The media buyers who swear their campaign is “optimized” but can’t explain the funnel. They’re quoting ADOTAT+ on calls like they thought of it themselves.

Because we don’t just break news—we expose it, dissect it, and serve it raw with receipts. Vendor drama? We’ve got the texts. Ad tech fraud? We name names. Holding company whispers? We’ve already heard them scream.

This isn’t a newsletter. It’s a crowbar for prying open the black box of this industry.

If you’d rather skip the recycled LinkedIn threads and get to the part where you actually know what you’re talking about, join ADOTAT+. Or don’t—and let your competitor use our reporting to crush your next pitch.

Welcome to the second act of this somewhat nutty series.

In Part I, we asked if Nielsen was still the king or just the last guy in the room with a landline.

Now let’s talk about the mess everyone is pretending not to see: the idea that “views” and “impressions” are worth anything in a world where your ad can “run” on a gas pump screen while someone’s swearing at their debit card.

This industry has been selling vapor and calling it rain.

And “reach”? It’s the biggest myth of all.

The Gross Rating Point—GRP for short, Great Ridiculous Placeholder for long—was the metric that launched careers, justified Superbowl ad prices, and kept media buyers smug and employed.

It didn’t matter if grandma left the TV on while asleep with the cat drooling on the remote.

If it was on? It counted.

Fast forward to now: digital has taken this flawed logic and turbocharged it with fraud, bots, autoplay, and a sprinkle of delusion.

A “view” can mean literally anything.
— One second? Sure.
— Muted? Why not.
— Hidden behind ten tabs and a broken iframe?

Congrats, you’ve been “reached.”

If the bar for success is barely registering in someone’s peripheral vision, then we should just start counting dreams and hallucinations too.

The problem? We’ve built an entire multi-billion-dollar industry on this house of sand. I get it. I got it.

Let’s move on.

Desperate for redemption, the measurement world has now discovered “attention.”

You can almost hear the collective relief at the trade shows.

Don’t worry, it’s not a total scam—just the same hype, but now with eye-tracking and buzzwords that sound like they came from a TED Talk held inside a WeWork.

Now we’re told: “We don’t just count views. We measure presence. We track attention signals. We know if they were breathing in your ad’s direction.”

But let’s not get ahead of ourselves.

The arms race for attention has created a new swamp of vendors, platforms, and “signals” that are barely standardized, totally siloed, and often less reliable than your aunt’s Wi-Fi.

Some of the new tools?

Brilliant. Others?

Just retargeting with a monocle.

Every sales deck now reads like it was written by a failed philosopher: “We don’t just drive reach—we deliver resonance.”

Cool story. Show us the receipts.

Marketers want incrementality, lift, and causal impact. What they get? Attribution spaghetti and dashboards that crash more often than the X app on Android.

And don’t even get me started on the Rube Goldberg machine that is digital attribution. Privacy changes, cookie deprecation, SKAN this, IP masking that—it’s all made the entire stack so opaque you’d think it was designed by Enron.

Here’s the truth: Most platforms can’t prove squat beyond “we showed something to someone somewhere and now your chart has a bump.”

Let’s talk tech bloat. Most marketers are running Frankenstacks of ad servers, pixel farms, viewability trackers, brand safety vendors, clean rooms, and attention dashboards. It’s like duct-taping 14 thermometers together and still not knowing if you have a fever.

Everyone’s got data—terabytes of it. But confidence? Not so much.

After all that spend, complexity, and slideware, here’s what most marketers are left with:

We built this industry on GRPs, then doubled down on impressions, and now we’re frantically trying to patch the holes with “attention.” But here’s the dirty secret:

More metrics haven’t made us smarter. They’ve just made us busier.

And unless this industry finds a way to connect what’s being measured to what actually matters—like sales, sentiment, and brand love—we’re just rearranging the pixels on the Titanic.

Yes, I made fun of their courtroom win. Yes, I laughed at the LinkedIn victory laps and the “Nielsen Killer” clickbait.

