Every year, the advertising industry gathers around the Super Bowl like it’s both a religious experience and a quarterly earnings call, then acts personally betrayed when it behaves like neither. We want spectacle and spreadsheets, culture and conversions, a moment that unites America and a dashboard that can explain it to the CFO by Monday morning. So we scream about performance, blame streaming fragmentation, declare resellers dead, resurrect AI measurement with Greek god names, and still somehow funnel most of the money to Big Tech while pretending we didn’t notice.

The result is an annual industry nervous breakdown: brands spend millions chasing attention, then spend the rest of the year arguing whether attention counts unless it clicks, converts, or politely raises its hand in a clean room. Meanwhile, the audience is drunk, distracted, and emotionally invested in laundry detergent commercials. Welcome to the Super Bowl economy — where everyone swears the game has changed, yet keeps running the same plays and blaming the refs when the math doesn’t line up.

Every February, marketers rediscover the same ancient truth and pretend it’s a breaking revelation: people don’t immediately buy stuff during the Super Bowl. Shocking, I know. Apparently when viewers are drunk, screaming at refs and negotiating whether they still love football, they’re not impulse-buying insurance or fintech apps. Cue the industry-wide panic attack about “performance.”

The new boogeyman is fragmentation. Too many streaming endpoints. Too many dashboards. Too many “access points,” as agency execs politely say instead of “this is a mess.” The Super Bowl still delivers mass reach, but now everyone wants proof that the vibes converted. Spoiler: the vibes still convert. Just not on your preferred Tuesday-morning attribution slide.

The smarter marketers are finally admitting the obvious. The game is the ignition, not the engine. The real money gets made before and after, when retargeting, search lift and social stalking kick in. The Super Bowl isn’t a four-hour media buy. It’s a month-long ambush. Pretending otherwise is how you end up yelling at Nielsen like it owes you rent.

Let’s bury this stat properly: almost nobody buys anything during the game. One percent. That’s not a funnel; that’s a rounding error. But forty-plus percent say the ads make them curious. That’s the ballgame right there. Curiosity is the trigger. Conversion comes later, when the hangover fades and the credit card reappears.

This is why the sudden obsession with QR codes during the broadcast itself remains deeply unserious. Asking someone to grab their phone mid-commercial is like interrupting sex to pitch a newsletter. Post-game? Fine. Social feeds? Sure. During the Super Bowl? Congratulations, you just turned your brand into homework.

Connected TV absolutely drives performance. It just does it on a timeline that makes growth hackers itchy. No clicks. No instant dopamine. Just delayed intent that shows up as searches, site visits and eventually sales. TV isn’t broken. It just refuses to cosplay as paid social.

The industry has decided resellers are evil now, which is adorable given how long everyone profited off them. Platforms like The Trade Desk have gone full scorched earth, crowing about transparency and efficiency while quietly squeezing anyone without leverage.

Here’s the part nobody puts in the press release: small and independent CTV publishers need resellers to survive. They’re already getting kneecapped by FAST platforms that take half their inventory, control ad decisioning and keep the data. Resellers aren’t a luxury. They’re the oxygen mask.

Yes, reselling can cheapen inventory. Yes, it can commoditize content. But it also pays the bills while publishers try to build direct sales they’re structurally blocked from scaling. When platforms declare “no resellers,” what they really mean is “no middlemen except us.”

And don’t kid yourself. Major advertisers won’t notice if these publishers disappear. They’re already buying the same ten platforms on repeat and calling it diversification.

Not all resellers are arbitrage goblins. Some actually do the hard work of building meaningful audience graphs for communities big platforms love to virtue-signal about and quietly ignore. Latinx viewers. LGBTQ+ audiences. Multicultural FAST channels. The stuff decks are made of but budgets rarely reach.

When anti-reseller policies nuke these players, the damage isn’t theoretical. Inventory gets mislabeled. Fraud filters go medieval. Entire audiences vanish because they don’t fit neatly into someone else’s taxonomy. Efficiency wins. Diversity loses. Everyone pretends it’s accidental.

