Which Adskeeper publishers send bots and invalid traffic?

Because Adskeeper syndicates its recommendation widget across thousands of publisher pages in the MGID network, the network can’t fully vet what happens upstream of each {widget_id} — and that gap is where most of its invalid traffic originates. Two mechanisms account for most of it, and they look different once you separate them. Scripted HTTP clients and stripped-down browser builds account for the more mechanical share: a widget click takes several hops to complete, and each hop is a chance for the tooling behind the request to leave a technical or behavioural trace that no real browsing session leaves.

The more common mechanism is commercial rather than mechanical: a publisher buys cheap pop or push traffic, drops it onto a content page carrying the widget, and lets the resulting impressions and clicks pad its numbers. Because that traffic arrives through an actual browser at an actual publisher URL, a plain IP blocklist can’t tell it apart from anything else — ValidVisit instead weighs the connection, the device and the session’s behaviour together, which is what catches it even when the click is routed through a residential-looking proxy. What’s left over is ordinary people who landed on the page and clicked the widget without much real interest; the quality score keeps that group apart from both of the above, since a bot-heavy {widget_id} is worth blocking outright while a merely low-intent one is often better handled with a lower bid. Whichever bucket a click falls into, {widget_id} is what the finding attaches to.