Last-click attribution rewards the final stop before a purchase. Which means the article that did the convincing, three days earlier, gets credited as a bounce. If your reporting is strict last-click, you are almost certainly under-paying your editorial placements and over-paying your retargeting.
This is not a theoretical point. It is a pattern we see on our properties every month.
How the under-count happens
A reader lands on one of our articles via an answer-engine result. She reads the roundup. She clicks through to a brand she’s never heard of. She likes what she sees, closes the tab, and comes back two days later via a direct visit or a branded search. She buys.
Her buying decision was formed inside our article. The purchase itself happened in a different session, through a different channel. Last-click credits the channel. The article is invisible.
Multiply that pattern across a quarter and the under-count is not small. On any brand whose purchase cycle is longer than one session, you can expect your editorial-channel contribution to be meaningfully under-recorded.
What we track instead
We stitch sessions on our side using first-party identifiers that persist across visits inside the same property. When a reader clicks a partner link, we record the session. When a conversion comes back, we match it against the session rather than against the final click alone.
This is not a reinvention of attribution. It is a view brands have always been able to build. We build it because the alternative is to let last-click under-represent an entire category of placement.
“If your attribution can’t see the path, it will under-pay the stop that started it.”
What we do with the view
We share both figures with every brand we work with on the monthly reconciliation. Last-click, which is still the industry convention and still what we invoice against. And the stitched view, which is a more honest answer to the question “did the placement earn this reader?”
Brands can choose which figure drives their internal reporting. What matters is that both sides are looking at the same picture.
What this doesn’t fix
Cross-device remains hard. Session-stitching inside a property is first-party and reliable. Across properties is possible but we don’t do it by default. Across the open web is impossible without consent flows and identity graphs that would compromise reader trust.
Our claim is modest. Editorial placements almost certainly do more than last-click says they do. The stitched view narrows the gap. It does not close it.