Podcast episode

Ep. 134 Why Programmatic Advertising Needs a Reset with David Nyurenberg of InterMedia Advertising

by Ken Rona ·June 4, 2026 ·The AdTechGod Pod

ai-in-adtech ctv measurement privacy programmatic

TL;DR

David Nyurenberg, SVP Digital at InterMedia Advertising (a 50-year-old TV performance agency), argues programmatic advertising's transparency problems are baked into the industry's economic incentives and that linear TV's discipline — buying at the show level, backed by ComScore/Nielsen research — offers a better model. He's a skeptic on the current AI/agentic/AdCP hype, comparing it to the blockchain and Flash cycles, and makes a substantive case that CTV needs show-level transparency to mature. Light on hard numbers or news; valuable mainly as a buy-side perspective on CTV transparency and AI hype.

What was covered

Nyurenberg's background: Entered the industry via an internship at mobile game publisher Gameloft, where he helped roll out their first programmatic product and became an early campaign manager. Later worked at multiple managed-service ad-tech vendors, ran his own data-engineering consultancy for four years, built CTV practice at Rain the Growth Agency (~3 years), and now leads digital at InterMedia Advertising, brought in to build a CTV-capable digital practice.
Why programmatic is "broken": He argues the core problem isn't inventory quality, targeting, or technology individually, but the industry's incentive structure — that certain non-transparent practices are "built into the system" to keep revenue flowing, and that fixing it would mean tearing the system down and starting over.
Linear TV as a model: Linear buying requires justifying spend with real research (ComScore, Nielsen) and buying at the show level, versus programmatic's habit of selecting audience segments that "just sound nice" with no visibility into how they're built or refreshed.
AI / agentic / AdCP skepticism: Nyurenberg says he hasn't seen agentic advertising or AdCP (the "Ad Context Protocol," a proposed standard for AI agents to transact ad buys) work at meaningful scale. He references AI working groups where the message is "don't believe everything in the trades," and notes test budgets of 20K–50K prove nothing about powering billions in spend.
CTV show-level transparency — the central thesis: He argues CTV must offer show-level data like linear does to reach its potential, because without it buy-side optimization is severely limited.
The VPPA as root cause: He attributes the lack of show-level data partly to the Video Privacy Protection Act (VPPA) — a ~30-year-old law passed after a politician's video-rental records were leaked — being retroactively applied to digital streaming, plus publishers' fear that buyers will cherry-pick top shows and hurt yield.
Optimism on talent: He sees a rising "contingent of nerds" — more sophisticated, vocal buy-side marketers openly sharing ideas, versus a past culture of guarding tactics.
How he stays current: Daily scans of Adweek, AdExchanger, and Digiday; the Marketecture episode every Friday (hosts Ari Paparo and Eric Franchi); and peer conversations.

Notable claims & predictions

On incentives: "It's the incentives that this industry runs on… these business practices… need to keep going for the lights to stay on. Digital introduced a lot of bad practices and bad thinking… to chart a better course would almost be [tearing] everything down and starting from scratch." — Nyurenberg
On linear vs. programmatic rigor: In linear, "everything has to be backed up by real data and research to validate why the money is going where it's going," whereas programmatic segment selection is often based on what "just sounds nice." — Nyurenberg
On AI/AdCP hype: "It feels like Flash all over again… I'm much more conservative when it comes to hype cycles." He drew an explicit parallel to the blockchain wave, noting that at Dentsu he audited many blockchain-built companies and "not one of those companies still exists." — Nyurenberg
On scale as the test: "Although something works at a minimal budget, that's great. So does a flat tire, but if you're going 80, it's shredding… for any solution to be rolled out, it needs to be scalable and tested at scale." — Nyurenberg
On CTV transparency unlocking spend: Media buyers "will be able to spend more budget when they're able to optimize and drive performance." Withholding show-level data "makes [publishers'] lives easier in the short term, but… handicaps them in the long term." — Nyurenberg
On the cherry-picking fear being misguided: A buyer who only wants one recognizable hit show is "a silly media buyer because they need to maximize reach" — with outcome data tied to shows, the algorithm will surface long-tail content that performs. — Nyurenberg

Why this matters for ad-tech operators

CTV show-level transparency is a live battleground with budget consequences. Nyurenberg frames the lack of show-level data — not just genre or app-level — as the binding constraint on CTV optimization. For SSPs, CTV publishers, and FAST aggregators, the str

Full analysis

Decision Council — Briefing Mode

Step 1 — Frame

A veteran TV-performance buyer (David Nyurenberg, InterMedia Advertising) makes two claims worth an operator's attention: (1) CTV will stay underfunded until publishers expose show-level data the way linear TV does, and the main excuses — a 30-year-old privacy law (the VPPA) and fear of buyers cherry-picking hit shows — are partly self-inflicted; and (2) the current AI / agentic / AdCP excitement is a hype cycle that hasn't proven it works at real scale.

