Can the Open Web Survive the Al Age? Kurt Donnell on Publisher Economics, Radical Transparency, and pubOS

Full analysis

The open web just took two hits in the same week, and only one of them was AI. Kurt Donnell, CEO of a publisher monetization and managed-services company, spent an hour with Signal & Noise host Rio Longacre arguing that generative AI breaks the old deal between publishers and platforms: search used to scrape your content and send traffic back, and now the AI answer gives the user what they wanted without the click. That is the structural argument. But the saved reading from the same week makes the case that Apple is the more immediate threat. iOS 27 quietly blocked a growing list of programmatic data companies from Safari's WebKit, including The Trade Desk's Unified ID 2.0, LiveRamp, ID5, and Permutive. So the open web is getting squeezed from two sides at once: AI eats the traffic, and Apple breaks the plumbing that lets publishers get paid for what traffic remains.

What's actually being decided here. Nothing, for the podcast itself. This is a practitioner's read on where publisher economics go, pegged to a product pitch for pubOS, a single-install header-bidding and monetization layer. The real decision sitting underneath, for every publisher and SSP operator listening, is whether to keep betting on anonymous open-web programmatic or to move money and engineering toward authenticated audiences and consolidated supply paths. That is hard to undo once you rebuild your stack around it. What sets the deadline is Apple, not AI. The iOS 27 blocklist is already live; the AI traffic erosion is gradual.

The Market Analyst

The supply-side is consolidating, and this episode tells you where. Donnell's model is six or seven large holding companies trading directly with six or seven large managed-inventory houses, citing a Jounce Media report he half-remembered that those houses run most managed inventory. He also notes The Trade Desk already buys directly from his company. Put those together and you get the direction of travel: the middle of the programmatic chain gets thinner. For a non-specialist: the many small toll booths between advertiser and publisher are collapsing into a few big ones. The losers are sub-scale SSPs with no aggregation story. The winners are whoever sits at the two ends and can offer the "one dollar in, see what comes out" clarity Donnell says the walled gardens already have. Apple's WebKit move accelerates all of it by making anonymous open-web impressions worth less.

The Skeptic

Donnell runs a publisher monetization business, so of course publisher economics are survivable in his telling. His own cohort was up 4 to 5% in traffic through 2024 while Longacre says he hears of 25 to 40% referral declines elsewhere. Both can be true, and the gap is survivorship: the publishers who hired a managed-service firm are the ones with direct audiences worth saving. The long-tail SEO sites that got "nuked" never showed up in his numbers. On the sub-$1 open-exchange CPM study from Permutive, Donnell says those figures only cover the open-exchange slice, not private deals. Fair, but that slice is exactly where long-tail publishers live. The optimism is real for enterprise publishers and false comfort for everyone else.

The Operator

The agentic trading story is pre-revenue and he says so plainly: more fingers on his hands than dollars that have run through real agentic trades, with tests to announce in 60 to 90 days. Treat any "AI agents negotiate directly" pitch as a 2027 problem, not a this-quarter one. The near-term break is Apple. If you run a publisher monetization stack, the iOS 27 WebKit blocklist means your Safari identity resolution is degrading right now, and no amount of header-bidding consolidation fixes an identity graph that the browser refuses to read. Donnell's own concrete advice is the cheat code worth acting on: build an email list. Authenticated subscribers reportedly command roughly 10x the CPM of anonymous traffic, per a prior guest, Scott McKinley of TriSet. That is the one move that survives both AI and Apple.

The Customer / End User (the buy-side)

From an agency or DSP seat, Donnell's "40-row spreadsheet" line is the whole problem. Buyers do not move budget to closed platforms because they love Meta; they move it because they can put a dollar in and count the dollars out. The open web asks them to reconcile transaction IDs across a dozen hops and trust publisher audience claims they cannot verify, which Donnell admits is why over 90% of publisher first-party data never makes it into the bid stream. For a non-specialist: advertisers pay a premium for simplicity and verified audiences, and the open web sells complexity and unverified ones. Every week the industry airs its own supply-path dysfunction in public, a rational buyer uses it as the reason to spend inside the walls.

The CFO

The pubOS pitch is total-cost-of-ownership: one install, consolidated billing, managed services, so a mid-sized publisher skips the in-house ad-tech engineering team. That math gets more attractive exactly as open-web yield gets harder, because the fixed cost of running your own stack does not fall when your CPMs do. But the payback depends on open-web programmatic staying a business worth optimizing. Donnell's loose citation of the old ISBA/PwC finding, that roughly half of advertiser spend reaches publishers, is the structural leak that consolidation is supposed to plug. If authenticated email is really a 10x premium, the better capital allocation is funding the audience-authentication work. That is where the revenue per visitor actually lives, and the anonymous programmatic optimization is not the path to get there.

Where the council splits.

The first disagreement is timing. The Market Analyst and the Operator both see consolidation coming, but the Operator puts Apple's blocklist at the top of the threat list while the Market Analyst treats the agentic future Donnell is selling as the bigger structural story. One is already in your P&L; the other is a 60-to-90-day press release.

The second is who survives. Donnell and the CFO case say managed consolidation plus authenticated audiences saves the publishers worth saving. The Skeptic says that framing quietly writes off the long tail, which is most of the open web by count, and that Donnell's cohort numbers only look good because the weak sites already died.

The third is whether simplicity is even winnable. The Customer says the open web loses budget on complexity and verification, full stop. Agentic trading and cryptographic audience verification might fix that someday, but nothing shipping today closes the gap with the walled gardens.

What this hinges on. Two beliefs. First, that authenticated audiences really do command the premium claimed, which would justify every publisher moving engineering out of anonymous programmatic and into email and first-party data. Second, that the supply chain genuinely compresses to a handful of aggregators, which would disintermediate sub-scale SSPs before the buy-side tooling is even ready. The council leans toward both being directionally right and toward Apple being the forcing event that makes them urgent rather than theoretical.

What to verify before betting on it. Watch whether the 60-to-90-day agentic tests produce actual dollars or another "tests planned" update. And measure your own Safari revenue line against the iOS 27 timeline before you assume identity resolution still works the way it did in August.

Prediction: Before the 2027 upfront negotiations begin in spring 2027, at least one more sub-scale independent SSP or exchange will be acquired by or fold into a larger managed-inventory aggregator, continuing the supply-path compression Donnell describes.

Confidence: Medium — the direction is clear; timing depends on individual deal cycles.

Why: Donnell's own account has The Trade Desk already buying directly from large supply houses and a Jounce report putting most managed inventory in six or seven hands, which means the many small SSPs in the middle add cost without adding clearing value. Apple's iOS 27 WebKit blocklist, live now and hitting The Trade Desk's UID2, LiveRamp, ID5, and Permutive, cuts the value of anonymous open-web impressions, and that is precisely the inventory sub-scale exchanges depend on. When your inventory is worth less and buyers want fewer hops, the exchange with no aggregation story cannot fund itself, so it sells or shuts. The opposite outcome, a sub-scale SSP raising money and staying independent, runs against both the buy-side push for fewer hops and a browser change already draining the value of its core supply.

Revisit by 2027-04-07: We're right if a sub-scale independent SSP or exchange announces an acquisition by or merger into a larger supply aggregator before spring 2027 upfront talks. We're wrong if no such deal is announced and the independent mid-tier SSP field is unchanged from today.