Live sponsorship with frequency capped at the stitcher

A frequency cap set inside a DSP does not survive server-side ad insertion. The stitcher decides what a household actually sees, and across a three-hour live game a cap of three per day can deliver eight or more identical spots to one television, because each break is filled by a separate auction and no household identifier is shared between them. Before I place a live buy, the cap goes into the insertion order at the SSAI layer, and it gets checked against post-log data by household. Where a seller will not enforce it there, I do not buy the property, whatever the audience number looks like. What gets negotiated, and it is not the CPM: Pod position and break structure. Break-length mismatch, where a 150-second break holds ninety seconds of sold inventory and the stitcher fills the rest with slate or house promos, which is a different viewing experience than the one you were sold. Competitive separation with a named list rather than a category. And the make-good language, which is the clause I read first. Make-goods belong in live inventory: A make-good delivered in on-demand catch-up is a different product with a different audience watching under different conditions. That gets written into the order before signature, with a named window and comparable audience, or the risk stays with you and nobody says so out loud until the game is cancelled. Canada is not one market and not one feed: Simultaneous substitution was removed from the Super Bowl by the CRTC for the 2017 to 2019 broadcasts; the Supreme Court found in 2019 that the Commission had exceeded its authority, and simsub returned in 2020. Rights are fragmented in the same spirit: Sportsnet, TSN, Amazon Prime Video carrying Monday Night Hockey in Canada since the 2024-25 season, with RDS and TVA Sports on the French side. Buying the game does not buy the audience. The French feed is a separate negotiation with its own break structure and its own creative, and treating it as translated overflow of the English plan is how a brand ends up sounding imported to a third of the country. Loudness inside a stitched break: Your spot is delivered to ATSC A/85 at minus 24 LKFS and then stitched next to programming and promos mixed to something else. I check the stitched output on an actual device before the flight goes wide, not after a viewer complains, because loudness jumps in live are the fastest way to make a brand unwelcome in somebody's living room. Twitch and creator live: Mid-roll bought as a schedule against a named category, with the cap enforced, yes. Creator-read only where the script arrives before air and a make-good clause survives a bad stream. I do not buy a logo bug on a stream overlay with no handoff and no measurement, at any price, and I will not build a reporting narrative around one. Excluded from the retainer: Talent negotiation and contracting for creator sponsorships, on-site event activation, and clearance for reusing broadcast footage afterwards. Those are legal and production workstreams with their own specialists, and I will hand you to people who do them properly. Do not hire me for this if: Your approval chain cannot swap creative inside 24 hours. Games get cancelled, weather moves schedules, and a spot referencing a result that went the other way is a bad evening for everybody. Also skip this if the plan is to buy live purely as reach at a lower cost than television. That trade exists, it is real, and it does not need me.