Let's start with a number: 61%.

Sixty-one percent of CTV impressions in one of Dr. Augustine Fou's recent analyses had no detectable source. None. Zip. Nada. They materialized out of the digital ether like a magician's assistant, except instead of a woman in sequins, what appeared was an invoice. Your invoice. For ads that ran, as best anyone can tell, absolutely nowhere on planet Earth.

If that doesn't make you put down your coffee and stare at the wall, check your pulse. We'll wait.

What did we actually find when we stress-tested all of this? That's Part Two. Subscribers only. Keep reading and you'll understand exactly why you need it.

Dr. Augustine Fou is either the most important person in digital advertising or its most entertaining menace. The answer is probably both, on the same Tuesday.

PhD from MIT. Runs FouAnalytics. Publishes charts that make CMOs reach for their blood pressure medication and send adtech sales teams sprinting toward calendar conflicts that absolutely did not exist five minutes ago.

But here is what his enemies, and there are many enemies, conveniently forget when they're composing their LinkedIn clapbacks:

This man cleaned up this industry in ways nobody else was willing to. Full stop.

When everyone else was celebrating programmatic's glorious infinite scale, Fou was in the log files. When agencies were packaging click-through rates into glossy decks and calling it performance, Fou was showing that half those clicks were bots in server farms located nowhere near your target demographic.

When brand safety became everyone's favorite buzzword, Fou had already documented, with actual receipts, that your ads were running next to content that would make your PR team change their names and move to another country.

He named names. Actual names. Not artful, pre-lawyered gestures at "certain bad actors." He put logos in his charts. He made specific companies look specifically, catastrophically terrible. He did it in public, in front of God and everyone, with the cheerful energy of a man who has absolutely nothing to lose.

Agency media directors loved him for it. His research was the document forwarded at 11pm with no subject line when a buyer needed ammunition to explain to a client why their programmatic campaign was performing like a three-legged horse at the Kentucky Derby. His work has saved real advertisers billions of dollars in fraud that would have sailed through undetected because nobody else had the nerve to go looking.

Advertisers revere him. The adtech industry despises him.

They've called him alarmist, a self-promoter, a crank, a menace, and our personal favorite, delivered with tremendous confidence by people whose research methodology is reading their own press releases: "not a real researcher."

His detractors also note, not without justification, that Fou is not audited by anyone. Not MRC. Not TAG. Not any independent third party with standing to validate or invalidate his work. He audits everyone. Nobody audits him. He is the one inspector in the entire ecosystem who answers to absolutely no one, which is either a principled stance against a corrupt certification apparatus or a remarkably convenient arrangement, depending on where you sit.

He also, and we are reporting this as fact because it is one, told this publication to go basically fornicate with itself when we invited him on our podcast to answer questions about his methodology. Dr. Fou does not do interviews where someone might push back. He publishes. He presents. He does not sit across from people with follow-up questions.

Whether this is the principled boundary of a serious researcher or the behavior of a man more comfortable making accusations than defending them is something we leave entirely to your judgment.

And yet. AND YET.

The fraud keeps showing up in the log files. Like a cockroach that survives every nuclear blast of industry denial, it is there, in the data, every time someone with the stomach to look actually looks.

How much of it is real? How much is methodology? How much is Fou's tool doing exactly what TVScientific says he privately admitted it cannot do? All of that is in Part Two. Subscribe. This is the part where it gets genuinely uncomfortable for everyone.

Fou has outlined five categories of CTV fraud happening right now, this quarter, while your vendor sends you a very confident performance report.

Your CTV ads are running on phones and websites. Mobile apps spoofing CTV device identifiers to collect CTV CPMs while delivering your ad to someone scrolling in a parking lot. The television is not a location. It is a concept. A line item attached to nothing.

Audience extension is eating your entire buy. You negotiate a direct CTV deal, shake hands, feel good about it. The seller silently routes 80% of your impressions to non-CTV inventory through "audience extension," which is the industry's elegant term for "we substituted something cheaper and hoped you wouldn't check." Eighty percent is not a rounding error. Eighty percent is the whole sandwich.

Your living room ads are going to cell towers. Impressions resolving to T-Mobile and Verizon wireless IP ranges instead of residential broadband. They almost certainly never reached a television in a home. They went into the programmatic equivalent of shouting your brand message into an empty parking structure at 3am and billing someone for the echo.

