VideoAmp says AI will connect TV planning to proof of what worked. In the same quarter it walked away from its MRC audit, one of its investors wrote its VideoAmp stake down by about 80%. In 2019 this was a company in serious growth mode. In 2026, I'm not sure it survives the rest of the year.

I liked VideoAmp. I've had many conversations with its founder, Ross McCray, and I thought the idea was right. Television needed a second currency, built on set-top box and smart-TV data instead of a panel, and somebody had to take a real swing at Nielsen.

In May 2019, VideoAmp raised $70 million, led by the Raine Group, to build out its team and buy companies. That was serious growth mode. By its 2021 Series F, Axios later reported, it was valued at $1.4 billion.

It's no longer a good idea. The idea didn't change. The company did.

On October 16, AdExchanger's Alyssa Boyle laid out VideoAmp's new pitch like she was paid: AI agents that connect media planning to measurement, recommend where money should move mid-campaign, and, as SVP of agency partnerships Andrea Zapata frames it, make sure media performs and then prove it.

The AI claims read like a desperate attempt to stay relevant in an industry that has started to ignore VideoAmp. A measurement company that wouldn't finish its own audit is now selling software to prove everyone else's outcomes.

Frank Kowalczyk: I was at VideoAmp from 2021 to 2023. Nice office. Good snacks. Somebody put the word currency on a slide my first week and nobody ever took it down, including after we stopped being able to explain it. I left before the AI showed up. Bought the boat instead of therapy. Neither one floats.

Neuberger Next Generation Connectivity Fund, a listed closed-end fund, bought 1,393,993 VideoAmp Series F1 preferred shares on January 4, 2022, for about $22 million.

For four years the fund carried them at roughly what it paid. Its schedule of investments showed $21,999,998 on October 31, 2025, and the same $21,999,998 on April 30, 2026.

On July 31, 2026, the fund's filing with the SEC put the same 1,393,993 shares at $4,400,000.

Same shares. About 80% less, in one quarter. That's the quarter VideoAmp withdrew from its MRC audit, and the quarter before it cut another fifth of its staff.

To be precise about what this is: one fund's mark on one class of VideoAmp preferred stock. It isn't a valuation of the whole company, and it isn't a cash balance. Nobody outside VideoAmp has published either. But funds don't cut a position by four-fifths for no reason, and whatever the fund learned between April and July, it wasn't good news.

Frank Kowalczyk: Twenty-two million to four point four. The Kowalczyk Ratio on that is about the same as my boat. Bought it for twenty-two grand. Guy offered me four-four. Didn't ask about the engine either.

Per AdExchanger, VideoAmp uses ACR data, the automatic content recognition built into smart TVs, for CTV targeting and measurement. Zapata's example: ACR shows what people watch when they're in market for a new car or planning a vacation.

Chief product officer Josh Hudgins says the priority is connecting planning and measurement with AI agents that spot in-flight opportunities: cut spend on one source of inventory, move it to another that delivers more efficiently.

Hudgins also said the quiet part: "you really need to work directly with publishers' log-level data." Those integrations, the article notes, take years.

So follow the loop. The agent recommends where the money goes, based on the inventory VideoAmp sees best. Then VideoAmp's measurement grades whether the move worked, with the same data. Inventory that's better observed looks like it performs better. It gets more money. The same system grades the result.

An AI that can't distinguish better performance from better visibility doesn't fix measurement. It automates the blind spot, faster.

CIMM's own ACR study found streaming apps contractually restricting measurement inside TV makers' environments. CIMM's Jon Watts says buyers should understand coverage, how unidentified activity is handled, and where the estimates depend on modeling. None of that is in VideoAmp's pitch.

The pitch does have takers. VideoAmp announced AI-powered reporting in June with Omnicom Media as its first adopter. One adopter is a press release. It isn't a comeback.

Frank Kowalczyk: We had a model once that always recommended the stations we had the cleanest data on. Those stations always won. Shocking. We gave the model an award. Little plexiglass thing. It's in the garage, next to the boat, holding up the end of the workbench that's shorter than the other end.

In June 2026, VideoAmp withdrew from its Media Rating Council audit, per the MRC. That audit is the outside check that tells buyers a measurement product does what it says it does.

In August, Unmeasured's first issue tested the number VideoAmp sells itself on: $6 billion in deal dollars. Built from the market up, using industry sizing for national linear TV and VideoAmp's share of the slice bought on advanced audiences, the math lands nearer $2 billion. A former VideoAmp employee described the $6 billion as an average total-day linear CPM multiplied by every publisher impression VideoAmp processed. That's throughput. It isn't money that changed hands on VideoAmp's numbers.

There is one real credential left. The Joint Industry Committee re-certified VideoAmp, along with Comscore and iSpot, in September. JIC certification means the networks and agencies on the committee consider the data usable for deals. It isn't MRC accreditation, and VideoAmp knows the difference, because it walked away from the second one.

Frank Kowalczyk: I sat in on a pre-audit once. Different company. Everybody wore the good shoes. The auditor asked for the one file nobody had, and the whole room looked at the intern. The intern looked at the window. You can f*** up a lot in this business. You cannot make an auditor un-ask a question. So you leave the room.

The investor mark is down 80%. The audit is gone. The pitch outruns the data.

Below the line: two cuts of a fifth of the staff, three people at the top in under three years, the people who keep writing to me from inside, a press operation that bounces, the 2024 sale talks VideoAmp denied, and why I'm not sure it makes it to January.

Enter your email to keep reading. It's free.