Shoppable CTV built around the phone, not the remote

Interactive connected TV runs on SIMID inside VAST 4.2. VPAID never ran on a television, so a vendor quoting VPAID for CTV is quoting mobile web, and a unit built that way serves as a plain linear ad on every device that matters to you. The second thing worth settling before budget moves: the interaction that performs is the one that finishes on a phone. A remote is a directional pad, and asking a household to arrow through a product carousel from eight feet away is asking them to do data entry with a stick. Every unit I build treats the QR handoff as the primary action and on-screen navigation as the fallback. What the code points at: A campaign-unique short URL, redirecting with source parameters you set, landing on a page that is dated and priced. For travel that means a fare with the dates already applied, not a destination hub. For retail it means one product with the size selector above the fold. A homepage destination loses the intent inside two taps, and no amount of on-screen design rescues it. If nobody on your side can ship that page before flight, the build waits. The fallback carries the offer: A real share of connected TV delivery strips interactivity somewhere along the supply path. The linear creative underneath therefore has to stand alone and make the same offer, with the code on screen long enough to be read from a sofa. Units that put the whole proposition inside the interactive layer are units that go dark for a chunk of the buy without anyone noticing. Supply path, confirmed before the build is scoped: Roku, Amazon and Samsung Ads behave differently, and hosting through Innovid or BrightLine does not guarantee the path passes SIMID intact. Vendor documentation is optimistic on this point. I confirm with a test impression on the path you are actually buying, and I scope the build against that answer rather than against a capability matrix. How it gets measured: Unique scans counted against ad server delivered impressions, not against vendor-reported interactions. Then scan to session, session to the thing you sell. Because the scan is a request to a URL you control, logged by your own infrastructure, you can audit every number in this campaign without asking me or the vendor for permission. Household sequencing off a television identifier is available and it is profiling: Quebec's Law 25 has required consent for that since September 2023, so it gets built where your counsel has documented the basis and not on an assumption. Where the money stops: I do not build your landing page, your catalogue feed or your payment flow. I do not report engagement rate, dwell time or interaction rate, because a household that opened an overlay and left is not a result. I will not build an interactive layer for a supply path that strips it, even where the vendor will happily invoice for one. Not the right buyer: Logistics. I work in freight and nobody scans a code from the couch to start a quote for a lane; the buying committee is at a desk on a Tuesday and the money belongs in in-stream. Also wrong for catalogues that need filtering — sizes, variants, availability by store. Television is a poor discovery surface, and a shoppable unit that opens a browsable catalogue is a worse version of the search ad you already run.