SSP vs. DSP Isn't the Right Fight

Two real shifts are landing on the same desk at once. In August, Taboola took over programmatic display ad sales for NBCUniversal's News Group, covering NBCNews.com and TODAY.com — the return of the sales house model, where a partner sells a publisher's inventory directly for a revenue share instead of the publisher running its own sales team. Ad tech veteran Andrew Kraft, who ran the equivalent playbook at AppNexus for Microsoft a decade ago, named it exactly: "the math stops being a philosophical question about 'core competency' and becomes arithmetic." Separately, programmatic curation, SSPs pre-filtering and packaging inventory with data and quality rules before a bid reaches a DSP, is booking measurable gains: shorter supply paths, earlier removal of made-for-advertising sites, and a larger share of budget reaching working media. Neither shift is vapor. And in the same window, The Trade Desk posted the worst quarter of its public life: Q2 2026 revenue grew just 3% year-over-year, down from the 20–26% growth rates it posted as recently as 2024, shares fell more than 20% in after-hours trading the moment results dropped, and the stock is down roughly 76% over the past year.

Neither shift actually removes the DSP's job, it just changes what arrives at its door. A sales house has to own no inventory of its own, or its reps face channel conflict and rationally favour their own supply — which means Taboola is now NBCU's commercial representative, not a replacement for reaching every other publisher on a media plan. Doing that publisher-by-publisher is exactly the manual process programmatic buying exists to replace. Curation works the same way: even the most advanced curation product on the market still ends its process producing a Deal ID "that can be activated in any DSP." That's not a limitation, it's the architecture. A curated Deal ID is a cleaner, pre-vetted version of the same unit a DSP has always transacted. Deciding which of hundreds of these deals across dozens of curators actually gets budget, in what sequence, against what frequency cap, measured against what independent standard, is still a demand-side job. Curation makes the inputs cleaner. It doesn't make the job disappear.

It's also fair to ask whether this debate is really about The Trade Desk specifically, and the honest answer is partly yes. CEO Jeff Green told investors "our revenue growth is below our expectations and below the standard we hold ourselves to," and pointed to genuine headwinds among TTD's largest legacy advertisers — but the deeper worry on Wall Street is a take rate that has sat within a point or two of 20% for a decade, which Green has so far declined to cut meaningfully even when asked directly. At the same time, Green is publicly attacking exactly the products a client is probably asking about: Amazon's DSP, which touts a zero-margin take rate, and Google's new Buyer Direct program, capping ad tech and data fees at roughly 10% while funneling deals through Google's own ad server. His argument, that both are built by companies which also own massive amounts of the media they'd be helping you buy, is structurally sound on its own terms, and his line that these approaches "look more like ad networks of 2006" is a fair description of what happens when the low-fee alternative also sells you the inventory. Both things are true simultaneously: the transparency argument holds up on its own logic, and Green is making it loudly because his business is under real, current pressure. A client shouldn't dismiss the argument because he's self-interested in making it, and shouldn't accept it uncritically because it happens to be correct — those are separate questions. Nothing in current reporting points to an SSP actually acquiring or absorbing DSP functions. What's happening is convergence from both directions: SSPs moving up-stack into sales representation and curation, DSPs defending ground that genuinely newer, cheaper buy-side products are eating at the edges. TTD's specific vulnerability, a battered stock and a take rate under open investor scrutiny, makes it a more exposed incumbent than it looked two years ago. That's a real story about one company. It isn't proof the DSP layer is becoming obsolete.

Separate the structural question from the market-noise question before it gets into a media plan. Structural question: does this plan need cross-publisher aggregation, independent frequency management, and measurement not controlled by the party selling the inventory? For any plan spanning more than a handful of publishers, the answer is still yes, curated Deal IDs and sales-house arrangements included, since both still need a demand-side layer to actually activate. Market-noise question, genuinely live and separate: is The Trade Desk specifically the right DSP at the right price, given what's happening to its growth rate and its take rate under investor pressure? If a client's "why do I need a DSP" is really "is TTD still the right one," say so directly, and price alternatives, don't let a fair question about one vendor get answered as if it were a question about the whole category.