YouTube Campaigns in Google Ads, Built Around a Holdout

A YouTube campaign inside Google Ads will not tell you whether it worked. The platform counts a view-through conversion after a ten-second view on a three-day window, and that count moves in your favour whether or not the video changed anything. So this engagement starts with a geo holdout: fifteen to twenty-five percent of your addressable regions kept out of the buy for the whole flight, matched to the exposed regions on the prior quarter's conversion volume, and read at the end as a difference between two groups rather than as a number the seller produced. If you are not willing to withhold spend from a region, stop reading, because the rest of this does not work without it. Which campaign type, and why the answer matters: Video Action Campaigns no longer exist as their own type; that inventory now sits inside Demand Gen and behaves like a bidded response channel rather than like video. I keep the two jobs apart. Demand Gen carries the response budget, with product feeds and image assets sitting next to the video. Video Reach Campaigns with a target frequency setting carry the frequency job. Video View Campaigns sit between them and I use them rarely, because a paid view is a metric the platform sells rather than one you need. Which of the three you get depends on whether your problem is that nobody has heard of you or that people who have heard of you are not converting. Those are different problems and I will tell you which one I think you have before any budget moves. Audiences built out of your own account: The segments that work here are custom segments seeded with the search terms already converting in your Search campaigns, not the affinity and in-market categories in the picker. For a machine tool distributor that meant a segment built from part-number and tolerance queries. For a mobile studio it meant competitor title searches and store page URLs. Customer Match from closed-won CRM records goes in as an exclusion, so you stop paying to reach the people who already bought. Creative: I do not produce the film. I do write the brief, and I will send a cut back if the product is absent from the first frame, if the brand arrives after five seconds, or if there is no reason to still be watching at second six. Where the flight is long enough, two cuts go into a Video Experiment rather than into an argument about which one is better. Placements and suitability: Every account gets an account-level placement exclusion list, reviewed against the placement report fortnightly, and content suitability set to the limited inventory tier for anything industrial or recruitment-facing. Google Ads does not offer pre-bid third-party verification the way DV360 does. If contracted IAS or DoubleVerify pre-bid segments are a condition of your media policy, this is the wrong buying platform, and you will hear that on the first call rather than at the invoice. Not included: Video production, voiceover, translation, subtitling. Brand lift studies, whose spend floor sits above the budgets most of these accounts run and which answer a weaker question than the holdout does. Buying the same inventory through DV360. Creator sponsorships on YouTube, which are a different negotiation and not one I take. Not for you if: Your flight runs shorter than four weeks, which is not long enough to read a holdout. You want a view count or a cost per view as the reporting outcome. Your conversion event fires on a page load rather than on anything a salesperson would recognise. Or your YouTube budget sits under roughly ten thousand euros a month, where the reach is too thin for the measurement to separate from noise and you would be paying me to confirm that.