Viant did not arrive in 2026 as a misunderstood innovator with a breakthrough product.

It arrived there needing a new story.

On paper, the last year looked passable. Revenue held. Earnings technically beat. CTV growth was still something you could cite without being laughed out of the room. But markets don't trade on "passable." They trade on belief. And belief drained out of Viant steadily across the year.

The stock lagged peers. The multiple compressed. Growth decelerated right as AI hype peaked and turned into expectation. ROCE slid into the kind of territory that makes capital allocators stop nodding politely and start sharpening questions.

None of this was fatal. All of it was corrosive.

2025 was the kind of year where fundamentals improved modestly while confidence quietly died.

That distinction explains everything that followed.

For most of its modern life, Viant sold like every other mid-tier DSP: better optimization, lower CPMs, smarter pipes, fewer hops, humans plus machines doing the right thing.

That pitch stopped working.

CPM optimization became a punchline. Traders went from heroes to overhead. And "AI-powered" became the most meaningless phrase in enterprise buying.

Meanwhile, the industry moved on. The Trade Desk stopped selling execution and started selling inevitability. Amazon sold closed loops. OEMs sold glass. Retail media sold receipts.

Viant was still explaining why its controls were better.

That's not a product problem. That's a narrative problem.

So Outcomes appears.

Fully autonomous buying. Business goals instead of knobs. No traders. No tuning. No second-guessing. Just set the goal and trust the machine.

This wasn't a random innovation. It was a response.

Outcomes is not the product of excess confidence. It's the product of a very specific realization: selling "how ads are bought" stopped working, so Viant needed to sell "what ads produce."

Not process. Not tooling. Results.

The word "outcomes" does a lot of work here. Too much.

The Timing Problem

Two days before Viant's push, ADOTAT published a funeral notice: "The Age of Outcomes is Officially Dead (And It Died Lying and Laughing)."

The piece featured former Google insiders—Erez Levin, 13 years at Google, top 2% of sales; Avinash Kaushik, 16 years as Analytics Evangelist, admitting they sold fraud, crap and calling outcomes what it actually is: attribution theater repackaged.

"Media is taking attribution credit for business outcomes it did not contribute to," Levin said, "and the propagation of easily gamed metrics masquerading as outcomes." Basically, you were lied to by the entire industry, if you are to believe him.

The numbers were brutal. Platforms tell you marketing drives 80-100% of sales. Real incrementality, according to Kaushik? Between 5% and 25%. The rest would have happened anyway.

Dr. Augustine Fou, MIT PhD, 15 years studying ad fraud: "The platforms and the fraudsters are claiming credit for sales that would have happened anyway, or that had already happened."

The industry consensus crystallized at exactly the wrong moment for Viant: "outcomes" is correlation in a lab coat. Coincidence mistaken for causation. The rooster thinking his crowing makes the dawn.

Viant launched a product called "Outcomes" directly into that.

No, really.

We didn't start here because we dislike automation. We started here because we read the documents.

Sales memos. Internal guidance. Framing language. The things customers never see but sellers repeat until they become muscle memory.

What we found wasn't fraud. It was something far more common and far more dangerous: the quiet loosening of definitions when old metrics stop persuading.

Outcomes optimizes to CPA and ROAS—the exact metrics the industry just called "easily gamed" and "masquerading as outcomes."

Incrementality is promised later. Causality is acknowledged but deferred. Control is reframed as friction.

That doesn't make Outcomes fake.

But it does make it not what it's being sold as.

Outcomes sounds like the future. It demos cleanly. It photographs beautifully. It gives Wall Street something to point at.

But the internal language tells a more complicated story about what gets optimized, what gets counted, and what quietly doesn't.

That gap is where embarrassment happens. That's where buyers get blamed later for trusting dashboards instead of causality. And that's the part nobody wants written down.

Here's where the logic breaks. And that's where we stop publishing publicly.