Some impressions skip humans entirely. Data centers. Fabricated bid requests. Servers talking to servers, exchanging synthetic impressions like trading cards, except you're paying market rate for cards that don't exist.

And then 61% with no detectable source whatsoever. Not mislabeled. Untraceable. Ghost impressions. Your media budget's equivalent of a missing person report where the person may never have existed.

Now. Honesty time.

Some of Fou's broader claims have real problems. His "turn off audience networks and eliminate 90% of fraud" rule is presented as universal law. The actual number varies enormously by campaign type and the 90% figure uses methodology he has not published for independent review. For a man whose entire brand is demanding transparency from others, this is a notable gap.

His hotel chain experiment, where a brand turned off programmatic for a week, saw no change, and he concluded the ads did nothing, is doing enormous narrative work on a thin empirical foundation. One week does not isolate causality. Brand effects have lag times measured in months. This is not an experiment. It is an anecdote wearing a lab coat.

And then there is what TVScientific told us. During a meeting with Fou, TVScientific representatives say he acknowledged that his product cannot actually do all of what he publicly claims it can do. Their words, not ours. A direct account of a direct conversation. A researcher who overstates his own tool's capabilities while criticizing everyone else's tools is a researcher whose findings need independent verification. Which, again, nobody is providing. Because nobody audits Fou.

In Fou's own data visualizations, when he breaks out where the untraceable impressions are coming from, one name keeps appearing at the top.

Index Exchange.

He did not editorialize. He simply put the charts out there with the logos visible, the way you leave evidence on a counter and walk out of the room.

Index Exchange, one of the largest and most respected SSPs in the business, appearing as a dominant pathway for CTV impressions that cannot be tied to a real screen, a real app, or a real human being watching anything.

If his methodology is sound, this is not a story. This is an earthquake.

They responded. We asked direct questions, they gave direct answers.

Their position: his claims are nonsense. His interpretations "rely on some pretty shaky interpretations." The suggestion that Index is a primary conduit for fraudulent CTV impressions does not reflect the evidence or how they operate.

They are, to be precise, not fans of the charts at all

And they get to say that. Index is not a pop-up exchange that appeared last spring. Real infrastructure, real publisher relationships, real standards work. Their point that they operate differently from the long-tail exchanges in Fou's typical case studies deserves to be taken seriously.

But "shaky interpretations" is what you say when the data is wrong. It is also what you say when the data is right and you need it desperately to be wrong. We cannot yet tell you which sentence you are reading.

Here is where we land honestly: we have real questions about both sides.

Millions of households now get home internet through Verizon and T-Mobile fixed wireless products, and those connections resolve to mobile carrier IP ranges. Which means some of Fou's "wireless IP equals suspicious" bucket might just be a family in Ohio watching football on their actual television. If so, his fraud numbers inflate significantly. How much? That math is in Part Two.

Index appearing in redirect chains might just be what happens when you're one of the most integrated SSPs on earth. Showing up in bad supply paths means something very different if you're in all supply paths. Does Fou do the denominator math? We checked. Part Two.

And pixel-based detection cannot see through server-side ad insertion, which is arguably the most legitimate delivery method in CTV. How much of Fou's "no detectable source" bucket is actually clean premium inventory his tool simply cannot reach? Part Two.

We asked Fou to come on and explain all of this. He told us, with characteristic efficiency, to go away.

So we kept digging without him.

Part Two is for subscribers. Here is what that means.

We go methodically through every category of Fou's framework and show you what holds up and what doesn't. We quantify the fixed wireless false positive problem. We examine whether Index's position in suspect chains is a market-share artifact or something that requires a harder explanation. We look at what Fou told TVScientific privately and what it means for everything he has published publicly. We let the critics with actual technical standing make their full case. And then we give you our honest, plainspoken verdict.

The CTV market is worth tens of billions of dollars. Your clients' money is in it. Your agency's reputation is attached to it. The possibility that a majority of it is unverifiable is not an academic question. It is a fiduciary one. You are either buying television or you are buying a very expensive story about television, and Part Two will tell you which one.

Subscribe. Part Two is already written. It will be worth every penny and several uncomfortable conversations with your programmatic vendors.

Seriously. Subscribe now. We did the work. Come read it.

ADOTAT is an independent publication. We received no payment from Index Exchange, FouAnalytics, TVScientific, or any related party. We reached out to all named parties prior to publication.