How to make a YouTube video, from idea to upload

Making a YouTube video on your own comes down to seven steps: pick one question to answer, outline it, set up so you can see and hear yourself properly, record it in takes, edit out everything you did not mean to say, add captions, then export and upload. Most of the time goes on the editing, and most of the editing is reading. That is the part worth getting right, and the part this guide spends longest on.

This is for the most common kind of YouTube video there is: one person on camera, explaining something, sometimes with their screen beside them. It assumes a Mac, because we make a Mac editor, but every step except the tool names applies anywhere.

1. Answer one question

A video that answers one question is easier to make, easier to title and easier for the right person to find. "How do I set up a home office for video calls" is a video. "Everything about working from home" is a channel.

Write the question down in the words a viewer would type. It becomes your title, it tells you what to cut later, and it stops the video drifting.

2. Outline it, then decide whether to script it

You do not need a full script, but you do need an order. Three to five points, each one a sentence long, is enough to stop you rambling and to make the editing obvious later.

Whether to write every word is a matter of temperament. A full script reads cleanly but can sound read; bullet points sound natural but produce more false starts. If you script it, read it from a teleprompter rather than a sheet of paper beside the camera, so your eyes stay near the lens. Rooster has one in its recorder (how); so do phone apps and dedicated hardware.

3. Set up for your face and your voice

Viewers forgive an ordinary picture much more readily than bad sound, so spend your effort in that order.

• Sound: get the microphone close to your mouth. A cheap microphone a hand's width away beats an expensive one across the room. Record somewhere soft - curtains, a sofa, a bookshelf - rather than a bare room that echoes.
• Light: face a window or a lamp, so the light falls on your face rather than behind you.
• Camera: at eye level, looking slightly down rather than up your nose. The camera built into a laptop is fine to start with if it is raised to eye level.

If the video shows your screen, close what you would not want a stranger to read: email, messages, other tabs. A recording catches everything on the display, including what you will later cut.

4. Record in takes, and do not stop for mistakes

The single habit that makes editing easier is this: when you fluff a line, pause for a breath and say the whole sentence again. Do not stop recording, do not start the video over, and do not try to pick up halfway through a sentence. A clean restart is a clean cut later.

Record more than you need. A second take of the opening is cheap; going back to re-record next week is not. This is also where a tool that charges by the minute of footage costs you most, because it charges for the takes you throw away as readily as the one you keep. Rooster records and transcribes on your Mac, so there is no meter on how much you record (how, or with your screen).

5. Edit by reading, not by scrubbing

Here is where the time goes. On a timeline, finding a fluffed sentence means scrubbing until you hear it. In a transcript you can read it in a second, which is why editing by text has taken over this kind of video.

The order that works:

1. Get a transcript. In Rooster it starts on your Mac the moment you stop recording, so it is usually ready by the time you have found a coffee.
2. Read it top to bottom once and delete what you did not mean. Every false start, every restarted sentence, every tangent that does not answer the question from step 1. Deleting the words deletes the video under them (how).
3. Remove filler words and long pauses in one pass, then listen to a few cuts. Removing every "um" can make someone sound clipped; keep the ones that carry the rhythm (why).
4. Watch it through once at normal speed. Reading catches what you said; watching catches how it looks, like a cut that lands mid-gesture.

Resist the urge to polish the picture before the words are right. A tight script with a plain picture beats a beautiful picture of someone repeating themselves.

6. Add captions

Captions help people who watch with the sound off, people who do not hear well and people whose first language is not yours. There are two kinds, and it is worth knowing the difference:

• Burned-in captions are drawn into the picture and always show. They suit short videos and social clips.
• A subtitle file (SRT or VTT) is uploaded alongside the video, and viewers can turn it on and off. It suits longer YouTube videos.

Since the edit already produced a transcript, captions are mostly done: check the spelling of names and terms, then style them or export the file (captions, subtitle files).

7. Export, upload and describe it

Export an MP4 at your recording's own resolution; it is the format YouTube recommends and the one almost every site accepts (how). Then, in YouTube, choose Create, then Upload videos, and pick the file.

While it uploads, fill in the parts that decide whether anyone finds it:

• Title: the question from step 1, in the viewer's words.
• Description: two or three sentences on what the video answers, then anything you mentioned. If you have a transcript, it is the raw material.
• Chapters: a list of timestamps in the description, starting at 0:00, turns into chapters on the progress bar. They help viewers jump to the part they came for. On a Mac with Apple Intelligence turned on, Rooster's export sheet suggests a description and chapters from the transcript for you to check and paste.
• Thumbnail: a clear face and a few large words beat a busy collage.
• Subtitles: upload the SRT or VTT file from step 6 if you made one.

The step that decides how many videos you make

The first video is mostly about learning the steps. Every video after that is mostly about how long step 5 takes, because the editing is what stands between recording and publishing, and it is the step people give up on.

That is the case for editing by the words, and for doing it on your own Mac. Reading a transcript is faster than scrubbing a timeline, and when the transcription runs on a Mac you already own, there is no upload to wait for and no meter deciding how many takes you can afford. Rooster is built around exactly that loop - record, read, cut, caption, export - and the YouTube page says what it does and does not do.