The useful bits
- The paper is dated 2 September; the 30 October roundup brought it back into this week's discussion. Its crawler and sampling design limit generalisation. The practical question is how an identity supplier distinguishes human audiences from automated traffic—not whether every matched identifier is invalid.
What the source covers
A study of more than 145 million header-bidding requests across 616,539 sites examined how extended identifiers were assigned to a crawler that declared itself a bot. The researchers found widespread EID use and limitations in how providers recognised that crawler. The paper is dated 2 September; the 30 October roundup brought it back into this week's discussion. Its crawler and sampling design limit generalisation. The practical question is how an identity supplier distinguishes human audiences from automated traffic—not whether every matched identifier is invalid.
Date and evidence
Paper: 2 September 2026 · Catch-up coverage: 30 October. Newsletter date is coverage/observation date; rollout scope remains qualified in the article.
Keep learning
What would you add?
Bring your experience to the conversation. The next useful perspective could be yours.
Talk with the collective
