Short version: YouTube advertising splits cleanly into two jobs, making the video and buying the media, and they are separate disciplines that fail for different reasons. We run the second: campaign type selection, audience targeting, bidding, exclusions, and the measurement work that tells you whether video actually contributed revenue or merely collected view-through credit for sales you were going to get anyway. Brands supply the creative, whether that is a proper production, existing product footage, or content they already have. This guide covers what the media buying side actually controls, why measurement is the hardest and most important part of YouTube, and how to judge whether video belongs in your account at all.
The division of labour matters more on YouTube than on any other Google channel. On Search, targeting is most of the battle. On YouTube, a brilliant media buy behind a weak video loses money, and strong creative pointed at the wrong audience with default settings loses money too. Being clear about which half you are buying, and from whom, prevents most of the disappointment brands report with the channel.
What YouTube ads are actually good at
YouTube sits at a different point in the funnel from Search, and expecting Search economics from it is the fastest route to concluding it does not work. Nobody on YouTube is searching to buy right now; they are watching something else when your ad arrives. That makes it demand generation rather than demand capture.
Where it genuinely earns its place: reaching people who have never heard of you at a cost per view that no other Google surface matches, remarketing to site visitors and past customers with something more persuasive than a banner, explaining products that need explaining, and feeding the top of a funnel that Search then converts. For considered purchases, higher ticket items, and anything where a demonstration does the selling, video moves numbers that text never will.
Where it disappoints: brands expecting last-click ROAS comparable to branded Search, accounts with small budgets that cannot sustain both prospecting and capture, and products whose appeal is entirely price-driven, where a video adds nothing a Shopping listing does not already say.
What the media buying side actually controls
Campaign type and format
The format decision comes first because it dictates cost structure and intent. Skippable in-stream is the workhorse for most direct response, since you generally pay only when someone watches meaningfully or clicks, which filters out disinterest automatically. In-feed placements reach people browsing YouTube with some intent to explore. Shorts inventory behaves differently again, with lower attention and lower costs. Non-skippable and bumper formats buy guaranteed impressions and suit awareness objectives rather than performance ones. Choosing the wrong format is how brands end up paying for views from people who would have skipped if they could.
Audience targeting and, more importantly, exclusions
YouTube offers a wide targeting toolkit: your own remarketing and customer lists, custom segments built from search behaviour and competitor interests, in-market and life-event audiences, plus placement and topic controls. The lists you own almost always outperform the ones Google supplies, so remarketing and customer-match usually earn the first budget.
Exclusions matter as much as targeting and get far less attention. Without deliberate exclusions, video budget drifts into children's content, made-for-advertising channels, and low-quality placements that spend real money on worthless attention. Regular placement reviews are to YouTube what search term reviews are to Search, and an agency that never shows you a placement exclusion list is not managing the channel.
Bidding and budget structure
Bidding strategy follows the objective: cost per view for reach, target CPA or conversion-focused bidding for direct response, with enough budget behind each campaign that the system can actually learn. Splitting a modest budget across many video campaigns is a reliable way to make all of them fail slowly.
Measurement, which is the real job
This is where most YouTube accounts are either won or quietly deceived, and it deserves its own section.
The view-through attribution trap
YouTube reports view-through conversions: someone saw your ad, did not click, and later converted. Counted generously, this makes video look spectacular, because a large share of those people were going to buy anyway. An agency reporting YouTube ROAS that includes unqualified view-through credit is showing you a number that flatters the channel and cannot be trusted for budget decisions.
Honest YouTube measurement takes a different shape. Separate view-through from click-through conversions in reporting rather than blending them. Watch what happens to branded search volume and direct traffic when video spends, because genuine awareness effects show up there. Where budget justifies it, run a geographic or audience holdout so you can compare like with like, which is the only reliable way to see incremental effect. And judge the account on whether total business revenue moved, not on whether the YouTube column looks impressive in isolation.
This is uncomfortable work because it frequently reveals that video contributed less than the platform claims. It is also the difference between scaling a channel that works and pouring budget into a reporting artefact.