A social clip has one job: stop the thumb. That job is done in the first three seconds, before any production value matters. GoodAvatar is useful here for a specific reason: the face is generated, so a small team can produce ten variants of the same opening line and let the feed decide which one works.
Feed video is watched muted, on a phone, in a hurry. Three consequences follow:
GoodAvatar charges 1 credit per second. A fifteen second clip is 15 credits, roughly $0.17 at the Starter pack rate, and a nine second hook costs nine. Uploads are capped at 600 seconds per task; anything past thirty seconds is split at pauses, generated in parts, and rejoined, which is slower and costs more credits for the same wall-clock length.
Start with one sentence and one portrait. Create a video, choose 9:16, and generate a nine second hook before committing to a longer script. If the hook does not hold, the problem is the sentence, not the face.
Nine to thirty seconds. The first three seconds decide whether anyone watches the rest, so the hook belongs in the first sentence, not after an introduction.
Short video is watched on a phone. A 9:16 portrait fills the frame without letterboxing, and a face generated for that frame will not fill a landscape slot cleanly later.
Fifteen seconds costs 15 credits, about $0.17 at the Starter pack rate. Ten clips a week for a month costs less than one hour of studio time.
The clip is a complete MP4 with sound. Most teams add captions and an end card in an editor, because feed playback usually starts muted.