Native campaigns commonly show a wide spread in results across the sites they run on, at similar click prices. The cause is that a click is not a uniform unit.
Source context sets expectations before the click
A reader arriving from a specialist publication has different expectations than one arriving from a general-interest page they were browsing casually.
Those expectations carry into the landing page, where one visitor is prepared to read and evaluate and the other is still deciding whether to stay.
The same landing page therefore converts at very different rates depending on where the click came from, without anything about the ad changing.
Some inventory generates clicks without intent
Placements on pages built for continuous scrolling, or units positioned where accidental taps are likely, produce clicks that reflect a mis-tap rather than interest.
These sources often look attractive on cost per click precisely because their clicks are cheap and plentiful.
They reveal themselves in post-click behavior: very short time on page, near-zero scroll depth, and effectively no progression to any further step.
Device mix explains part of the spread
Mobile clicks dominate most native supply, and mobile visitors abandon slow-loading pages faster and complete long forms far less often.
A source that is disproportionately mobile will underperform on any conversion that requires sustained input, regardless of audience quality.
Separating source performance by device usually shows the problem is a mismatch between the destination and the device rather than the publisher.
Blocking is the primary optimization lever
Native platforms let buyers exclude individual sites or publisher identifiers, and systematic exclusion is where most of the improvement in these campaigns comes from.
The discipline is to judge sources on a post-click outcome rather than click volume, and to allow enough delivery on each before deciding.
Blocking too early on thin data removes sources that would have performed, which is the most common way an optimization pass makes a campaign worse.
Supply changes continuously
Native networks add and remove publishers constantly, and traffic can shift toward new sources without any change on the advertiser's side.
A block list built once will drift out of date, and performance can decay for reasons that appear nowhere in the campaign settings.
Reviewing source-level reporting on a regular cycle is therefore ongoing maintenance rather than a setup task, and it is what separates stable native accounts from volatile ones.