Amazon product video shows about 3 days after approval: plan back

Amazon's seller guidance says a shoppable product video typically displays within three days after it is approved, and videos are reviewed against community guidelines first. So a video that must be live on launch day has to be submitted at least three days earlier, plus however long review takes. Add your own production time for generation and revisions, and count backwards from the launch date.

What the Amazon page states

The Amazon product video guide, read on 2026-10-03, gives these points for shoppable video.

Approval time is not the same as display time

The page states the three days as the typical delay after approval. It does not give a length for the review itself, so this post cannot either. Treat review as a variable you set from your own experience, and keep the three days as the fixed part of the schedule.

Sume lead time is the part you control

Generation is asynchronous, so a video is a job you submit and poll. The video docs say a clip typically takes 30 seconds to several minutes depending on model and resolution, and recommend polling about every 30 seconds. They also note a model's maximum resolution matters here: Amazon's page states up to 1080p, and the docs list 1080p among supported resolutions, with each model advertising its own subset.

One more catch from the same table: the page states videos of 1 to 12 minutes, while a single generated clip is seconds long. Timeline 1.0 assembles ordered clips plus an audio spine into one MP4 of up to 1800 seconds at $0.10 per output minute, rounded up. Confirm the length rule on the current Amazon page before you plan a short clip.

A planner that counts backwards

This script takes the launch date and your own estimates, then prints the dates to hit. The three-day figure comes from Amazon's page. The other numbers are inputs you choose.

from datetime import date, timedelta

def plan(launch, review_days, revision_days, generation_days):
    display_delay = 3  # typical, from Amazon's page
    submit_by = launch - timedelta(days=display_delay + review_days)
    final_cut_by = submit_by - timedelta(days=revision_days)
    start_generation = final_cut_by - timedelta(days=generation_days)
    return {
        "start generation": start_generation,
        "final cut": final_cut_by,
        "submit to Amazon": submit_by,
        "launch": launch,
    }

for label, day in plan(date(2026, 11, 20), review_days=2, revision_days=2, generation_days=3).items():
    print(f"{label:>18}: {day.isoformat()}")

What to keep in the schedule

Keep a buffer for rejection. A video that fails review needs a revision and a fresh submission, which restarts the clock, so a launch with no slack depends on the first submission passing. Store the Sume job id, the prompt and the source image with each clip so a requested change is a re-run with a one-line edit rather than a rebuild, and keep the source files that Amazon asked for in the same folder as the finished export.

Four clips, one schedule

With up to four videos per ASIN, stagger them rather than submitting all at once. Put the clip most likely to pass review first, so something is live while a more demanding one is still in review. The four-clip plan covers which four to make.