Supply Path Optimization for Property and Marketplace Media

On open-exchange display the distance between what your DSP records as media cost and what the publisher records as revenue sits between 15 and 35 percent. Roughly a third of that pays a second or third intermediary reselling an impression the first one already held. This engagement measures that distance on your own account, path by path, and closes it by changing what your seat is permitted to bid into. Where we start: We read the sell side before the buy side. For every domain that took more than half a percent of your ninety-day spend we pull ads.txt and sellers.json, reconstruct the SupplyChain object hop by hop, and separate DIRECT declarations from RESELLER ones. On a property portal or a classifieds marketplace the same impression is commonly reachable through eight to twelve paths. One or two of them are the publisher's own. The rest are re-brokered, and the only reliable difference between them is who takes the margin. What we measure: Win rate, clearing CPM and post-bid outcome per seller ID, not per exchange. An SSP is not a unit of analysis; a seller ID inside an SSP is. Paths are compared on the same domain, format, geography and hour of day, because comparing a Magnite path against an Index path across different inventory tells you which one bought cheaper traffic and nothing else. Where a path wins more often and delivers worse measured outcomes, we say so and show the bid-level rows behind it. What you need to give us: Log-level data. The Trade Desk Raw Event Data Feed, DV360 exported to BigQuery through the Data Transfer service, or Xandr log-level feeds. Ninety days is enough, thirty is workable, and we will tell you which conclusions thirty days cannot support. If your seat is held by an agency that will not release the schain field, we will say on the first call that this cannot be done honestly, and end the conversation there rather than sell you a version of it. What we hand over: A path allowlist written into the seat itself, as inventory source settings in DV360 or supply vendor and public inventory controls in The Trade Desk, plus the seller and deal IDs to keep, the ones to cut, and the reason attached to each. The reasoning document matters more than the list. Paths change, and your team has to be able to re-run the decision in six months without calling us. What we do not do: We do not select fraud or IVT vendors, write brand safety keyword lists, or advise on attention metrics. We do not review creative. We do not stay to operate the seat once the paths are set. We will not accept a fee based on a share of media, here or anywhere else, because that fee moves in the opposite direction from this work. Where this is not worth buying: Below roughly forty thousand US dollars a month of open-exchange spend, the recovery does not pay for the engagement and the money is better spent on two direct deals. If ninety percent or more of your display already runs through programmatic guaranteed and preferred deals, your paths are already short and there is little here for us. Tell us that before you brief us and we will confirm it in an hour rather than in a report. Who this is not for: Buyers who want a supply audit as a negotiating instrument against an incumbent agency, with no intention of changing the buy. We have taken that engagement twice. Both times the document was filed and nothing moved. We now decline it.