“More valuable decisions are moving toward the supply side.” That single sentence is the thesis of Magnite’s quarter, and it is worth more attention from publishers than the earnings numbers that came with it.
CEO Michael Barrett is pushing the SSP into decisioning work that has historically belonged to DSPs — packaging inventory, audience enablement, optimisation — with SpringServe as the vehicle. He is careful about the framing: “we, in no way, shape or form are trying to replace the DSP.” The claim is narrower and, for publishers, more interesting: the choices that determine which impression wins, at what price, with which data attached, are migrating from the buy side of the exchange to the sell side.
The numbers give the thesis credibility. Q2 revenue came in at $193 million, up 11% year over year; contribution ex-TAC was $190 million, up 17%; and CTV contribution ex-TAC hit $97 million, up 36%. In a week when The Trade Desk grew 3% and its stock fell 20%, an SSP growing contribution at 17% and CTV at 36% is not just outperforming — it is making an argument about where the industry’s margin is moving.
The numbers in this piece
01Whose decisioning layer is it?
Strip away the earnings-call language and the mechanics are familiar to anyone who has run a wrapper: curation, deal packaging, audience matching and supply-path optimisation are all decisions, and every decision in the chain has an economic owner. Barrett’s thesis is that those decisions increasingly get made where the inventory and the first-party data live — the sell side. He is probably right. The question for a publisher is not whether decisioning moves sell-side; it is whether “sell-side” means you, or means your SSP.
That distinction is the whole game. When your SSP packages your inventory into a curated deal, matches an advertiser’s audience against segments built from your first-party data, and optimises delivery across your supply — that is real value creation, and it will be paid for. The publisher-favourable version is one where you drive the packaging, own the resulting segments, and see the fee for every layer. The publisher-unfavourable version is one where a curation and optimisation stack you never asked for quietly widens the spread between what the buyer pays and what you clear — the same take-rate opacity the industry spent a decade fighting, rebuilt one layer down.
There is one genuinely good feature of Magnite’s version of this, and it is worth naming: non-exclusivity. Sell-side decisioning that works across many DSPs strengthens the publisher’s position in a way that single-buyer lock-in never does. A decisioning layer that routes to every buyer is leverage; a decisioning layer that routes to one is dependency. When Google’s Buyer Direct is simultaneously offering to collapse the chain inside the ad server, an SSP-operated layer that stays buyer-agnostic is the version of consolidation publishers should prefer — provided the fees stay visible.
02CTV is where this gets proven first
The 36% CTV growth is the tell about where the contest is being fought. CTV is where deal IDs, curation and audience packaging already dominate transaction structure, where content signals are scarce and valuable, and where SSPs are competing hardest to demonstrate that their decisioning adds margin. That competition is a window: SSPs need publisher supply to prove the thesis on, which means CTV and streaming publishers currently have unusual leverage to set the terms — data ownership, fee disclosure, packaging control — under which that proof happens.
Barrett's thesis is probably correct: decisioning is moving sell-side, because that is where the inventory, the context and the first-party data live.
03Why this matters for publishers
| Value migration is fee migration | "More valuable decisions moving to the supply side" has historically meant more fees accruing there too. If you cannot itemise what your SSP charges per deal type — open market, PMP, curated, audience-enabled — you cannot know which side of this migration you are on. |
|---|---|
| Your first-party data is the fuel | Audience enablement at the SSP layer runs on publisher data. Who owns the segment built from your data, who can resell it, and what you are paid when it transacts are contract questions, and right now they are being answered by default rather than negotiation. |
| Non-exclusive beats exclusive | An SSP decisioning layer that reaches all buyers strengthens you; proprietary paths that reach one buyer trade short-term yield for long-term dependency. Magnite's framing is the better pattern — hold every partner to it. |
| CTV publishers have a leverage window | SSPs need premium supply to prove sell-side decisioning works. Terms offered while they are competing for that proof will be better than terms offered after the layer is entrenched. |
04What publishers should do
05The bottom line
Barrett’s thesis is probably correct: decisioning is moving sell-side, because that is where the inventory, the context and the first-party data live. But “the supply side” is not one place — it is a stack, and publishers sit at the bottom of it. The difference between this being the good version of consolidation or just a new fee layer with better branding will be decided in unglamorous places: QBR agendas, fee schedules, and data-ownership clauses. Publishers who negotiate those now, while SSPs still need them to prove the model, will own a piece of the decisioning layer. The rest will rent it back.