Halloween 2026: five clips matched to NRF's top activities

NRF expects $13.5B of Halloween spending, with 74% celebrating. Match five 5-second clips to the five top activities for about $3.55 in Sume costs.

5 min readSume
All posts

The National Retail Federation's Halloween page, read on 2026-10-03, says consumer spending is expected to reach $13.5 billion in 2026 and that 74% of consumers plan to celebrate. The page's top five activities are handing out candy (61%), dressing in costume (50%), decorating the home or yard (42%), carving pumpkins (42%) and watching scary movies (42%). Five 5-second clips, one for each, cost about $3.55 on Sume.

Halloween is Saturday 31 October, 28 days after 3 October. Older posts here cover the budget and the single-product teaser; this one starts from the survey and picks what to film.

One clip per activity

Treat each percentage as a prompt for a hook, not a promise about your customers. The NRF figure also says 49% of consumers began shopping in September or earlier, so late-October posts are for the people who have not.

Halloween activities per NRF (read 2026-10-03) and a clip for each
ActivityShareClip idea for a shop
Hand out candy61%Bowl of treats on a porch, your product in the bowl
Dress in costume50%Costume reveal, a 'last-minute' accessory
Decorate home or yard42%Doorway before and after
Carve pumpkins42%Tools or snack made for a carving night
Watch scary movies42%Cozy couch, blanket, snack, a flicker of screen light

The cost

Generate each clip with wan-3.0 through POST /v1/videos at 720p and 5 seconds: the pricing table lists $0.10 per second at 720p, billed list x 1.25, so $0.125 per second and $0.625 per clip. Five clips are $3.125. Assemble with one Timeline job ($0.10), add one music bed ($0.125) and burn the five hook lines in a single caption job ($0.20). The total is $3.55.

If you would rather have a first-frame you control, make a still with the Image API and pass it as image_url. Check each result before you use it.

import os, sys, uuid, requests

def main():
    key = os.environ.get("SUME_API_KEY", "")
    if not key:
        sys.exit("Set SUME_API_KEY first")
    body = {
        "video_url": "https://media.sume.com/artifacts/YOUR_ARTIFACT/clean.mp4",
        "style": "slam",
        "cues": [
            {"text": "Treats sorted", "start": 0.0, "end": 3.0},
            {"text": "Costume: done", "start": 5.0, "end": 8.0},
        ],
    }
    resp = requests.post(
        "https://api.sume.com/v1/video-captions",
        json=body,
        timeout=60,
        headers={"Authorization": "Bearer " + key,
                 "Idempotency-Key": str(uuid.uuid4())},
    )
    print(resp.status_code, resp.json())

if __name__ == "__main__":
    main()

Order of work

Day 1: write five hook lines and five prompts. Day 2: generate the five clips. Day 3: review, then assemble and caption. Posting a clip a day from 26 to 30 October means the series finishes before the night. Each clip is its own job, so a failed one is regenerated alone.

What Sume does and does not do

  • Does: generate the clips, music and captions and put them in one MP4.
  • Does not: forecast your sales. The NRF numbers describe consumers in aggregate.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume