DSPs & buy-side platforms

Outcome in this segment means a downstream business result attributed to a campaign the buyer ran — conversions, sales, revenue, store visits — plus hours saved inside the buying team: 153 claims across 41 pages, all 10 companies read.

50 claims are outcome-bearing — 28 state a metric, 20 imply one, 2 are guaranteed — across four constructs: attributed conversion or sale (9 companies), cost per acquired outcome (5), physical-world action (4) and buying-team productivity (3).

Evidence is thinner at the point of claim than in the walled gardens: 19 of the 50 carry no evidence on the page, 16 a named client, 11 a bare number, 3 a method, 1 a third party, and 24 of 38 pillars hold only on a different page.

One line states how a number was arrived at — “Leveraging Conversion Lift reporting, available directly within the Viant DSP, the global pharmaceutical company saw a 104% increase in site visits from target audiences who visited the site after exposure to the CTV ad.” — and the same site names ghost bidding for control groups on its DSP page.

The recurring gap is the home-page multiplier: “4x incremental reach” (madhive.com) asserts a causal construct from a metric tile with no baseline, population or period, a condition the capture record also flags.

Both guarantees cover fraud, not results: “The Fraud Free Guarantee” (madhive.com) and DV360's refund for impressions determined to be fraudulent; no page attaches a make-good to a conversion, sale or visit figure.

Price is absent as evidence: 0 of 41 pages state a rate, 1 of 10 companies has a pricing page, and it names the mechanism — “Most DSPs charge a percentage of media spend as a platform fee.” — without a number.

Open ground: a published fee beside an outcome an outside party measured, with a make-good attached; Viant holds the method, Basis the outside study, Yahoo the certifier and Madhive the guarantee, but no page holds two of them at once.