Download websites that use Taboola

About Taboola

A recommendation widget is not installed to make a website better. Installing one is a decision to sell the attention of readers who have finished reading, taken by a business that needs the revenue more than it needs the page to stay clean. The marker therefore describes a commercial posture rather than a technical stack.

Taboola has been in that business since 2007, delivered by Taboola as a Proprietary JavaScript tag that publishers place below their articles. As of September 2026 we detect it on 110,471 sites. Every one of those properties belongs to somebody who decided that monetising the end of the page was worth the cost in appearance and page weight.

The population that produces is unusually well defined. Content sites with enough traffic to be worth monetising, publishers rather than brands, and operations with some kind of ad ops function, because the tag has to be placed and maintained. Sites selling their own product almost never carry it. As a filter for publisher-shaped businesses, presence of the tag is more reliable than most inferences drawn from a single script.

United States accounts for 66,645 of the detections and the .com zone carries the bulk of the rest, which follows the advertising market rather than the reading public. Native advertising money is concentrated where the buyers are, and the supply side arranges itself around the demand side.

Three groups buy a list like this for different reasons. Competing recommendation networks use it to size a displacement market, since a publisher already running one widget has proved willing to run the format at all. Advertisers and agencies use it to understand where their placements can actually appear, because the inventory is defined by the publisher network rather than by any list a vendor chooses to publish. Brand safety and research teams use it in reverse, to see which sites carry syndicated content and where a campaign might surface next to something awkward.

The caveat here is structural rather than technical. Recommendation tags live on article pages and not always on home pages, so any crawl weighted toward the front door of a site will undercount. A publisher running the widget on one section of an archive and nowhere else can be missed entirely. We record where we found the tag and do not extrapolate, which makes the count a floor rather than an estimate.

What a downloadable list answers that a per-domain check cannot is the population question. Checking one publisher tells you what one publisher does. Holding every domain where the tag appears, alongside the CMS, the ad stack, the hosting country and the other advertising tags on the same row, turns that into a market map, and the file is rebuilt daily, so a network gaining or losing publishers shows up as a difference between two copies instead of an impression.