People hand their YouTube watch history to random web tools far more casually than they would hand over their browser history, even though the two are closely comparable. A timestamped log of everything you chose to watch is not a list of videos; it is a behavioural record. This guide sets out what can actually be inferred from the export — not to alarm, but because the answer is the reason our analyzer runs entirely in your browser.
Your schedule, in detail
The timestamps alone reconstruct a routine. Cluster views by hour of day and day of week and a shape appears almost immediately: when you wake, when you commute, when you stop working, when you go to bed, and which days break the pattern.
Extended gaps are as informative as the clusters. A week with no activity reads as travel, illness, or a holiday. A sudden shift in the evening peak reads as a new job or a new schedule. A run of 3am views reads as insomnia or shift work.
None of this is hidden or clever. It falls out of a histogram of a single field.
What the topics imply
Channel names and video titles carry more than interests. Watch histories routinely contain searches for symptoms before a diagnosis, tutorials for a skill someone is quietly learning for a career change, content in a language someone is studying, material tied to a religious practice, political commentary, and content associated with a specific life event — pregnancy, bereavement, addiction recovery, a breakup.
Several of those categories are treated as sensitive under privacy law precisely because inference from them is consequential. And inference here does not require sophistication: a run of twenty videos from one channel over two days is legible to anyone reading the list.
It is worth noting what the export does not settle. It records that a video was opened, not that it was watched, agreed with, or enjoyed. A history is evidence of attention, not endorsement — a distinction that matters if this data is ever read by someone inclined to judge it.
Why an upload is different from a download
A file on your disk has one copy under your control. The moment it is uploaded, that stops being true in ways that are hard to reverse.
The file may be retained after the page closes. It may sit in server logs or backups outside whatever the interface deletes. It may be readable by staff, by a future acquirer of the service, or by anyone who breaches it. And a privacy policy is a promise about intent, not a technical constraint — it can be amended, and it does not survive a bankruptcy sale or a subpoena.
Free tools have a particular version of this problem: the processing costs money and the tool charges nothing, so it is fair to ask what the business model is. Sometimes the answer is benign. Sometimes the data is the answer.
What local processing changes
Our analyzer parses the file in your browser tab using ordinary JavaScript. The file is read from your disk into memory, analysed, and discarded when you close or refresh the page. It is never transmitted, because there is no server behind this site to transmit it to — the whole site is static files.
You do not have to take that on faith, which is the point. Open your browser's developer tools, switch to the Network tab, and load your export. If a tool is uploading a file, a request carrying it appears there. This check works on any tool, including ours, and it is the only privacy claim that can be verified from outside.
A stronger version of the same test: disconnect from the network after the page loads, then run the analysis. Local processing keeps working.
Sensible handling either way
Treat the export like a password manager backup rather than a photo. Delete it from your downloads folder when you are finished, since the file will happily sit there for years otherwise. Be careful about screenshots of results, which often contain video titles you did not intend to share. And if you want to share findings, share the aggregate numbers, not the underlying file.
Who realistically ends up reading this file
The dramatic threat models are not usually the relevant ones. The export ends up exposed through ordinary carelessness far more often than through attack.
It sits in your Downloads folder indefinitely, which is a problem on any shared or work machine and on any laptop that gets sold or repaired. It gets swept into cloud sync if Downloads is synced, which quietly copies it somewhere with a different access model. It lands in automated backups that outlive the original. And on a work device it may sit within reach of endpoint management software and IT staff who have no interest in your viewing habits but do have access.
None of that requires anyone to be malicious. It requires only that the file still exists somewhere you stopped thinking about.
The export identifies you even with your name stripped
A common instinct is that the file is anonymous because it contains no name, email, or account ID. That intuition is wrong, and it is worth understanding why.
A viewing history is close to unique. The specific combination of channels someone follows, at the specific times they watch, is rarely shared by anyone else — and much of it maps to public information. Local news channels indicate a region. School or university content indicates an institution. A niche hobby channel with a small audience narrows the population sharply. Language and posting-time patterns narrow it further.
Research on de-anonymisation has repeatedly shown that a handful of behavioural data points is often enough to single out an individual within a large dataset. Treat the export as identifying, because in practice it is.
Sharing results without sharing yourself
Watch-history stats are fun to post, and the screenshots people share usually give away more than intended.
Aggregate numbers are safe: total views, views per year, the shape of your hour-of-day curve. Top-channel lists are much less so, because a small set of channels is close to a fingerprint and often reveals politics, health, religion, or sexuality by implication. Individual video titles are the most revealing of all, and they appear in screenshots more often than anywhere else.
Before posting a screenshot, read every visible title as though a future employer, a relative, or a stranger with a grudge were reading it, because in principle all three can. If that feels excessive, crop to the charts and leave the titles out.