Episode 156: Does quality really matter? Erez Levin weighs in

2026-01-16

Download

Ari Paparo and Eric Franchi are joined by Erez Levin, a former Googler who focuses on media and inventory quality. They dig into what “quality” really means in programmatic advertising, why short-term outcomes can be misleading, and how incentive structures have pushed spend toward lower-value impressions. Takeaways Quality is best understood through effectiveness, but most measurement overweights short-term signals. “Not all impressions are created equal.” quality varies by context, format, and goal....

Takeaways

Quality is best understood through effectiveness, but most measurement overweights short-term signals.
“Not all impressions are created equal.” quality varies by context, format, and goal.
Video definition loopholes led to premium pricing for lower-attention formats and contributed to market confusion.
MFA, SPO, and curation are connected symptoms of incentives that reward cheap scale and vanity metrics.
Verification helps, but quality needs to be addressed across the full media workflow, including experimentation and MMM.
Agentic buying could either improve quality controls or make it easier to optimize only to what’s measurable in the near term.
Publisher traffic declines reinforce the difference between commoditized content and differentiated journalism or creator-led media.

Chapters

00:00 Welcome and introduction to media quality

01:29 Marketecture Live updates and announcements

04:18 Erez Levin on why advertising quality matters

06:00 Defining quality vs outcomes in digital advertising

08:30 Brand impact, long-term effectiveness, and mental availability

09:43 Lessons from Google AdX and DV360

10:58 Video misclassification, IAB definitions, and market fallout

14:03 Outstream video, pricing, and mobile gaming use cases

17:00 MFA, SPO, and the real causes of inventory quality problems

19:03 Tools, verification, and the role of measurement frameworks

20:30 Agentic buying, AI, and control over media quality