Own the signal, move the auction, fix the cache: what publishers should do next
BY HAZEL BROADLEY, BEELER.TECH
Our second edition of The Pub Report, co-hosted alongside our friends Prebid.org, brought together digital media leaders John Rosendahl (Optable), Mike Racic (Prebid), Erik Svilich (Encypher), Brian Sardo (Microsoft), Kieran Greene (Shinka), and Karim Mourra (JWX).
The session moved across identity, provenance, Prebid Server, and Prebid Cache. However, the common theme was that the next set of adtech defaults are being shaped now.
So publishers, you need to get involved.
Own the audience before someone else does
Mike pointed to the issues publishers should be watching before Q4: agentic revenue opportunities, AdCP, Prebid Server, the Google antitrust trial, GAM, and unified auction changes are all connected to the same question. Publishers need to be in the groups where these systems are being discussed, including Prebid’s LLM group, which is open to publisher input this year.
John Rosendahl put identity back in its commercial context. Before programmatic, a publisher’s audience was the product, and buyers made assumptions based on the publication itself. Programmatic gave buyers more precision, but it also made impression value depend heavily on whether the right identifier was attached. Different IDs solve pieces of the problem across the open web, apps, clean rooms, people, households, and devices. They also come with different incentives, especially as holding companies buy identity infrastructure and build buyer-side fortresses.
John’s answer was to own the graph, not pick one ID and hope it wins. Publishers can still interoperate with external IDs, but rented infrastructure shouldn’t become the foundation of their data strategy. A first-party graph can collect publisher signals, add third-party inputs, support provenance declarations, and deliver a 4-7% eCPM lift in open programmatic when used for enrichment.
Stamp your content to prove authenticity
Erik Svilich brought the same control argument to content authenticity. C2PA lets publishers embed machine-readable, cryptographic proof into their content. It also provides buyers, licensing partners, and platforms with transparency over where content came from.
That kind of proof is becoming crucial because content is becoming harder to trust. Buyers need to know whether content comes from a real publisher, or if it’s been scraped, generated, repackaged, or detached from its original source. C2PA gives publishers a way to keep identity and rights information attached to the content as it moves, rather than letting that context disappear the moment the content leaves the page. The signing can happen inside the publishing workflow, and Prebid can then pick up the signal, cache the verification, and carry that provenance into the bidstream.
Erik was careful not to overclaim the CPM upside. The market is still early, and there isn’t enough public data to claim a firm uplift number. The stronger case is protection, licensing leverage, and brand-safety differentiation, especially as C2PA moves into other areas, such as AdCP and social environments that preserve provenance metadata.
Move the auction, then fix the plumbing
Brian Sardo’s Prebid Server update focused on another control point. In a typical GAM-led setup, Prebid Server runs an upstream auction, prepares bids with key values, and sends them back so GAM can make the final decision. That leaves Prebid demand sitting inside line items and price buckets, with limited transparency or publisher control.
The Prebid Server subcommittee is exploring whether it could become more of an orchestrator with basic decisioning. It could gather demand from bidders, an ad server, and potentially ADX if separation from GAM becomes possible, then make a basic decision about which kind of ad demand should win. The aim isn’t to recreate a full ad server, because campaign management, pacing, and frequency capping can still live elsewhere.
That could reduce line-item wrangling and improve performance by reducing client-side back-and-forth. It could also short-circuit an auction when a direct campaign will clearly win. Brian was clear that questions remain around legal developments, ad server sources, APIs, real campaign logic, ADX performance outside GAM, and agentic demand.
Kieran Greene and Karim Mourra then made the infrastructure point. A production-grade Prebid Cache needs real infrastructure behind it, with the database, routing, monitoring, scaling, and operational support to keep it reliable under load. The code is open source, but running it properly is still infrastructure work.
That doesn’t mean every publisher should try to build it themselves
If you already run your own Prebid Server instances, you may have the operational muscle to manage cache as well. If you don’t, the smarter move is probably to work through an existing partner, such as an SSP, managed platform, or independent provider. Karim also pointed to local cache as a useful alternative because it keeps the cache in the browser, removing a remote point of failure and helping more bids render. The caveat is GAM unified auction, where the workaround can reduce ADX spend even if the higher render rate makes the trade-off worthwhile for some publishers.
Participants agreed this is about practical control for publishers
Firstly, own the graph where you can and stamp the content before proof becomes a buyer requirement, then watch Prebid Server decisioning closely. Also, make sure that migration away from Xandr’s Prebid Cache doesn’t become a Q4 fire drill. The next 6-18 months will set a lot of defaults, and publishers need to show up with a point of view now – before they get built around someone else’s priorities.
👉Want to stay in the loop with what’s happening in the Beeler.Tech community? Subscribe to our newsletter. If you’re interested in attending or sponsoring a future event, you can explore our upcoming events here.