But here’s the thing:

HyphaMetrics isn’t trying to kill Nielsen. They’re trying to license data to everyone who wants a shot at replacing it.

Which, frankly, might be the smarter play.

So let’s stop the snark for a second and talk about who Hypha actually is—and why they might matter more than you think.

At first glance, HyphaMetrics looks like a measurement startup clinging to the bottom rung of the ladder. Small panel. No seat at the currency table. Court cases with Nielsen. Easy to dismiss.

But that’s not the business model. They’re not trying to be a currency. They don’t want to be your next rate card.

Hypha licenses viewing data—meticulously gathered from a growing, person-level panel—to everyone else in the measurement space: VideoAmp, iSpot, Comscore, publishers, brands, calibration vendors, planning teams, media mix modelers, and yes, maybe even Nielsen someday. As CEO Joanna Drews puts it, “We’re more than willing and eager to sell our data to Nielsen—or any currency provider.”

They’re not fighting to replace the system. They’re building the data plumbing behind it.

Hypha’s secret sauce is a piece of software called UNIe—short for Unified Neuromedia Identification Engine. Yes, it sounds like a late-90s cypherpunk project, but it’s actually one of the more quietly impressive pieces of tech in ad measurement.

Most companies use ACR (automatic content recognition) data from smart TVs. It’s fast, broad, but shallow. You’re only seeing what people with a certain brand of TV are watching, and only if that TV is turned on and connected.

UNIe works differently. It doesn’t rely on a static content library. Instead, the system grows its catalog with each new exposure, constantly updating and classifying media content in real time. According to Drews, “Our library grows with us each video exposure, which enables limitless TV omnichannel measurement.”

Think of it less like Shazam, more like a custom neural net mapping every second of screen time in a household—across streaming, linear, gaming, and digital. It knows what was on, who watched it, and how long they engaged. That includes co-viewing, Spanish-language content, and even episode-level tracking.

And they do it without pretending to be omnipotent. They’re currently in 50 households, aiming for 5,000 by Q1 2026. It’s small, but it’s surgical.

This is the smartest part: HyphaMetrics doesn’t care who wins the currency war. Because their business model is designed to survive all of them.

“Hypha is a Data-as-a-Service company, not a currency,” Drews emphasizes. “Our data is available to every single company in our industry, and it can be used for any bespoke data modeling, like calibration, MMM, planning, and so much more.”

Translation? They’re the infrastructure layer. The middleware. The API of measurement. If Nielsen holds, Hypha still wins by licensing niche solutions. If VideoAmp, iSpot, or Comscore gain ground, Hypha fuels their backends. And if the entire market eventually runs on open modeling? Hypha becomes the underlying data everyone plugs into.

They don’t need to own the tower. They’re selling the blueprints, the bricks, and the mortar.

The recent court victory over Nielsen wasn’t just a win for Hypha—it was permission to exist without legal sandbags tied to their feet. Drews called it “thrilling,” but didn’t miss a beat pivoting back to business. The company’s strategy hasn’t changed, even with leadership shuffles: keep growing the panel, keep licensing the data, keep feeding the machine.

And they’re not naïve about how long this will take. “It takes time to change a system that’s underpinned TV ad buying for decades,” Drews says. “Unseating an established player takes a lot of work and an appetite for playing the long game.” She doesn’t see the challengers as failing, just grinding—as any upstart has to.

The same applies to Hypha itself. There’s no instant scale. No IPO buzz. Just quiet, deliberate growth.

Because the fight for the future of measurement isn’t just about who gets to be “the currency.” It’s about who owns the inputs that define what counts.

And Hypha’s trying to be the source of truth beneath the truth. Not a judge—but a record keeper. Not a disruptor—but a connective layer between the old world and the one that’s (maybe) coming.

Also, let’s be honest—they’re kind of weird, in a good way. They’re not trying to dazzle you with size. They’re not launching Super Bowl campaigns. They’re building, slowly, for people who actually care about accuracy.

