Download websites that use YouTube

About YouTube

Not every technology found on a page belongs to the people who built the page. YouTube is a plain example of the difference. What we detect is an embed, which is a block of third-party markup and script that a publisher pasted in to avoid hosting video themselves. The video, the player, the bandwidth and the measurement all belong to Google. The site contributes a rectangle and a decision.

The decision is what makes the signal worth having, because publishing video carries a budget behind it. As of September 2026 we detect embeds on 1,878,044 sites. Somebody at each of those organisations concluded that video was worth producing, or at least worth borrowing, and then chose a free hosted route rather than a paid video platform. For anyone selling video hosting, production, captioning, localisation or audience analytics, the population splits neatly into prospects who have proven the need and have not yet paid a specialist to meet it.

A second reading matters more to a different buyer. An embed loads third-party script and sets identifiers from a domain the publisher does not control, which is exactly the pattern that consent management and privacy compliance work exists to handle. A list of sites carrying embeds is, from that angle, a list of sites with a cookie question to answer, and the intersection with sites that run no consent platform is the part worth filtering for.

Detection deserves a caveat here, because embeds are easy to hide from a crawler. We read the public HTML of pages we fetch. An embed injected by script after a user gesture, or held behind a consent banner until permission is granted, or placed on a page we did not fetch, will not appear in the count. The figure is a floor rather than a total. Sites that do the privacy-conscious thing are exactly the ones likely to be missing from it, which is an awkward bias to carry into a privacy analysis and better said out loud than discovered later.

United States accounts for 865,569 of the sites where we find embeds, and .com carries most of the domain mix. The platform is Proprietary and free to embed from, so the signal describes editorial behaviour rather than spend. Media companies, universities, ecommerce product pages, agencies and one-person businesses all land in the same bucket, which means the technology on its own is a weak segment. Combined with a content platform, a commerce stack, a country and a suffix from the same record, the segment becomes usable and the weakness disappears.

One page can be inspected by hand, and for one page that is the sensible approach. The dataset exists for the opposite question. Which sites in a sector embed video and which conspicuously do not, how many of them run a consent tool alongside the embed, whether embed usage is rising or falling between snapshots, and which populations appear to have moved to a paid platform since the last one. Available since 2005, the embed has been a default choice for long enough that its absence on a commercial site is often the more informative finding, and only a full list makes absence a question you can ask at all.