
Opinion. A response to Dave Morgan's November 18 op-ed in The Current, "The TV upfronts need a major overhaul. It could happen sooner than we think."
There are maybe five people in American television whose read on where the ad market is heading I would stop what I'm doing to read. Dave Morgan is one of them, and on the specific question of how video audiences get counted and sold, he may be first on the list.
He built Real Media. He built TACODA and sold it to AOL. He has spent the better part of thirty years inside the actual machinery of audience measurement and transaction, not adjacent to it, not commenting on it from a consultancy, inside it. He runs Simulmedia because somebody had to make linear and streaming buyable in the same breath and he decided it would be him. When Morgan says the upfront is giving way to an always-on forward market, that is not a provocation. That is a man describing a building he has helped wire three times.
So let me say clearly what this is before anyone decides otherwise. This is not a teardown. This is a discussion.
A word about how this is supposed to work, because our industry has gotten genuinely bad at it.
Professionals disagree. That is the job. When two people who have spent decades in the same business read the same numbers and come to different conclusions, the adult move is to say so plainly, show your work, and let the other person answer. You discuss. You talk. You take the reply seriously. Most of the time you both come out of it knowing something you did not know going in, which is the entire point.
What you do not do is insinuate, slander, or make it about the person. That is what children do when they cannot argue the substance, and there is far too much of it in ad tech, usually from the people with the least to say and the smallest record to point at.
So let me be exact about what this is. Two people who like and respect each other, looking at the same evidence, disagreeing about what it proves. Dave has forgotten more about television measurement than most of this industry will ever learn, and he may well be right and I may well be wrong. If he convinces me, I will say so in print, under my own name, as plainly as I am disagreeing with him now. Knowing him, if I convince him, he will do exactly the same.
Morgan published that argument this week in The Current. The structural reasoning holds up. What I want to talk about is the evidence sitting underneath it, because every load-bearing number in the piece is doing work it was not designed to do, and not one of those numbers is Dave's fault.
They are the industry's numbers. We all use them. I have used them. He is simply the most credible person in the room repeating what the room says, which is exactly why his piece is the right place to stop and ask whether the room is right.
And honestly, there is nobody I would rather have this argument with. You can only have a real conversation about measurement with someone who has actually built measurement. Everyone else is reciting.
Where Dave is right, where I disagree, and why
Let me put the ledger up front so nobody has to guess at my position.
Where he is right. The direction is settled, and he called it earlier than most. Dollars are moving to streaming and they are not coming back. His account of why the upfront survived six decades, that sellers wanted guaranteed volume and buyers wanted wholesale pricing and neither side would move first, is the best one-sentence explanation of that institution I have read anywhere. And the agentic plumbing is real, with actual transactions behind it rather than slideware.
Where I disagree. Not about the destination. About the clock, and about what we are reading it on. It is one disagreement with three faces.
One. The evidence cannot carry the claim. The viewing figures come from a Nielsen report that Nielsen disclaims for advertising purposes, and the dollar figures lost their most important qualifier somewhere in transit. Streaming won primetime. It did not win the upfront.
Two. The yield story runs backward. His always-on market promises media owners higher prices. The 2026 upfront delivered more money at lower CPMs. A continuous market dissolves the scarcity the upfront manufactures, and scarcity is what holds price up.
Three. The prerequisite is missing and almost nobody is building it. Continuous trading needs one currency both sides compute against, live, with an agreed tiebreak. We have four certified currencies, an incumbent carrying nearly all the volume that spent this spring working through its accreditor's list, and no tiebreak at all. The annual upfront hides that problem by settling it in a room once a year.
Why it matters. If I am right, the automation arrives before the agreement does, because automation has vendors and agreement has committees. What we get is not the clean market he describes. It is a continuous one running on contested numbers, where whoever has the better model quietly wins the arbitrage and nobody can prove it for years.
Everything below is the evidence for each of those, starting with the part where he is right.
Start with what Morgan has dead right
The parts that hold are the more interesting half of the story.
Consolidation is real and accelerating. Holding companies are contracting. Entertainment companies are merging, spinning and restructuring at a speed that makes annual planning cycles look faintly ridiculous. His read that both sides of the table now have to move faster is not forecasting, it is observable from any quarterly call you care to open.
