Private marketplace deals that actually bid

A deal that will not spend has failed in one of five ways, and an hour with the logs tells you which. The deal was trafficked against a seat ID the SSP has not authorised. The publisher's floor sits above your line's maximum bid. The deal's own targeting and your line's targeting intersect to nothing. Your identity match rate against that publisher is too low to bid on the traffic that arrives. Or the deal was created as a private auction while the buyer planned around guaranteed delivery. We start every engagement here, on your existing deals, before we sell you new ones. Setting deals up so they do not fail: One deal ID per publisher, placement and format, named to a convention that still parses in six months. The deal type chosen for the job rather than for convenience: programmatic guaranteed where you need the impressions and have committed the money, preferred deal where the price is agreed and delivery is not, private auction where you want first look and nothing more. Floors negotiated in the publisher's own currency and in their own language, which for Japanese publishers means Japanese and means several meetings rather than a rate card by email. Japan: Domestic supply moves through Ad Generation, fluct, GENIEE, MicroAd and OpenX, and the publishers worth the negotiation, the national dailies, the recipe and parenting properties, the video services, do not have their inventory represented well in the global tools. Deals here are relationships and they are conducted in Japanese, on Japanese terms, with a schedule that expects patience. We do that part. It is most of why foreign advertisers hire us. Curation: We read the domain list before signing. A curated package that will not show you what is inside it is open exchange with a marketing name and a margin, and the answer is no. Where a curator will show the list, we check it against sellers.json, ads.txt and the schain object, and we ask which of those domains the curator has a direct relationship with and which they resell. That question ends a surprising number of conversations. After launch: A deal health monitor runs daily and compares bid requests received, bids submitted, bids won and impressions delivered per deal. A deal that receives requests and never bids is a targeting or match rate problem. A deal that bids and never wins is a floor problem. A deal that receives nothing is a trafficking problem. Naming which one it is on day two rather than in the month-end review is the whole value of watching it. Not included: DSP seat management, which is a separate engagement and priced separately. Creative. Ad server trafficking on the publisher side. Contract and payment terms with publishers, which stay between you and them; we advise on floors and structure but we are not a party to the deal and we take nothing from the media. Not for you if: You want a stack of deal IDs delivered as a volume of options, which is how deal sprawl starts and how nobody ends up accountable for any one line. You want us to accept a curated list unseen. You expect Japanese publishers to negotiate at the pace of an email thread. Or your buying team cannot access delivery logs at deal level, in which case nobody can tell you why a deal is quiet and you will be paying us to guess alongside you.