Video Ad Sequences Built Where Identity Holds

A sequence loses most of its audience before the last step. Where each step retains forty to fifty-five percent of the one before it, which is the range we see on signed-in YouTube inventory, step five reaches a single-digit share of the people who saw step one and anything concluded from it is noise. We build three or four steps, sized so the final one still lands on an audience large enough to read, and we run them only where the same person can be recognised across days. Where sequencing quietly fails: Sequencing depends on identity, and identity is exactly what connected television does not reliably provide. A shared household profile, a signed-out session, a move from phone to laptop: each one delivers step three to somebody who never saw step one, and a step-three message that assumes its own premise is worse than no sequence at all. We measure step-completion rates by environment and drop the ones below the threshold set at the start, rather than letting them dilute the read. In practice most sequences we build run on signed-in mobile and desktop, with the television screen carrying a single standalone message instead. The order is the argument: Step one describes the situation the viewer is already in and never names the advertiser first. Step two puts one number on the cost of doing nothing: the annual difference between a standing tariff and a fixed one, the fee that rises after the application deadline, the repeat callout charge. Step three is a specific action with a date attached. Where a fourth step exists it is triggered by a skip rather than a view, because a skip is information. The shorter, blunter version earns more from someone who has already refused the long one than a repeat of the ad they refused. Not one film cut three ways: Three lengths of the same edit is not a sequence, it is a frequency plan with extra reporting. Each step is written to be understood by someone who missed the previous one and to reward someone who did not. Six-second bumpers work as a closing step and almost never as an opening one. Frequency, reset and re-entry: We set the campaign cap and the sequence reset window deliberately at the start rather than accepting defaults, and block re-entry inside a flight so completers are not returned to step one. Every one of those settings goes into the plan document with the reason attached, because whoever inherits this account in a year needs to know why it was set that way. How we test whether the order earned anything: The control here is not a dark market. It is the same audience receiving one strong message at matched frequency and matched spend. The question is narrow: did the order add anything over a single good advertisement shown three times? Often it does not. We have told clients exactly that and moved the budget, and you should expect the same answer if the data gives it. What we do not take on: Products with a same-day purchase cycle, where a sequence has no room to run. Plans that are television-only, for the identity reasons above. Storyboards already approved and locked, since the ordering decisions are the work and arriving after them leaves us grading somebody else's homework. Production, casting and edit: we brief and review, we do not shoot. What you receive: A sequence map with the trigger logic on every branch, per-step drop-off by environment read weekly, and the raw exports underneath it. At the end, a written recommendation on whether sequencing belongs in your plan at all.