Shoppable CTV Units Measured by Site-Visit Lift

QR scan rates on shoppable CTV run between 0.1 and 0.5 percent of impressions in every campaign we have measured. That is the honest ceiling, and anyone quoting you two percent is counting something other than scans. So we do not sell the unit on scans. The unit exists to shorten the distance between seeing and searching, and the measure that survives scrutiny is a holdout read on branded search volume and direct site visits from the treated geographies. Device testing before the build is commissioned: A large number of CTV publishers will accept a VAST tag, ignore the interactive payload and render the linear fallback, and their reporting will not tell you this happened. You pay a build fee and serve a standard spot. So we test-render on hardware first: Roku, Fire TV Stick, a Samsung Tizen set, an LG webOS set and Sky Stream at minimum, on the exact tag, through the exact supply path. VAST 4.2 with a SIMID payload and OMID for measurement is the target; where a publisher is still on a VAST 2.0 wrapper we treat the placement as linear-only and tell you to buy a plain spot and keep the build money. Roughly a third of the allowlists we are handed cannot carry interaction at the scale that would justify a build, and saying so early is most of the value in this engagement. What we actually build: Roku Action Ads where Roku is a real share of the plan, Amazon's interactive formats on Fire TV and Prime Video, Innovid or KERV where the publisher accepts a third-party build. Design decisions get made against ten-foot viewing conditions rather than a laptop: a QR code below about nine percent of screen height does not scan reliably from a sofa, and it needs to hold on screen for six seconds or more before the average person has found their phone. A short spoken and written vanity URL runs alongside it for the majority who will never scan anything. Remote-navigable units get built only when the interaction survives being two button presses deep, because almost nobody goes further. The landing experience is part of the media buy: A scan puts somebody on your site one-handed, standing up, with a television still playing. If that lands on a generic homepage, the unit will underperform and the report will blame the creative. We specify the destination, and we will decline to build if nobody on your side owns getting that page changed. What we refuse: We do not accept scan rate as the campaign KPI, in the brief or in the report. We do not run interactive without a household frequency cap across the entire CTV buy, since the same interaction shown six times is an irritation with a measurable cost. We do not build for inventory we have not test-rendered on. And we do not run pause-ad formats as a stand-alone line, because their measurement story is currently weaker than their sales deck. Not for you if: You have no ecommerce or lead capture worth sending someone to. Your CTV spend is under about two hundred thousand euros in the flight, below which a lift read on branded search cannot be separated from noise. Or you want the interactive unit primarily so it can be shown internally, which is a real reason brands buy these and not one we will take money for.