An early obituary for The Trade Desk: how big tech broke media

At GroupM I built GSK's programmatic framework. The work only made sense if I could see how the impression cleared. Jeff Green sold that same job to every CMO who was tired of buying blind.

389 agency clients in 2015

In 2015 The Trade Desk did USD 113.8mn of revenue, up 156% on USD 44.5mn the year before, on USD 552.3mn of gross spend. 389 clients, almost all agencies. If you rolled those shops up to the holding company, Omnicom and WPP each accounted for more than 10% of the billings.

They listed on Nasdaq the next September. Priced at USD 18, opened at USD 28.75, closed at USD 30.10.

The product was a desk that did not own media. Agencies that were tired of Google's exchange could buy through someone who was supposed to show them the other side of the bid. That was the antitrust argument, turned into a company. It worked because the holding companies needed a desk they could point at when a client asked who was watching Google.

Agencies lost the terms

That company only works if agencies still set the terms of the buy.

On 2 September Judge Leonie Brinkema turned down the Department of Justice's breakup. AdX stays. DFP stays. Google Ad Manager is still one company. She took most of the conduct remedies and gave both sides 30 days to file a proposed final judgment. The independent desk was what you used while you waited for a breakup. The breakup did not come.

PepsiCo already skipped the pitch once the asset had no rate card. Michael Nathanson at MoffettNathanson wrote the quieter version for investors: agencies are going direct to media-owner sellers, and both sides have a reason to step over the middle.

The FTC says Amazon Ads told retail advertisers they were in a second-price auction and then charged nearly the full bid 80% of the time, more than USD 20bn since 2019. Amazon disputes it. You cannot do the GSK job on inventory that will not publish a clearing rule.

Wieser told AdExchanger where the share came from. The Trade Desk grew for years by taking it from other third-party desks. It did not take it from Amazon, Google or Meta. It did not grow programmatic's slice of agency spend. The company that listed as the answer to Google never took a garden's share.

Kokai missed, then Zuma walked it back

In 2023 they launched Kokai and billed it as the largest overhaul of the platform. Green later told investors they rolled it out slower than planned, and that some of that was deliberate. Q4 2024 was the first miss in eight years as a public company: USD 741mn against guidance of USD 756mn. The stock dropped 33% the next day. Green called it a series of execution missteps.

In August they shipped Kokai Zuma. Adweek confirmed the periodic table is no longer the default view. Three years after the overhaul, the signature interface is the one they are walking back. Samantha Ross, who markets the product, said buyers sitting in an ad group wanted to know why it was not delivering.

AdExchanger's 11 September file said The Trade Desk has not changed how it scours the web. Challenger desks call that bloat. The investor worry is Amazon, Google and Meta taking the budget Kokai was supposed to defend.

Almost USD 70bn to under USD 10bn

In December 2024 The Trade Desk was worth almost USD 70bn. On 11 September AdExchanger had them under USD 10bn. About seven times smaller, in 21 months.

They still trade on Nasdaq. S&P 500 drops them on 21 September, onto the SmallCap 600. Green announced a 15% layoff after cutting the rest-of-year forecast. Dan Salmon at New Street noted Green's super-majority, so a cheap stock does not even pick up a takeover bump. He decides whether the company is sold.

In 2020 Walmart left Xandr for The Trade Desk because TTD paid to see more of the bidstream. That was the product working.

On 10 September Amazon put ChatGPT ads into Amazon DSP, a US pilot, Delta Vacations in the first group. OpenAI keeps delivery. Amazon's team builds the campaigns. Digiday's fee line is the one I recognise from the buy side: Amazon often lands near zero on programmatic guaranteed, against The Trade Desk's historical 15 to 20%. Some of Amazon's growth, they wrote, has already come at The Trade Desk's expense.

OpenAI had talked to The Trade Desk earlier this year. I have sat in enough of those rooms to know what the paper looks like when the independent desk is supposed to win, and they still signed Amazon.