What's actually being decided: Not a single transaction. For the reader, it's two posture questions — how hard do I push (or resist) show-level CTV transparency in my 2025 roadmap and renewals, and how much do I invest now in agentic/AdCP buying versus waiting for proof?
Reversibility: Both are Type 2 (easy to reverse) at the pilot stage, Type 1 once you've rebuilt pricing, packaging, or your buying stack around either bet.
Forcing function: None urgent from this episode. It's a buy-side opinion, light on numbers or news. Treat it as a temperature read, not a trigger.

Honest impact assessment: Low-to-moderate. No data, no deal, no regulatory move. The value is that a credible buyer is naming the friction point — show-level transparency — that determines how fast CTV budgets grow. That friction is real and worth pressure-testing.

Step 2 — The Council

The Customer (the media buyer) This is Nyurenberg's home turf, and he's largely right: without knowing which shows an ad ran against, optimization is guesswork dressed up as targeting. Linear buyers justify every dollar against Nielsen/Comscore show ratings; CTV buyers pick audience segments that "sound nice" with no view into how they're built. Plain version: TV buyers know exactly what program their ad sat next to; streaming buyers often don't, and that blindness caps how much they'll spend. If a publisher hands me clean show-level outcome data, I move budget toward it — fast.

The Operator (publisher / SSP yield) Easy to say "just expose show-level data." Try doing it Tuesday. The VPPA isn't only an excuse — class-action plaintiffs have hammered publishers over exactly this, and legal will veto anything that smells like tying a viewer to a title. Then there's yield: if buyers can see the hits, the long tail gets starved, my floor prices erode, and my packaged bundles fall apart. Nyurenberg's "a smart buyer maximizes reach" assumes buyers behave well. They don't. The cherry-picking fear is rational, not silly.

The Skeptic The load-bearing assumption is that transparency unlocks net-new spend rather than just reallocating the same dollars toward premium shows and away from everything else. That's unproven. Linear's "rigor" is also partly nostalgia — Nielsen panels are tiny and contested; the discipline is cultural, not technical. And "the whole system needs tearing down" is the kind of total critique that's unfalsifiable and conveniently lets a buy-side agency look pure. Bad incentives are real, but agencies live inside them too.

The Engineer On AdCP and agentic buying, his scale point is the sharpest thing in the episode: something working on a $20–50K test says nothing about powering billions in spend — "a flat tire works until you go 80." That's correct. The gap between a demo where an AI agent negotiates a buy and production systems clearing billions of auctions a day is enormous: latency, fraud, deterministic billing, auditability. The blockchain-in-adtech comparison is fair — most of those companies are gone. But "unproven now" isn't "never." Protocols mature quietly while the trades overhype them.

Step 3 — The Tensions

Does show-level transparency grow the pie or just reslice it? The buyer says budgets expand once optimization is possible. The publisher operator says it collapses the long tail and erodes yield. Both can't be fully right — and which one wins depends on whether outcome data actually surfaces performing long-tail shows (Nyurenberg's claim) or whether buyers just chase the same five hits.
Is the VPPA a real wall or a convenient one? The Customer treats it as an excuse; the Operator treats it as live legal exposure. Reality: it's both, and that's exactly why it persists — genuine liability gives cover to a yield-protecting instinct.
Is AI/agentic skepticism wisdom or a buyer protecting the status quo? The Engineer validates the scale critique; the Skeptic notes that a managed-service agency has every reason to distrust automation that could disintermediate it.

Step 4 — Synthesis

The episode hinges on two beliefs you can actually test:

Belief 1: Show-level data expands CTV budgets net-net. This is the high-value question. Don't accept it on faith and don't dismiss it. For publishers/SSPs: run a controlled test — expose show-level outcome data to one or two sophisticated buyers under privacy-safe aggregation (clean room, k-anonymity thresholds) and measure whether their total spend rises or merely shifts toward premium. If spend grows, transparency is a growth lever and you should productize it ahead of competitors. If it just reslices, the operator's caution wins and you protect the bundle. For buyers: the agencies pushing hardest for this data should win share as it arrives — build the optimization muscle now.
Belief 2: Agentic/AdCP buying is years from scale. The council leans toward Nyurenberg here, but with a caveat. The right posture for an operator isn't "ignore it" or "bet the roadmap" — it's cheap optionality: run small pilots, sit in the standards working groups, but don't restructure pricing or headcount around agentic buying until something clears real volume with auditable, deterministic results. The flat-tire test is the right gate.

My view: The transparency thesis is the more important and more actionable of the two, and it's underrated precisely because it's boring legal-and-yield plumbing rather than AI sizzle. Whoever solves privacy-safe show-level reporting first — likely via clean rooms rather than raw data — gets a real edge with serious buyers. The VPPA is a solvable engineering-and-legal problem, not a permanent ceiling. The AI skepticism is healthy but should be held loosely; "the trades oversell it" and "it will never matter" are different statements, and the smart operator keeps a foot in the door.

The thing to be wary of: a buy-side agency declaring the whole system corrupt while operating comfortably within it. Take the diagnosis seriously, discount the purity.

What did we miss? Is there a persona we should add for this specific decision? A General Counsel would sharpen the VPPA question — the difference between "we can't" and "we won't" on show-level data is mostly a litigation-risk judgment, and that's the real swing factor for publishers.

Listen to the episode →