The agentic plumbing genuinely is being laid. He is right that this is not vaporware. NBCUniversal, FreeWheel and Newton Research ran a cross-platform agentic buy with RPA in January, using Model Context Protocol so buy-side and sell-side agents could negotiate with each other, including against live sports inventory. That is a transaction, not a slide. The subsequent NBCU, FreeWheel, Newton and Carat work aimed at upfront commitments is also real.
His diagnosis of incentive lock is the sharpest thing in the piece. The upfront survived six decades not because anyone enjoyed it but because sellers wanted guaranteed volume and buyers wanted wholesale pricing, and neither side would move first. That is the correct explanation for why a widely disliked institution persists, and most people writing upfront obituaries never get within a mile of it.
He is also right about direction. Nobody serious thinks the dollars are flowing back toward linear. The 2026 book is not a blip, it is the continuation of a line that has been bending the same way for a decade.
The disagreement here is not about where television is going. It is about whether we can currently prove anything about how fast.
The disclaimer Nielsen prints on every single release
Morgan grounds his viewing argument in Nielsen's Gauge. So does everyone else. So does the trade press. So do the investor decks. So, frequently, do we.
Here is the sentence Nielsen attaches to the Gauge, month after month, inside the release itself. The Gauge and Media Distributor Gauge "do not reflect Nielsen's currency TV ratings that inform advertising sales." Nielsen adds that the Gauge reflects total TV viewing, ad-supported and non-ad-supported together.
Read that twice, because the industry has been politely ignoring it for about four years.
The Gauge is a time-spent snapshot. It is not the currency. It counts hours a household spends in front of a television set and sorts them into broadcast, cable, streaming and other. It does not separate a Netflix subscriber watching with no ads from a Tubi viewer sitting through an eight-minute pod. It is not what advertising is bought and sold against. Nielsen says so on the tin, every time.
Which means that when any of us uses the Gauge to argue about where advertising dollars should go, we are using a ruler that measures a different object. Morgan's piece is an argument about how ad commitments get made. Its evidence is a report whose publisher disclaims it for that exact purpose.
I want to be precise about who that criticism lands on. Not Dave. All of us. The most-cited number in television advertising is a number its own publisher tells you not to use for television advertising, and we have collectively decided to live with that because it is the only figure that arrives monthly with a clean chart attached.
It gets more uncomfortable. Nielsen has said it is rebuilding the Gauge for the fall season to reflect its currency enhancements, with back data to help clients through the transition. The metric the entire debate rests on is under reconstruction while the debate is happening.
The 52 percent is not in the report
Morgan writes that per the Gauge, cable and broadcast accounted for "nearly 52% of ad-supported TV viewing time" in the first quarter.
The Gauge does not publish that figure.
What the Gauge published in that window was broadcast and cable as shares of total television usage. In the February 2026 edition, broadcast sat at 21.7 percent and cable at 20.0 percent. That is roughly 42 percent of all TV time, not 52 percent of ad-supported TV time.
To reach 52 you have to strip non-ad-supported viewing out of the denominator. Out goes ad-free Netflix, ad-free Disney+, ad-free Max, every premium tier where a commercial never runs. Nielsen does not publish that cut in the Gauge.
Here is the part that matters, and it is a point in Morgan's favor rather than against him. The derivation is almost certainly directionally correct. Remove subscription-only viewing from the base and linear's share of genuinely ad-supported time climbs steeply, because linear is nearly all ad-supported and streaming is emphatically not. His underlying claim, that ad-supported viewing still overindexes on linear, is probably true and is the single most useful observation in the piece. I would go further and say it is the observation the rest of the industry keeps skating past while it writes linear's obituary.
My question is narrower. A derived figure got presented as a Nielsen figure. In an argument about the market becoming more data-driven, the provenance of the data is not housekeeping.
There may well be a Nielsen product behind it that I have not seen. If there is, the correction is mine and I will print it with my name on it.
July is not a normal month, and everyone in this conversation knows it
The second viewing number is streaming at 49 percent of total TV in July, the most recent month measured when he wrote.
That is accurate. It is also the most flattering month on the board.
Nielsen's own framing of that report credits the final weeks of the FIFA World Cup together with streaming's traditional summer boost. Streaming climbs every summer. It climbed in May too, to 48.6 percent. July stacked a global tournament on top of the seasonal peak.
So the comparison in the op-ed sets a seasonally peaked, World Cup inflated July streaming number against a first-quarter linear number. Two periods, two denominators, two definitions of what is being counted. Stacked, they read as a trend line. Separated, they read as a seasonal wave riding on a real but slower structural drift.
