Hook rate is computed as 3-second video plays / impressions, expressed as a percentage. The 3-second threshold is the Meta convention; TikTok analysts often use 2 seconds against TikTok’s 2-second video view event. The metric isolates the opening — whether the first frame stopped the scroll, not whether the rest earned a click or a purchase.
What counts as a “3-second play” is platform- and version-specific. Meta has redefined its video-view events more than once (older 3-second view vs. newer ThruPlay-family definitions), and TikTok exposes both 2-second and 6-second view metrics. Two reports labelling the same number “hook rate” can be measuring different events — the threshold moves with the platform’s event definition, not just operator convention.
Hook rate sits at the top of the video-ad funnel. Thumb-stop ratio is a common synonym; on autoplay surfaces the calculation is the same. Hold rate (or watch-through rate) is the complementary half — of viewers who stopped, what share finished. CTR sits further down, measuring post-view click intent.
Benchmarks circulate in DTC commentary, but they vary so much by category, placement, audience, and platform-event definition that the metric’s reliable use is a within-account baseline for variant testing, not a corpus-wide target. Hook rate is the cheapest signal for whether the first frame is working, and the right cycle for iterating at the variant level before downstream metrics outrun creative fatigue. It does not determine whether an ad works on its own: a strong hook can drive zero purchases, and ROAS remains the binding read.