The UGC & AI video glossary.

AI UGC is user-generated-content-style video created with AI avatars and tools instead of filming a human creator. It keeps the authentic, hand-held, "made by a person" feel of UGC while letting brands produce and test dozens of variations in the time one traditional creator delivers a single clip.

A UGC (user-generated content) video is short-form, creator-style footage that looks organic rather than like a polished ad. The format that performs is typically vertical, under ~15 seconds, script-led, and shows the product in the first few seconds.

A hook is the first 1–3 seconds of a video — the part that decides whether someone keeps watching or scrolls. It's the single biggest lever in short-form ad performance. Strong hooks are specific and natural; showing a human face in the first 3 seconds can lift performance dramatically.

Hook rate is the percentage of impressions that watch past the first 3 seconds (3-second views ÷ impressions). It's a fast, platform-agnostic predictor of whether a creative will scale. A healthy benchmark for short-form ads is roughly a 30%+ 3-second hook rate.

Retention is how much of a video viewers actually watch — the curve from the hook to the end. Platforms reward high watch-time with cheaper distribution, so retention is both a quality signal and a cost lever. Disciplined teams structurally test edits to hold attention.

ROAS (return on ad spend) is revenue generated per unit of ad spend — e.g. 4× ROAS means $4 back for every $1 spent. In creative-led accounts, ROAS is downstream of hook rate and retention: cheaper, more-watched creative lowers CPMs and lifts ROAS.

Creative testing is the disciplined process of launching multiple ad variations to find winners, then iterating on them. A common system is a 3×2×2 matrix (3 hooks × 2 lengths × 2 CTAs = 12 variants); a practical pipeline keeps 5–10 fresh creatives live and refreshes every 7–14 days.

A pattern interrupt is an unexpected visual or audio moment that breaks the scroll and resets attention — a jump cut, a bold on-screen claim, a surprising motion. Used in the hook (and to re-hook mid-video), it lifts both 3-second view rate and retention.

B-roll is supplementary footage layered under a voiceover or talking head — product shots, hands using the product, lifestyle clips. In UGC it keeps the eyes busy and reinforces the script, which protects retention. The trick is matching visuals to what is being said.

A slideshow ad is a short-form post built from a sequence of static frames (often an educational listicle) rather than filmed video. It doesn't feel like an ad, is cheap to produce at volume, and is a strong complement to talking-head UGC for top-of-funnel.

A digital twin is a consistent AI character — a fixed face and voice — that represents a brand across every video. Because it stays identical clip to clip, it builds the recognisable, repeat-creator effect that random stock avatars cannot.

Content on autopilot is an automated loop where an AI agent researches a niche, writes scripts, produces character-led videos and publishes them on a schedule — with a human approving or letting it run fully autonomously. It collapses the generate → edit → deploy loop to near-zero.

AEO is structuring content so AI answer engines — ChatGPT, Perplexity, Gemini, Google AI Overviews — can understand, trust and cite it. Its goal is mentions and citations in AI answers, not just blue-link rankings. With ChatGPT at ~800M weekly users and zero-click searches rising, it is where discovery increasingly begins.

PAS (Problem–Agitate–Solution) is a short-form script structure: name the problem clearly, make the viewer feel its frustration, then introduce the product as the logical fix. It maps cleanly onto a UGC hook → body → CTA and is one of the most reliable converting structures.