One thing worth retiring while we are here, and I am retiring it from my own notebook. The old complaint that the Gauge inflates streaming by dumping virtual MVPD viewing into the streaming bucket is no longer true. Since the February 2023 interval, live programming watched through YouTube TV or Hulu + Live TV is credited to the network that aired it rather than to the app. Nielsen fixed it. Anyone still running that argument, very much including ADOTAT, should stop.
Streaming did not pass linear. It passed linear in primetime.
This is the correction that matters most, and it did not start with Morgan.
The op-ed says streaming surpassed linear in upfront dollars for the first time, streaming up 30 percent, linear down 9. That traces to The Current, reporting estimates from Media Dynamics Inc, the consultancy run by Ed Papazian that has been counting this market since before most of the people arguing about it had jobs.
The Media Dynamics estimate is $17.2 billion for streaming against $16.6 billion for linear. Those are primetime figures.
Media Dynamics also estimated roughly $12 billion more flowing into linear upfront buys outside primetime: early morning, daytime, weekend, early news, late night, plus national syndication. Their own conclusion is that linear still captures more upfront dollars in aggregate than ad-supported streaming does.
The accurate sentence is therefore: streaming took a majority of primetime upfront dollars for the first time, while linear in total still took more money. That is a genuine milestone. It is not the milestone the headline promised.
And the headline is where this started. The Current's own framing dropped the primetime qualifier. Morgan was writing for The Current, citing The Current, trusting what The Current had published. A man reading his own outlet's reporting and believing it is not a failure of rigor. It is how a qualifier vanishes from an entire industry's shared understanding in a single hop, and it is the most quietly alarming thing I found all week.
It also matters for the thesis. If linear still takes more upfront money in aggregate, then the institution Morgan is describing as obsolete is still, in dollar terms, mostly a linear institution. That does not make him wrong about the direction. It does change the clock.
Two forecasters, two sets of numbers, one milestone
Here is a detail that did not make any headline and probably should have.
Media Dynamics put the crossover at $17.2 billion streaming against $16.6 billion primetime linear. eMarketer, forecasting the same market, put it at $17.73 billion CTV against $16.98 billion primetime linear.
Same milestone, same year, same claim of a first. Roughly half a billion dollars of daylight between the two estimates on the streaming side alone, and different definitions underneath, since one is counting streaming and the other is counting connected TV, which are overlapping but not identical categories.
Neither is wrong. Both are estimates, built from different inputs, by serious people. But notice what happens next: the trade press picks whichever figure it saw first, the qualifier falls off somewhere in the retelling, and within a month the industry has a shared fact that nobody can reproduce from a primary source.
This is the actual condition of the market Morgan wants to automate. We cannot presently agree on how much money moved through the upfront that just happened, to within a rounding error of half a billion dollars, in a market where the whole thing is roughly $34 billion.
Now imagine agents transacting against that in real time.
Dollars up, prices down, and what that does to the yield argument
Morgan's forecast for media owners in the always-on market is attractive: higher priced premium inventory, better yield, more durable advertiser relationships.
The 2026 upfront went the other way on two of the three.
Media Dynamics put total primetime upfront volume near $33.8 billion, up about 9 percent, while CPMs fell. Streaming's average adult CPM dropped 4.9 percent to $25.90, with free ad-supported services cutting hardest to buy share. Primetime CPMs across linear and streaming together fell about 6.2 percent. Inside the linear number, broadcast slipped about 5 percent to $8.6 billion and cable fell about 7 percent to $8.0 billion.
Streaming's 30 percent dollar gain arrived alongside falling prices. That is volume growth, not yield growth. More inventory, sold cheaper.
Which is the real counterargument to the always-on thesis, and it deserves an answer rather than a wave. When you move from an annual negotiation that artificially constrains supply into a continuous market where every impression is available every second, the default direction for price is down. Scarcity is the product the upfront manufactures. Remove the pageantry and the scarcity leaves with it.
There is a version of this argument that favors Morgan, and I want to state it fairly because it is the strongest thing his side has. If continuous trading lets a seller find the buyer who values a given impression most, instead of selling it in an undifferentiated block nine months early, then average price can fall while realized yield per impression rises. That is the entire promise of liquidity. Programmatic display made exactly that promise. We all know how that priced out, and the people who made the promise were not lying, they were just wrong about who would capture the surplus.
This is the piece of his argument I would most like to hear him defend, because if anyone has a good answer it is him, and I cannot construct one myself.