This is how “clean supply paths” start looking a lot like cultural monocropping.

While the industry argues about philosophy, Tubi just stacked Nielsen receipts. Bigger ad-supported share than Netflix. Younger demos. Tenfold growth in key age groups. Turns out free content plus ads still works, especially when you’re not ashamed of it.

The real flex isn’t the ratings. It’s legitimacy. Expanded Nielsen measurement means Tubi gets taken seriously in rooms where “FAST” used to trigger eye rolls. This is what happens when distribution, pricing and audience reality align instead of fighting each other.

YouTube quietly crossed sixty billion in revenue and still doesn’t get the respect it deserves from TV purists who think a remote control is a moat. Ads grew. Subscriptions exploded. AI-driven search made queries longer, messier and more commercial. The machine keeps eating.

The lesson here isn’t that YouTube is digital. It’s that it understands intent better than anyone else and monetizes it without pretending it’s art. While TV debates attribution windows, YouTube just counts the money and upgrades the model.

The Interactive Advertising Bureau rolled out Project Eidos, because apparently no measurement initiative is complete without classical cosplay. Meanwhile the ANA has Project Aquila, because why have clarity when you can have mythology?

AI-powered measurement is coming whether the industry likes it or not. Everyone admits current MMMs are garbage. Everyone swears AI will fix it in a year or two. Nobody agrees on standards. This is less a roadmap and more a race to control the narrative before someone else defines “truth.”

Hearst finally said the quiet part out loud: brand-safety tools are punishing news inventory for existing. Decades of journalism reduced to a blocked keyword list built for convenience, not accuracy. Context be damned.

By leaning into contextual analysis instead of blunt-force blocking, Hearst is betting advertisers might rediscover courage. Or at least curiosity. The irony is brutal. Brands happily buy news audiences inside walled gardens, then blacklist the same content when it’s sold transparently.

If advertisers really care about effectiveness, attention and value, local news is a bargain. If they care about not getting yelled at on the internet, they’ll keep hiding behind default settings and call it policy.

The common thread here isn’t streaming or AI or resellers. It’s fear. Fear of being measured wrong. Fear of standing out. Fear of admitting that marketing works on human timelines, not dashboards.

The brands that win aren’t chasing instant gratification. They’re playing offense before, during and after the moment everyone else obsesses over. The rest will keep blaming fragmentation while the audience moves on.

Stay Bold. Stay Curious. Know More Than You Did Yesterday.

Here's a fun number for you: 38 percent. That's how much of Hearst's newspaper inventory was being misclassified by the brand safety vendors advertisers pay good money to trust. More than a third of perfectly fine ad space, marked as radioactive because some keyword bot saw the word "death" and panicked like a golden retriever hearing thunder. Did it mean the death of a celebrity? The death of a TV character? The death of the third-quarter earnings forecast? Who cares! Block it all. Safety achieved. Invoice sent.

Hearst just rolled out Mobian's contextual and sentiment technology across its local newspapers and TV stations, and in doing so, told the legacy brand-safety industrial complex something it badly needed to hear: you're broken, and we have receipts.

The receipts are brutal. On the TV side, over 97 percent of Hearst's 2.5 billion impressions landed in safe environments. Eighty percent-plus sat next to neutral or positive content. The actual risk? Negligible. The perceived risk — the one that keyword blocklists have been selling to frightened CMOs for a decade — was enormous.

And that gap? That's not a bug. That's the business model.

One in three U.S. advertisers has blocked news entirely. Let that marinate. A third of the market looked at the Wall Street Journal, their local NBC affiliate, and the newspaper that's been covering their community since before their great-grandmother was born, and said: too scary. Can't do it. What if someone sees our toothpaste ad next to a story about city council zoning?

Meanwhile — and I cannot stress this enough — those same brands are gleefully running pre-roll on TikTok rage-bait and buying inventory next to conspiracy content inside Meta's walled garden, where the exact same topics exist but get laundered through platform-controlled "guardrails" that nobody audits and nobody questions because, well, it's Meta. They have a dashboard. Dashboards mean safety, right?

