How Online Video Downloaders Actually Work

Why the same link works on one site and fails on another.

You paste a link, you press Download, and either a file arrives or an error does. The difference is almost never the button. It is the method behind it. This guide walks through what happens between the paste and the file, the four methods downloaders use, where each one breaks, and how FileShot combines them so that one broken method does not end the download.

What happens after you paste a link

1. Identify the page. The downloader works out which site the link belongs to and which video on that page you mean. Short links, share links and tracking parameters all have to resolve to one video.
2. Find the media. The page you see is not the video. Somewhere behind it is a media address: a single file, a playlist of segments, or separate audio and video streams.
3. List the formats. Most platforms store the same video at several resolutions. The downloader reads that list so you can choose a quality.
4. Fetch the bytes. The downloader requests the media with whatever the platform expects: the right headers, a fresh signed address, sometimes the same cookies the player used.
5. Assemble one file. Segments are joined in order, and separate audio is merged with the video so the file has sound.
6. Deliver it. You get a normal MP4, WEBM or audio file that plays anywhere.

A downloader can fail at any of these six steps. Most failures happen at step 2, finding the media, because that is where platforms differ the most and change the most often.

How platforms actually deliver video

There are three common delivery styles, and a downloader has to handle all of them.

• Progressive file. One MP4 with video and sound together. TikTok and X serve most clips this way. This is the easy case.
• HLS. A playlist file ending in .m3u8 that points to hundreds of short segments. News sites, sports sites and live replays use it widely. The downloader has to fetch every segment and join them.
• DASH. Separate lists of video segments and audio segments. Nearly every YouTube format above 360p, Facebook HD and Reddit video arrive this way. Save only the video list and the file is silent.

On top of that, many platforms protect the media address. Addresses are signed and often expire within hours, some only work with the same headers the player sent, and some only appear after the page has run its JavaScript. Each protection defeats a different kind of downloader.

Method 1: one platform, one script

Sites like SSSTik and SnapTik understand exactly one platform’s link format and fetch the file that platform serves. Because they only have one target, they can be fast and simple.

Where it breaks: when the platform changes its page or API, the script stops until someone updates it, and it does nothing at all for any other site. You end up with a separate bookmark for every platform.

Method 2: an extractor list

Projects such as Cobalt keep a dedicated extractor for each supported service; Cobalt lists about 21. An extractor knows how one site stores its formats, so when it works it returns clean, complete results.

Where it breaks: if the link’s site is not on the list, nothing happens. If the site changed last night, its extractor fails until it is fixed. The list is only as current as its last update.

Method 3: sniffing in your browser

Extensions such as Video Downloader Plus and Video DownloadHelper watch what your browser loads and offer the streams they see. That catches players no extractor list covers, because the extension sees the real requests.

Where it breaks: it needs an install with permission to read the sites you visit, it only sees what plays in your own tab, and on phones it is mostly unavailable because Chrome on Android and iPhone does not run extensions.

Method 4: a real browser on a server

A headless Chromium opens the page on a server, presses play and captures the stream the player requests. It reaches videos that only appear after the page runs its JavaScript, and it needs nothing installed on your device.

Where it breaks: it is slower than a direct extractor, so used alone it wastes time on sites a simple extractor reads in a second.

Why chaining wins

Each method fails in different places. An extractor fails when a site changes its API. A one-site script fails on every other site. A sniffer fails when you are on a phone. A server browser is slow on easy links. A downloader that chains several methods only fails when every one of them fails on the same link, and that is far rarer than any single method failing.

How FileShot chains them

FileShot runs five engines in order on every link:

1. yt-dlp with rotating client profiles and browser impersonation, for the large set of sites it supports.
2. Cobalt, as a second extractor with its own per-service code.
3. A headless Chromium browser that loads the page, plays the video and captures the stream.
4. An HTML media sniffer that reads video addresses straight out of the page source.
5. Nextractor, FileShot’s own extractor for sites the others miss.

FileShot remembers which engine won on each site and tries that one first next time, so easy sites stay fast. It merges separate audio and video into one file, and the quality menu says “Video + Audio” so you know the download has sound. In a live sample on October 4, 2026, 22 of 112 successful reads, about 1 in 5, came from an engine other than yt-dlp. A downloader built on yt-dlp alone would have failed on every one of those links.

What no downloader can do

No tool can reach private posts, videos you would need to log in to see, or streams protected with DRM such as Widevine, FairPlay or PlayReady, which paid streaming services use. If a video is private or deleted at the source, every tool stops there. Download only videos you own, have permission to save, or that the platform’s terms allow.

How to pick a downloader

• It covers the sites you actually use, not just one.
• It uses more than one method, so a site change does not break it.
• It merges audio, so you never get a silent HD file.
• It needs no install for a one-off download.
• It publishes security results you can check.

Frequently asked questions

Why does a link work on one downloader and fail on another?

Because they use different methods. A one-site script, an extractor list and a server browser each fail on different links. A tool that chains methods tries the next one instead of stopping.

Why do some downloads have no sound?

The platform stored video and audio separately and the tool saved only the video. See Downloaded video has no sound?

Do I need to install anything to use FileShot?

No. You paste the link in your browser on any device. An optional extension exists for Chrome, Firefox and Edge, but the website works without it.

Compare: FileShot vs Cobalt, vs SaveVideo.me, vs Video DownloadHelper. Related: Why your video downloader is not working. Try it: paste a link.

Sorry, we couldn't load this post. Please go back to the blog index.