Are PLAs Enough? A Conversation with Ari Paparo on the Future of Retail Media Ad Serving

Retail media is growing fast, but the tech powering it is still in its earliest chapters. To unpack where things are headed — and where RMNs are already repeating old mistakes — we sat down with Ari Paparo, one of the most widely respected operators in ad tech and the author of Yield, the best book yet written on how digital advertising actually works.

This episode started with a simple question about retail media ad servers… and very quickly turned into a much bigger conversation about PLAs, personalization, log files, attribution, and the future of the entire retail-media stack.

Why PLAs Are Easy Tech — But Not a Strategy

Ari is blunt about PLAs (Product Listing Ads / Sponsored Products): they are technically simple, but they’re not a strategy. They solve SKU-matching — not personalization, not creative sequencing, not identity, not attribution, and not frequency or pacing. RMNs relying entirely on PLAs risk building a search-only ad product in an ecosystem that increasingly expects cross-format relevance.

SKU-Matching Is Not Personalization

A key distinction Ari makes is between SKU-based relevance (what PLAs do) and user-based relevance (what modern ad servers and CDPs do). RMNs are now discovering the same gap publishers hit in the early open-web era: product-level signals don’t equal understanding the shopper.

Log-Level Data Still Determines Leverage

Ari has long argued that log files are the foundation of truth in advertising. In this episode we get into why retailers hesitate to share LLD with brands, how this echoes early Google/DFP dynamics, and why LLD is still the backbone for incrementality and comparisons across RMNs.

Attribution Is Breaking (Predictably)

Retail media is running headfirst into the same attribution issues the open web spent years wrestling with: overcounting, last-click bias, broken MTA, walled-garden measurement, and the difficulty of isolating lift. RMNs that want to differentiate will need to build measurement that works beyond their own walls.

Do We Even Need Ad Servers?

One of the most interesting turns is Ari questioning the premise itself: do we need ad servers in retail media? The discussion hits where decisioning lives, what logic belongs with the retailer vs with the advertiser, whether RMNs are becoming small walled gardens, and what the minimum viable “ad server” even is.

Is Retail Media Repeating Open-Web History?

In many ways, yes. Retailers are now confronting the same tensions that shaped digital advertising: who controls the auction, who owns the logs, how personalization is defined, and what counts as transparency or fairness. The parallels to the early DFP era are hard to ignore — and meaningful for any RMN building out its stack.

Was Ari Interviewing Us?

Maybe. But that’s what you get when your guest has built bidders, worked at Google and AppNexus, founded Beeswax, written the definitive book on ad serving, and interviews half the industry on Marketecture. The result is one of our clearest conversations yet about how retail media actually works.

Listen to the Episode

Are PLAs Enough? — with Ari Paparo

(video/audio embed here)

Ari’s book Yield:

https://www.amazon.com/dp/1646871580/

‍