Wrong.

Publishers say roughly 30 percent of their inventory goes unfunded because of "no news" directives. The annual cost to U.S. news publishers? Low billions. Not millions. Billions. That's not a rounding error. That's an entire industry being slowly strangled by a system that was supposed to protect brands but ended up protecting nobody — except maybe the verification vendors cashing the checks. They're doing great, thanks for asking.

I sat down with Jonah Goodhart recently — co-founder of Moat (sold to Oracle for a tidy sum), now building Mobian — and the guy did not mince words. "Brand safety, we've not gotten right so far," he told me. "We've given brands a false sense of security. We've told them they have insurance, when in fact, I would argue we've not been doing a good job as an industry on getting that right."

Ouch. But he's right, and everybody in ad tech knows he's right. They just don't want to say it out loud because the current broken system is very, very profitable for the people running it.

Goodhart's thesis is straightforward and, frankly, kind of embarrassingly obvious once you hear it: brand safety was built in another era, using keywords and early machine learning trained on offensive words and images. It was fine for 2014. It's absolutely absurd for 2026. "No longer do we need to look for the word 'death' on the page and try to determine in some old machine learning kind of way whether it meant death of the Knicks' season, death of the future, actual human death, or death of a star," he said.

What large language models can do now — understanding context, source credibility, sentiment, intent, even the emotional register of a piece of content — makes keyword blocking look like using a metal detector to perform heart surgery. Sure, technically you're scanning. You're just not helping.

But here's the part of our conversation that really stuck with me. Goodhart borrowed a line from his father's book, Mobian Nights (yes, that's where the company name comes from — the man named his ad tech company after his dad's book, which is either deeply sweet or deeply nerdy, and I'm going with both): "We cannot sidestep context."

The same words mean completely different things depending on where they appear and who's reading them. A story about a house fire on the Wall Street Journal is journalism. The same keywords in a random forum post might be arson instructions. The brand safety industry has spent years pretending that distinction doesn't exist. Mobian's entire bet is that it's the only thing that matters.

And honestly? He's got a point. This is a guy who did 800 meetings in two years when he was building Moat because his advisor told him to go learn the market. Eight hundred. Most founders can't do eight before they start pitching. So when Goodhart says the measurement paradigm is broken, I tend to pay attention.

Hearst and Mobian's modeling suggests that reclassifying mis-flagged inventory could lower CPMs by up to 27 percent for buyers. Read that again. Twenty-seven percent. That's not a marginal optimization. That's the kind of number that should make every media buyer in America spit out their oat milk latte — if they're actually optimizing for performance and not just for the comfort of never having to explain a screenshot to their CMO at the next all-hands.

And that's the real test here. More than 65 percent of Hearst's local news audiences come back daily. Daily! That's an attention profile most "premium" digital environments would commit light fraud to achieve. Research consistently shows advertising in trusted news drives stronger lifts in brand trust and consideration than the programmatic slop most budgets end up swimming in. These are high-trust, high-attention impressions being sold at a structural discount because a keyword bot couldn't tell the difference between a war crime and a crime drama recap on CBS.

So Hearst is setting up a beautifully clean experiment: follow the performance signal back into local news, or keep clinging to "no news" defaults because your brand safety vendor — the one charging you handsomely for the privilege of blocking a third of your available inventory — told you journalism is dangerous.

As Goodhart put it to me, the companies doing brand safety today "have an opportunity to change their technology, to suddenly do something different, to embrace a new approach, to be transparent." They can evolve. Or they can keep selling keyword insurance policies on a house that isn't on fire.

The advertisers who figure this out first are going to scoop up a ton of cheap, high-performing inventory while everyone else is still blocking the word "explosion" on a story about a fourth-quarter earnings explosion.

Your move, CMOs. The data's on the table. The only question left is whether you're buying media based on performance or based on fear. And if it's fear — well, there's a whole industry of keyword vendors who'd love to keep selling you that.