Let's get something out of the way: nobody in programmatic woke up one morning and decided to build a fraud-friendly, margin-devouring, metric-hallucinating Rube Goldberg machine. We just kind of… got there. One DSP at a time. One vanity KPI at a time. One "but the ROAS looks great" at a time.

Liz DeAngelis — who cut her teeth in mobile app marketing, helped run the Ad Council's COVID-19 vaccine campaign when the stakes were actual human lives and not quarterly earnings calls, and now leads programmatic thinking at Brainlabs — has a way of describing this that should be printed on a plaque and hung in every trading desk in America:

"Programmatic is just capitalism at its worst. We set machines and told them to do a certain thing and they did that. However, the means for getting there and whether it's actually driving impact has yet to be determined."

Read that again. We built the machines. We told them what to optimize for. They did exactly what we asked. And then we had the audacity to act surprised when the whole thing turned into a swamp.

And who built the swamp? Everyone. "There were the buyers and the publishers. And then obviously this was a huge market. There was a lot of money to be made. And so anytime that's the case, people are going to find a way to weasel themselves in." She's not wrong. When Liz started in programmatic, she couldn't have named an SSP. The supply path was a black box nobody thought to open. "Anytime that you're not sure what's happening in the middle, it's just kind of a breeding ground for people to come in and exploit it."

She doesn't even think most of them are evil. Just opportunistic. Which, in this industry, is basically the same thing with better LinkedIn bios.

Everyone has a moment where the job gets a little too real. For Liz, it came early — and it came with a spreadsheet full of credit scores.

She was launching a B2C arm for a large investment firm, not long after 2008, when the financial industry was still picking shrapnel out of the drywall. The targeting strategy? Find people with high credit scores who were simultaneously drowning in credit card debt. The client had bought all the first-party data. Names. Addresses. Financial profiles. And Liz's job was to aim the cannon.

"We were renting America's money back at them," she said. The client saw themselves as offering a solution. Liz saw something closer to predatory lending with better creative. And the worst part? "Because they had all these addresses, we were able to do addressable TV and it performed incredibly well."

There it is. The original sin of programmatic, distilled into a single campaign: it worked beautifully, and it felt terrible. Performance metrics don't have a conscience column. That was, as she put it, her first time balancing "just because I can do something well, I need to actually feel confident in what it is that I'm doing."

Funny how the industry never built a KPI for that.

I gave Liz a multiple choice question — because this is the ADOTAT show and we treat serious topics with the gravity of a BuzzFeed quiz. What's the misconception about programmatic that causes the most long-term damage?

A) Automation equals optimization. B) Cheap equals efficient. C) Attribution equals truth. D) All of the above, and we keep funding it.

She picked D, obviously. But then she got specific: "B is probably my least favorite. I think B is absolutely the worst." And then the kicker — "I think it is equally fueled by C. Those two go hand in hand. This idea that cheap is better and then we use our attribution and we use ROI to kind of validate that."

This is the ouroboros at the center of programmatic: we buy cheap inventory because the numbers say it works, and the numbers say it works because we designed attribution to reward cheap inventory. It's a closed loop of self-congratulation. The snake eating its own tail, except the tail is made of bot traffic and MFA sites, and the snake is wearing a Patagonia vest at a conference in Cannes.

Every industry has its security blanket metric — the number you put on a slide when you need the room to nod and stop asking questions. In programmatic, that metric is video completion rate, and Liz is done pretending it means anything.

"Video completion rate makes no sense half the time," she said. And she's right. A video that autoplays muted in a 300x250 unit buried below the fold on a recipe site "completes" just as enthusiastically as one a human actually chose to watch. The videos running in doctor's office waiting rooms? Completing all day long, whether or not a single human eyeball is pointed at the screen.

But we cling to it. Because, as Liz put it, these vanity metrics "make people feel comfortable. They make it feel like their creative investment was worth it." They exist to soothe, not to measure. They're the weighted blanket of media reporting. The teddy bear you bring to the QBR.

Her broader point is the one that should keep CMOs up at night: "We are bombarded with thousands of ads every day. And maybe you remember two." And just because a video completed doesn't mean anything happened in anyone's brain. It means a file finished loading. Congratulations. So did my system update.

Most people in ad tech will tell you attribution is imperfect but useful — a necessary approximation in a chaotic system. Liz doesn't bother with that diplomatic framing.

"I kind of think all of attribution is theater."

There it is. No hedge. No "well, it depends on your model." Just: theater. She went further: "This idea that we can say confidently, this worked, this didn't, this drove, this sailed… and not account for every single other touch point that these people have, whether it was talking to a friend or hearing about it through some sort of organic method."

What she's describing isn't a measurement problem. It's a theological one. The industry built an entire belief system around the idea that we can map the exact moment a human being decided to buy something, and then attribute that decision to an ad impression served at 2:47 AM on a weather app. We didn't just drink the Kool-Aid. We built a factory for it.

"This idea that we can come up with the perfect sequence of ads and touch points and that we can personalize that to every single person is just theater. It's a great story and it's a fun story to tell, but it's not realistic."

And when I asked her about campaigns where attribution said it was a win but reality said otherwise, she barely had to think: "I think of most of them." Specifically, she pointed to retargeting — the industry's favorite trick of showing someone an ad for a product they already decided to buy and then claiming credit for the sale. "You're actually just looking at one single solitary metric. You're discounting all of these other touch points."

The uncomfortable truth Liz is circling is the one nobody wants on a slide deck: we might be dramatically less important than we think we are. "You talk to any brand and the amount of revenue that's driven from their marketing spend is usually not the lion's share. We're not the ones out there saving the world and saving these businesses and yet we certainly act like we are."

Here's where it gets fun. I asked Liz: if you were the supreme ruler of programmatic and could enforce one structural rule, what would it be?

Her answer was elegant: "Start at the end first. Think about where it is that you want to end up. And that is how you start your programmatic plan."

Simple. Revolutionary. And apparently terrifying — because the people who scream first? The 200-plus SSPs currently clipping the ticket between buyer and publisher. The content farms nobody's ever actually visited. "Findagrave.com," she said, and honestly that's the most devastating thing anyone has said on this show.

And then the punchline — the one that explains why nobody will ever actually do this: "If we start buying only ads where actual humans are looking at the screen and it's not littered with 50 ads at a time, the price is gonna go up."

Of course it will. Clean inventory costs more than the garbage we've all been swimming in. But nobody wants to tell the CFO that the reason CPMs just doubled is because we finally stopped buying ads that run on haunted websites where the only visitors are bots and the ghosts of abandoned browser tabs.

Liz knows this makes her unpopular: "I'm not gonna be a very popular" empress, she admitted. But that's the thing about thrones — they're not built for people who want to be liked. They're built for people willing to tell 200 SSPs to go find honest work.

That's Part 1. The free stuff. The appetizer. In Part 2 — behind the paywall, where the real conversations happen — Liz gets into the Pinterest/CTV Scientific acquisition and why "performance CTV" makes her nervous, Google's inability to reconcile its own platforms, what breaks first when agencies scale too fast, and the opinion she softens in public but holds like a grudge in private: that advertising has ruined the internet. Subscribe if you want the version she wouldn't say on a panel.