The currency itself spent this year under review
Everything above is about the Gauge, which is not the currency. So let us talk about the currency, because the currency had a year.
Nielsen's National Big Data + Panel service is the thing that actually informs advertising sales. It was accredited by the Media Rating Council in January 2025, the first hybrid panel and big data product to get there, and Nielsen endorsed it as currency going into that upfront.
On March 3, 2026, weeks before the upfront season that produced the numbers Morgan is citing, the MRC published a status update telling the market it had found "seemingly unusual changes in measurement results" in that service, going back to at least early 2025. It identified four priority areas requiring action, including representation levels in the panel and the variability between Big Data + Panel results and panel-only results.
The VAB, representing the sellers, was blunter, calling the currency unstable, unpredictable and decimating demographics.
Nielsen then delayed two of the changes the MRC wanted until August, pushing fixes past the upfront. Accreditation held throughout, and on May 27, 2026 the MRC reaffirmed it and shifted back to its ordinary annual audit cycle, saying Nielsen had addressed the key concerns on representation and weighting.
So the honest summary is: the currency was never withdrawn, it was questioned in public by its own accreditor, during the exact buying season whose results we are now using as evidence about the future of television.
I am not claiming the 2026 upfront numbers are wrong. Those dollar estimates come from Media Dynamics and eMarketer, not from Nielsen ratings. What I am saying is narrower and, I think, harder to argue with. In a single twelve-month stretch, the industry's time-spent report carried a disclaimer against using it for advertising, and the industry's actual advertising currency carried an open MRC action list. Those were the two instruments available to anyone trying to describe what happened this year.
Worth noting, and I mean this as credit rather than a jab: The Current covered the MRC situation. The reporting was in the same publication that ran the op-ed. The information was not hidden from anyone, including from the argument.
We do not have a currency. We have four.
Morgan's always-on forward market has a prerequisite he does not spend much time on, and it is the whole ballgame: a currency both sides accept, live, continuously.
Here is what we have instead.
In September 2026 the U.S. Joint Industry Committee re-certified Comscore, iSpot and VideoAmp as national currencies for the 2026 to 2028 term, after what it called its most comprehensive evaluation yet, with expanded testing on stability, sports and long-tail programming. That is real work and it deserves respect.
It also means the market now has four certified national currencies, plus the Gauge sitting outside all of them as the thing everybody quotes in public.
And the concentration is lopsided. Nielsen is estimated to hold somewhere around 80 to 90 percent of transacted volume as a national TV and video currency, leaving VideoAmp, iSpot and Comscore to split the remaining ten to twenty points between them.
Sit with what that combination means. We have certified competition and we do not have a competitive market. The alternatives cleared a rigorous bar and still transact a sliver. The incumbent carries nearly all the volume and spent the first half of this year working through an accreditor's priority list.
Now put the always-on market on top of that. Continuous trading does not just need a number. It needs the buyer's agent and the seller's agent to be computing against the same number, at the same moment, with an agreed tiebreak when they disagree. The annual upfront papers over multi-currency chaos because humans negotiate it away in a room once a year. They argue, they settle on whose number governs the deal, they sign, they go home.
Take the room away and you have not removed the disagreement. You have automated it.
The agentic upfront currently has one luxury retailer in it
Morgan cites Fox's agentic operating system talk and the NBCU, FreeWheel, Newton and Carat agent as evidence the automation is arriving.
Both are real. Both are also smaller than the sentence implies.
The January NBCU work with RPA was explicitly a proof of concept, demonstrated by video, with first live deployment planned for one brand's first-quarter campaign. The Carat version debuts with a single unnamed luxury retail client before any expansion across dentsu.
That is a pilot, and pilots are how everything real begins. I am not mocking it. The Model Context Protocol work in particular is the right architecture, and the people building it know exactly what they are doing.
But the distance between one custom agent running a luxury account and an always-on forward market absorbing the national video economy is measured in years and in standards nobody has written yet. Agent-to-agent negotiation needs a shared vocabulary for what an impression is, whose measurement governs, what counts as delivery, and what happens when the two sides' numbers disagree by eight percent. None of that is a modeling problem. All of it is a governance problem, and governance is the slowest thing this industry does.
What an always-on market actually requires
Here is where Morgan's argument and mine arrive at the same place from opposite doors.
The annual upfront is a blunt instrument, but it is a negotiated truce: both sides agree on a measurement, agree on a price, and go home. Replace it with continuous trading and that measurement has to be live, trusted, and identical for buyer and seller, every second, permanently.
