Screen-recording tutorial Shorts: what YouTube expects you to add

A screen recording of your own software is your own content, so the reused-content rules are not the issue. The risk is the other list: YouTube's channel monetization page names content with minimal variation, and slideshows or scrolling text with minimal narrative, as not allowed (read 2026-10-03). A silent recording that shows a click path with a few labels is close to that description, even though every pixel is yours.

What turns it into a tutorial is the explanation: why you click there, what goes wrong, what to do instead. After the October 1 recommendation change, which Search Engine Journal reports does not count 'voice-over descriptions of what's happening on screen' as original for other creators' clips (read 2026-10-03), the safest standard for your own recording is the same: say something the screen does not say.

What makes a recording a Short

Cut the recording to one task and one reason. A 40-second Short about one setting beats a 3-minute Short that touches six.

Trim, caption, check

Video trim takes one range from one hosted recording: start plus exactly one of end or duration, 0.2 to 900 seconds, $0.02 per job. Recordings are usually landscape; if your Short must be vertical, crop first with Video filter and then trim, as in this guide.

Video captions burns text onto the cut. If you have narration, it transcribes the audio; if you have a silent recording and a written explanation, pass cues with your own text and times and no speech-to-text runs. A standalone caption job costs $0.20 for a clip up to 60 seconds.

This sketch submits a trim and prints the job id. It is a starting point: confirm the media is already on media.sume.com, as the trim doc requires, and poll the job per the jobs docs.

import os, requests

def submit_trim(video_url, start, duration, key=None):
    key = key or os.environ["SUME_API_KEY"]
    r = requests.post(
        "https://api.sume.com/v1/video-trim",
        headers={"Authorization": f"Bearer {key}",
                 "Idempotency-Key": f"trim-{start}-{duration}"},
        json={"video_url": video_url, "start": start,
              "duration": duration})
    r.raise_for_status()
    return r.json()

if os.environ.get("SUME_API_KEY"):
    job = submit_trim("https://media.sume.com/example/setup.mp4", 12, 38)
    print(job)

A checklist before you post

Sume does not guarantee reach or recommendation. It trims and captions what you recorded; the explanation is yours to write.

• Can a viewer say what the one lesson is after 10 seconds?
• Does the audio or caption say why, not only what?
• Is this task different from your last tutorial Short?
• Did you hide secrets, tokens and customer data in the recording?
• Have you described the result you actually saw?

Cutting one task from a long recording

Most tutorial recordings are 10 to 20 minutes with dead time. Mark the timestamps where each task starts and ends as you record, using a note or a clap, so you do not have to scrub later. Each timestamp pair becomes one trim.

If you have the full recording on media.sume.com, Video inspect can pull up to 24 still frames at times you choose with frames: {at: [...]}, so you can check where a menu opens before you commit to a start. Frames are stills of the video, not a screen-reader, so label them yourself.

Keep each Short to one task. If the lesson needs more than about 60 seconds, consider a regular video instead and let a 30-second Short point to it. YouTube's help lists a 3-minute maximum for vertical Shorts (read 2026-10-03), but nothing says a Short should use it all.

Writing the caption cues

Write the reasoning as short lines, about a dozen words each, and give each a start and end. A cue like 'Turn this off: it re-indexes the whole drive' says why. A cue like 'Settings, then Advanced' only repeats the screen. When a caption is only a menu path, delete it and let the picture carry it.