Supply Path Optimization Down to a Named Seller Shortlist

Ask how many distinct paths your DSP found to a single impression on a single publisher. On open-exchange display in this region the answer is usually between five and twelve. Each of those paths costs you a bid, and the winner is decided on a price net of fees you cannot see rather than on the quality of what is being sold. The ISBA and PwC supply chain study left 15 percent of advertiser spend in an unattributable delta, money that left the advertiser and could not be traced to a publisher or to a named fee. Shrinking that is what this work is for, and it shrinks by naming sellers, not by haggling over take rates. How the analysis runs: Bid request logs, keyed on domain, ad unit, user and a one-second timestamp bucket, produce the duplication map: which sellers offered the same impression, at what price, through which chain. sellers.json resolves each seller identifier to publisher, intermediary or confidential, and the publisher's own ads.txt states what it declares as direct. Where the two disagree, the path is either misdeclared or unauthorised, and both are grounds to cut before any fee conversation begins. On server-side paths we check whether the request reached the seller through the publisher's own header bidding stack or through a wrapper the publisher does not control. What we will not do: We hold no rebate, referral arrangement, curation revenue share or seat resale with any seller, curator or platform. When a shop recommends a shortlist and also earns on a name inside it, that shortlist is a media plan for their own accounts. We do not cut paths on fee percentage alone: the cheapest route to a publisher is frequently the one with the least inventory control and the most reselling behind it, and two points of take rate is a poor trade for a thirty percent made-for-advertising rate. We do not build or operate a curated marketplace for you. The deliverable is your configuration, in your seat, under your contracts. The decision handed to you: A named shortlist per market. Which sellers to keep at full weight, which to keep for a specific publisher set only, which to throttle, which to remove. For every removal, the reach and unique-publisher coverage it costs, so the decision stays yours with the price visible. Where a path survives only because it is the sole route to a publisher you want, we say so and move it into a direct deal conversation instead of pretending the path is good. Implementation: Changes land in DV360, Xandr Invest, Beeswax or The Trade Desk as supply source settings, inventory lists and deal allocations, phased across three to four weeks so each cut gets a clean read against reach, win rate and clearing price. Duplication is then re-measured on the same join, so before and after are comparable rather than rhetorical. Who should not book this: Advertisers running under roughly 400,000 dollars a quarter through open exchange, where recovered fees will not clear the engagement. Teams inside a holding company principal buying arrangement, where the path decision has already been made for you and no amount of analysis reopens it. And advertisers who want a take-rate figure as negotiating material but have no intention of changing the buy. We will produce it, you will not act on it, and we will both have spent a quarter to confirm something you already suspected.