YouTube Measured by Search Lift, Not View-Through

Two readings of YouTube inside Google Ads survive a finance director's questions: a geo holdout over six to eight weeks, and a Search Lift study, which counts whether the people who saw the ad later searched for you by name. View-through conversions are not one of them. They are the number that keeps a video budget alive for one quarter and gets it cut in the next. If the plan is to justify this channel with view-throughs, this is the wrong listing to open. Campaign selection: It follows the job rather than the format menu. Video action campaigns where there is an existing audience to convert and a landing page able to take them. In-feed video where the buyer is researching and will watch three minutes of a lecturer if the first fifteen seconds earn it. Bumpers only as a reminder layer against people who already saw something longer. Sequencing where the addressable audience is small enough to reach repeatedly, which in Ireland happens more often than people expect. Audiences: Custom segments built from search terms and competitor URLs rather than affinity categories, because affinity is a demographic proxy and a search term is a stated intention. First-party lists for exclusion, current students and existing owners in particular, and for seeding. Placement targeting is used sparingly, only where a specific channel has a demonstrated audience overlap. A hand-picked list of forty channels is a smaller and more predictable buy than a topic. Assets, and what I will say about yours: Three lengths matter: a six-second bumper, a fifteen-second cut, and one long asset that deserves its watch time. In this practice's education accounts, the unpolished piece to camera from the person who actually teaches the course has consistently outperformed the produced film, and you will hear that before you commission the film rather than after. Briefs and cut reviews are included. Production is not. The lift test: Search Lift and Brand Lift each carry a Google-set minimum spend and a minimum study window, and both need the rest of the account stable. Nothing else launches and no budgets move on other campaigns while a study is open. That constraint is agreed in writing at the start, because the most common reason a study returns nothing readable is that somebody changed something else in week two. Reporting: The lift result with its methodology stated, the geo read placed beside it, and cost per incremental enquiry instead of cost per view. Where the two readings disagree, both are shown and the disagreement is explained rather than averaged away. Not included: Video production, editing or motion work. Channel management, uploads, thumbnails and organic strategy. Creator and influencer deals. YouTube bought through DV360 or an agency inventory deal. Any campaign whose success metric is views. Who this is not for: Advertisers who want cheap views. That is widely available and it is not this. Accounts with no existing search presence, where video creates interest that a competitor with better search coverage then collects; fix search first and come back. And anyone whose approval chain cannot leave a test undisturbed for six weeks, because the test is the deliverable here, and an interrupted one is worth nothing to either of us. I would rather decline the engagement than run a study we both know will be unreadable.