, the database increments one aggregate count for each UTC day, source and step. The public website can increment approved combinations but cannot read or edit the table. No video titles, channel names, timestamps, email addresses, or report content are sent, and there is no field in which they could be stored.
When you arrive through one of our own campaign links, your browser keeps a fixed source label (such as “youtube” or “dataisbeautiful”) and the date it was captured for up to 14 days. It also remembers which fixed steps it has already counted during that window, so a reload does not ordinarily increment the same step again. We use those steps—for example, checker used, upload opened, and upload parsed—to see whether a campaign led to meaningful use. We never store or send the raw query string or referrer, and the label is not a visitor identifier.
On our public pages only — the home page, the live example, the guides, and these legal pages — we also record which buttons were pressed, roughly how far down the page you scrolled (in quarters), and roughly how long you stayed (in bands such as “30–120 seconds”). This tells us which parts of our own copy are working. Buttons are recorded as fixed labels we wrote ourselves, never as the text on screen.
We deliberately do not record clicks, scrolling, time-on-page, or screen content on pages that can show your own data—the upload page, the questions, your dashboard, and shared links. Those pages may emit only the fixed funnel step and source label described above, plus the fixed operational counts listed earlier in this section. Behavioural observation is excluded by an allowlist, so a new page is silent by default until it has been explicitly judged safe. We do not use session recording, screen replay, or heatmap tools anywhere on the site.
The extension uses no cookies, analytics, or error reporting at all — see section 1.
9. Data retention
We retain nothing about you. Your watch history and your answers to the optional questions never reach us in the first place (sections 2 and 3), so there is nothing on our side to keep, delete or hand over. The only things we hold at all are the operational error reports and anonymous usage counts described in sections 7 and 8, neither of which contains personal information.
If you used Decode My Feed when it still produced written reports, your browser may also hold a short summary of each report (persona name, three headline numbers, top topics and channels) and, for Google Takeout reports, a voucher for a free second scan. Both live on your device only, were never sent to us, and expire on their own — the summaries after about a year, the voucher after 30 days. Clearing this site’s data in your browser removes them immediately.
Your browser also keeps the watch history you analysed, together with your answers to the optional questions from section 2, so that your dashboard is still there when you come back to it and you don’t have to upload the same export twice. This is the analysis itself, not a summary — the titles, channels and dates from your file. It is stored on your device only. It is never sent to us, never logged, and never counted; we have no server that could read it, and the analysis that produces your dashboard runs entirely in your browser. It is deleted automatically after 90 days, which is the same window the analysis covers, and the “Forget my history on this device” link on your dashboard deletes it immediately. If you copy a prompt from that page and paste it into ChatGPT or Claude, that is a message you send to them under their privacy policy, not ours — we are not involved in it and never see it.
Finally, if you start a Google Takeout export from our upload page, your browser records a single timestamp noting that you did — nothing else. Its only purpose is that when you come back with the file, possibly days later, the page can open on the right step instead of starting you over. It contains no account name, no export, and no watch data; it is stored on your device only, is never sent to us, logged, or counted, and it is deleted the moment you upload a history or after two weeks, whichever comes first. The optional calendar reminder we offer at that step is likewise built entirely in your browser and contains only a date and a link to our upload page — we never see it, and it is not an email reminder because we do not hold your address.
10. Your rights
Depending on where you live (for example, under the GDPR or CCPA), you may have rights to access, correct, or delete personal data a service holds about you. Because your watch history, your answers and your prompts never reach us at all, and because there is no account, in practice there is nothing here for us to retrieve or erase — the copies live in your browser, where the “Forget my history on this device” button deletes them immediately and clearing your browser data removes them too. If you have pasted a prompt into an AI assistant, any rights over that conversation are exercised with that provider. For anything else, contact us at the address below.
11. Children
Decode My Feed is not directed to children and is not intended for anyone under 16. Please do not use the service if you are under the minimum age in your country.
12. Changes to this policy
We may update this policy as the service evolves. When we do, we will revise the “Last updated” date above. Material changes will be reflected here before they take effect.
13. Contact
Questions about this policy or your privacy? Email us at info@decodemyfeed.com.