Angle and Hook Development for Paid Social and YouTube

We will not send you a hundred hooks. An engagement produces a small set of angles, usually five or six, and for each the executions that actually test it, plus the number at which we stop testing it. Volume is what a shop sells when it has not decided which claim is true. The distinction we work from: An angle is a claim about why somebody buys. A hook is the first two seconds of an execution arguing that claim. Most accounts that feel starved of creative are not short of hooks — they have one angle and forty ways of saying it, which is why every test lands in the same place. Our first deliverable is usually the list of angles the account has never run. How a hook gets judged: Hook rate, three-second plays over impressions, tells you the thumb stopped. On its own it is a trap. A hook that lifts three-second rate while hold at fifteen seconds falls has lied about the ad behind it, and it will buy cheap video views and no conversions. We read the pair. We kill hooks that win the first number and lose the second, including ours, and we do not argue for them. Per platform, because they are not the same job: On skippable in-stream you are not charged for the first five seconds unless someone stays or interacts, so the hook there is not built to prevent the skip. It is built so the skip costs you nothing: product and reason in frame and audible before the button matters. On Meta feed, captions are burned in as standard and the framing is 4:5 or 9:16 with the safe zones respected — not the twenty per cent text rule, which Meta dropped in 2021 and half the industry still designs around. On TikTok the hook is more often the first audio beat than the first image, and a silent-safe cut still gets made. Kids and family, specifically: Platform targeting for under-eighteens is restricted to a short list of signals, so in practice the hook argues with the parent while the product is used by the child. Hooks that speak to a child in a feed the child is not in are the most reliably wasted spend in that category, and we do not write them. What you receive: The angle set, each with the buying claim it makes and the evidence behind it. The executions per angle, written to be shot, with the opening two seconds specified frame by frame rather than described. A kill criterion per angle, stated before the test runs. And the running order, because testing six angles simultaneously on a modest budget returns noise and a meeting about noise. Not included: Full scripts beyond the opening, editing, media buying, and building the test itself. If you need the complete scripts, that is separate work and we will say so rather than pad this one. Not for you if: You need a monthly quota of hooks to fill a content calendar. Your budget cannot separate variants cleanly, in which case the result will not be readable and the honest recommendation is to spend the money on one good execution instead. Or you want a competitor's hook reproduced — we will explain why theirs works and then decline, because you do not know what it costs them to run and neither do we.