We are nowhere close. We are so far from it that the most cited measurement in the business carries a printed warning against using it for advertising, is being rebuilt mid-argument, sits alongside four certified currencies of which one carries nearly all the volume, and generated a headline figure this year that shed its most important qualifier somewhere between the research firm and the trade press.
Morgan's future needs a referee. The current referee has a sign on its desk saying it is not the referee, and the one that is spent the spring answering an accreditor's questions.
That is not an argument against his thesis. It is an argument that his thesis has a prerequisite nobody is pursuing with half the enthusiasm they bring to agent demos. Building AI that transacts video in seconds against a currency the industry cannot agree on is building a very fast car before anyone has settled which side of the road we drive on.
And here is the uncomfortable part for both of us. The automation will probably arrive before the agreement does, because automation has vendors and agreement has committees. Which means the most likely version of Morgan's future is not the clean one he describes. It is a continuous market running on contested numbers, where the party with the better model quietly wins the arbitrage and nobody can prove it for three years.
I would very much like him to tell me I am wrong about that.
One more thing, about where this ran
The Current is The Trade Desk's publication. It carries a disclaimer saying op-eds represent the author's views and are not endorsed by The Current or The Trade Desk, and that disclaimer sits right at the bottom of Morgan's piece. Nothing was hidden from anybody.
But look at where the argument lands. In the always-on market he describes, the TV ad economy is not run by walled gardens. It is run by a limited number of integrated DSP and SSP platforms handling the plumbing, data and transactions for the largest buyers and sellers.
That is a specific, contestable prediction about market structure, and it happens to describe the future that is best for the company that owns the outlet it appeared in.
Morgan believes it. He has believed versions of it for years and published them in plenty of places The Trade Desk does not own. This is not a payola story, I am not insinuating one, and anyone who reads it that way is reading badly. The disclaimer is honest, the byline is honest, and Dave Morgan has never in his life needed a publication to tell him what he thinks.
It is worth noticing out loud, though, that our trade press increasingly runs through outlets owned by the companies with the most to gain from particular conclusions. The right response to that is not suspicion of writers. It is more disclosure, and more people checking the arithmetic underneath.
Which is all this was.
So let's actually talk about it
I have been in this business long enough to know what it feels like when someone picks apart your numbers in public, and I know it is not always received as affection. Let me be plain about it, then.
I think Dave Morgan is right about the destination. I think the evidence available to him, to me, and to everyone else writing about this cannot yet establish the timeline, because the instruments we are all reading are not built for the question we are all asking. One of them says so in writing. The other one spent the spring under review. That is not a disagreement about television. That is a disagreement about whether we currently know what we claim to know, and it is the most important unsettled question in this industry.
Everything else we argue about downstream, the agents, the yield, the holdcos, the walled gardens, is downstream of a number none of us can fully vouch for.
I would rather argue that with him than agree about it with almost anyone else.
Dave, the drinks are on me. I think you are early, not wrong, and I would like to hear why I'm the one who has it backwards.
This is opinion and commentary, a response to a published argument rather than a reported news story, so no comment was sought from the parties named. Each load-bearing figure in Morgan's November 18 op-ed was checked against primary sources: Nielsen's Gauge releases and their methodology notes, the Media Rating Council's March 3 and May 27, 2026 status updates on Nielsen's National Big Data + Panel service, the U.S. Joint Industry Committee's September 2026 recertification of Comscore, iSpot and VideoAmp, and 2026 upfront estimates from Media Dynamics Inc and eMarketer as reported by MediaPost, Señal News and others; agentic buying claims were checked against the January 2026 NBCUniversal, RPA, FreeWheel and Newton Research announcement and later trade coverage of the NBCU, FreeWheel, Newton and Carat work. Limitations: upfront dollar figures from Media Dynamics and eMarketer are estimates rather than audited spend and the two do not agree; we worked from published Gauge releases rather than underlying data files; the 80 to 90 percent transacted-volume share attributed to Nielsen is a market estimate, not a disclosed figure; and the source of the 52 percent ad-supported viewing figure could not be identified and may come from a Nielsen product we have not seen. The author and Dave Morgan are friends, and the author has known many of the executives named here across two decades in this business; ADOTAT has no financial relationship with Simulmedia, The Trade Desk, Nielsen, NBCUniversal, Media Dynamics or any JIC-certified measurement provider, is reader-supported, and accepts no paid placement in editorial.

