YouTube Shorts Buying Inside Demand Gen, Measured by Lift

Shorts inventory is bought in Google Ads through Demand Gen and video view campaigns; there is no separate Shorts campaign type to ask for. The number that decides the outcome is retention at second three. Below roughly thirty percent of viewers still present at that point, no bid strategy, audience or budget rescues the asset, and the correct action is to replace the creative rather than restructure the account. We measure that threshold on every asset before it gets budget. The consequence is that this engagement is mostly a creative testing programme with a media system attached, and it is priced that way. The build: Demand Gen campaigns with channel controls set deliberately rather than left open, so Shorts placements can be isolated and read separately from in-feed and Discover. Product feeds attached where the catalogue justifies it. Video view campaigns used where the objective is genuinely reach and stated as such, not disguised as performance. Audience signals built from first-party lists rather than affinity segments, which flatter the report and move nothing. The creative test: Four propositions, each shot vertically at 9:16, each legible with sound off, each with burned-in subtitles that match spoken French rather than an automatic translation of an English script. We do not cut a thirty-second television asset into vertical and call it Shorts creative; the pacing is wrong at the only moment that matters. Assets are compared on second-three retention and completed views before any conversion data is trusted, because conversion volume at this stage is too thin to rank creative reliably. The measurement: A geo holdout, or a Google conversion lift study where the account qualifies, decides whether the channel is producing sales or recording them. View-through conversions are reported for completeness and excluded from the decision. For automotive, where the conversion happens in a dealership weeks later, we import offline conversions against GCLID so that a test drive booked in Lyon in March can be traced to a Short watched in February, and we tell you the size of the match rate rather than quietly assuming it is high. What we will not do: We will not run a lift study on a budget that cannot detect an effect. If your spend is below the level where the test is readable, we will say so, run the geo holdout instead, and describe the result as directional. We will not report view-through conversions as sales, and we will not accept a target CPA carried over from search, because the two are not measuring the same thing and matching them is how good prospecting gets switched off. Not for: Brands with a single hero film and no appetite to produce more. The programme depends on replacing losing assets within a fortnight, and a client who needs six weeks and three committees to approve a cut will pay for inventory that is already spent. Also not for anyone who wants a channel launched and left running. Shorts creative decays faster than any other format we buy; an unattended account is a declining account within two months. Not included: Production, casting, music licensing and dubbing. We write shot-level briefs, review edits and reject the ones that fail the retention threshold, and we keep production outside our invoice so that rejection stays honest. YouTube channel strategy, organic publishing and creator agreements are separate work. How it runs: One senior partner owns the programme and presents to your marketing director monthly. Google Ads sits on your billing. The retention threshold, the holdout design and the decision date are agreed at kickoff and written into the scope, so the conclusion is not negotiable after the fact by whoever is least happy with it.