Ad Tech Briefing: Publishers are turning to AI-powered mathmen, but can it trump political machinations?

"Can the carrot of better empirical measurement, as opposed to the beat of a political drum, woo advertisers back to news content? For much of the past decade, brands' retreat from advertising next to news content has been framed as a means of risk mitigation in response to political polarization, as well as to other ills of the digital media sector, such as misinformation. Alternatively, as many media-buyers call it, the "I just don't want to get fired approach.""

"Inevitably, this caution leads to a "set-and-forget" mindset that leaves money on the table, often directed into the pockets of Big Tech rather than legacy media owners, contributing to the subsequent decline of the news publishing sector. In practice, it has often been enforced through blunt technical defaults: keyword blocklists, conservative suitability thresholds, and verification rules that treat "news" as a proxy for danger."

"The result has been a structural shift of budgets toward Walled Gardens, where advertisers continue to appear alongside news-adjacent content, but under different measurement regimes and with limited transparency - just think of how news content is often the topic of trending topics on social networks. That dynamic is now being questioned - less on ideological grounds than on economic ones."

Brands retreated from advertising next to news to mitigate risks from political polarization and misinformation, adopting a 'I just don't want to get fired' approach. That caution produced a set-and-forget mindset that diverts spending to Big Tech and accelerates decline of news publishers. Brand safety enforcement uses keyword blocklists, conservative thresholds, and verification rules treating 'news' as inherently risky. Budgets shifted to Walled Gardens with different measurement regimes and limited transparency. Publishers and ad-tech challengers argue legacy systems systematically over-block news, suppress supply, inflate prices, and deny advertisers engaged audiences. Hearst's use of Mobian contextual measurement found many impressions misclassified despite neutral or positive contexts.

Read at Digiday