As Drews puts it, “Imagine a media marketplace where all participants—advertisers, content creators, platforms, buyers—are working from the same data set with shared visibility into consumer behavior.” She believes Hypha is already making that possible.

Hypha isn’t betting on a title fight. They’re building the data spine that supports the whole ecosystem, quietly powering the players clawing at Nielsen’s crown.

That’s not just smart—it’s necessary.

And for an industry obsessed with replacing Goliath, it’s worth paying attention to the one company everyone might have to license from to do it.

Stay bold. Stay curious. Know more than you did yesterday.

Let’s not sugarcoat it: AI didn’t fix attribution—it just gave it a prettier interface and an even fuzzier confidence level.

The truth is, if attribution was broken before, AI simply added more gears to the Rube Goldberg machine.

But that doesn’t mean it’s useless. It just means we need to stop treating machine learning like divine prophecy and start treating it like what it is: an incredibly fast intern with a PhD in correlation and a total lack of emotional intelligence.

So, here’s the setup: Google and Nielsen got together like two corporate Avengers and ran a mega-analysis of 50,000 brand campaigns and 1 million performance campaigns.

Their weapon of choice? AI-powered Marketing Mix Modeling (MMM)—designed to parse ROAS, isolate sales effectiveness, and control for everything from economic shifts to someone sneezing during your Super Bowl spot.

The outcome? Predictably impressive. The kind of results that fill conference slides and drive demand gen managers to send excited Slack messages.

Let’s be clear: this isn’t trivial. It means AI can chew through billions of data points, find patterns that would make a human analyst’s brain melt, and spit out smarter budget allocation suggestions—sometimes in minutes.

Synergy > Silo: The biggest flex of AI-powered MMM is showing that combining AI formats (not siloing them) gives better results than running single-channel strategies. Shocker: collaboration works—even when it’s between soulless machines.

Speed, Baby: Traditional MMM moves like a freight train hauling molasses. AI MMM is more like a Tesla in Ludicrous Mode. Faster insights, more granularity, and shorter test cycles.

Budget Whisperer: AI doesn't just look at spend—it reallocates it. On a good day, it can optimize media mix in near real-time, recommending which channels are worth doubling down on and which should be sent to the farm upstate.

Here’s the rub: prediction is not causation, no matter how many TensorFlow models say otherwise.

AI-powered MMM still can’t tell you why something worked. It can highlight a pattern, sure, but it doesn’t know about your Black Friday disaster, the competitor’s surprise drop, or the fact that your product had a recall in Ohio. Context is still king, and AI is still context-blind.

Interpretability Black Hole: Many AI models are basically horoscopes with code—black boxes that spit out a number without explaining the reasoning. Want to know why “OOH” just got a 30% weight increase? Ask your data scientist, and prepare for a 90-minute lecture on Bayesian priors.

Garbage In, Garbage Out: AI is still only as good as your input data. If you’re feeding it junk, don’t be surprised when it tells you to double down on that display campaign no one saw.

MMM’s Altitude Problem: Even with AI, MMM lives at 30,000 feet. It looks at macro trends, not individual journeys. So it might tell you that Q4 looked great for branded search, but won’t tell you if that was because of your new CTV campaign or your viral TikTok involving a llama.

AI MMM shines at simulation, not storytelling. It’s best for asking “What if we spent 20% less on Meta and shifted it to YouTube?” It’s not great at answering, “Did this campaign actually drive those incremental purchases from Gen Z women in Dallas?”

That line between prediction and proof? Still razor-thin.

And here’s where human beings (yes, we’re still needed) come in.

AI-powered MMM is a huge leap forward—but it’s not the moon landing.

It gives marketers more speed, better granularity, and smarter “what-if” tools. But it doesn’t solve the biggest problem in advertising: knowing what actually moved the needle, and why.

The attribution game isn’t over. It’s just more complicated, more automated, and more full of caveats than ever.

So before you crown AI the new oracle of marketing, ask yourself: Are you seeing insight—or just very convincing math?

And if your model says the best thing to do is double your spend on podcast ads in Alaska at 2AM... maybe double-check the inputs.