On this page
The short version
- 1Pick one task, one learner, and what they should be able to do afterward.
- 2Choose the format: a screen recording for software, a filmed demonstration for hands-on work.
- 3Write an outline, then a script: the goal, the steps, the common mistake, a check.
- 4Record in short takes in a training account, then cut the waiting and add captions.
- 5Publish it where your team learns, and check it again whenever the task changes.
Pick one task and one learner
A training video works when someone can do something afterward that they couldn't do before. So start from a task, not a topic. “Submit an expense claim with a receipt” is a task you can teach in one short video. “Our expense policy” is a topic, and it needs several.
Then picture one learner: a new hire in their first week, a manager approving requests, a support agent on a new tool. What they already know decides what you can skip and what you have to explain. Write two lines before anything else:
- Who it's for: the role, and what they already know.
- What they'll be able to do: one sentence that starts with a verb, such as “approve a time-off request”.
If the goal doesn't fit in one sentence, you have more than one video. For your first tasks, look at the questions new starters ask, the steps people get wrong, and any process that changed recently.
Choose the kind of training video
There are four common kinds, and the task decides which one fits. Software is easiest to teach by showing the screen; a physical process needs a camera.
Screen recording of software
- Fits when
- You're teaching a tool: a form, a report, a setting
- You need
- A screen recorder, a microphone and a training account
- To update it
- Re-record the steps that changed
Talking presenter
- Fits when
- A welcome, a policy, or anything where a familiar face helps
- You need
- A camera, a quiet room, soft light and a script
- To update it
- Film it again, so keep it short
Slides with a voice
- Fits when
- Concepts and rules with no screen to show
- You need
- A slide deck, a microphone and a screen recorder
- To update it
- Edit the slide and re-record its part
Filmed demonstration
- Fits when
- Hands-on work: equipment, safety, a physical process
- You need
- A camera on a tripod, close-ups and clear sound
- To update it
- Film it again
| Kind | Fits when | You need | To update it |
|---|---|---|---|
| Screen recording of software | You're teaching a tool: a form, a report, a setting | A screen recorder, a microphone and a training account | Re-record the steps that changed |
| Talking presenter | A welcome, a policy, or anything where a familiar face helps | A camera, a quiet room, soft light and a script | Film it again, so keep it short |
| Slides with a voice | Concepts and rules with no screen to show | A slide deck, a microphone and a screen recorder | Edit the slide and re-record its part |
| Filmed demonstration | Hands-on work: equipment, safety, a physical process | A camera on a tripod, close-ups and clear sound | Film it again |
You can combine them: a short presenter intro that says why the process matters, then a screen recording of the steps. This guide follows a screen recording most closely, but the outline, script and editing advice apply to all four.
Write the outline, then the script
An outline keeps the video to the task. Fill in one line for each part, and do the task yourself as you write the steps, in a fresh account. You'll catch what experienced people forget: a permission a new hire doesn't have yet, a field that fills itself in for you, a menu that only appears later.
- Learner
- New staff in their first week, using the expense system for the first time
- Goal
- Submit an expense claim with the receipt attached
- Before you start
- Your login, and a photo or PDF of the receipt
- Steps
- 1. Open Expenses and click New claim. 2. Choose a category. 3. Enter the amount and the date. 4. Attach the receipt. 5. Click Submit.
- Common mistake
- Filing a team lunch under Travel. Meals have their own category.
- Check your work
- The claim shows as Submitted, with a paperclip next to it
- Getting help
- Questions go to the finance team's channel
Then write what you'll say: one or two sentences per step, with the exact labels on the screen so people can find them on their own. Give the reason when a step isn't obvious, warn about the common mistake just before it happens, and read the script aloud to catch the lines you stumble on.
The video script templates include a timed training template with the same parts, and the software tutorial video guide goes deeper into naming buttons and steps.
How long should a training video be?
As short as the task allows. A learner who's halfway through a form wants the next step, not an introduction. As a rule of thumb:
- One software task: one to three minutes.
- A process with a few parts: up to about five minutes, with each part clearly marked.
- A whole topic: a series of short videos, each ending with something the learner can do.
Check the length on paper first. A common rule of thumb for speaking pace is 130 to 150 words a minute, so three minutes is about 390 to 450 words of narration, fewer if the screen needs time to load. The free script timer times it at your own pace.
Record it
For software training, how you set up the screen matters more than the equipment.
- A training account. Use a sandbox or test account with realistic sample data, so no real names appear and nothing real changes when you click Submit.
- A readable window. Record one window about 1280 pixels wide rather than a large screen, and zoom the page if the text is small.
- A quiet screen. Notifications off, other tabs closed, bookmarks hidden.
- Short takes. Record one step at a time, pause, then start the next. A mistake then costs you one take, not the whole video.
- A slow cursor. Rest it on the button and say its label before you click.
You don't need special training video software: any screen recorder will do. How to screen record covers the one built into each computer and a free one that runs in your browser. For a presenter or a demonstration, put the camera on a tripod at eye level, face a window or a soft light, and use a microphone close to the speaker rather than the one in the camera.
Check every frame for real data
Edit for learning
Edit for someone who is following along, pausing the video and repeating each step on their own screen.
- Cut the waiting, not the steps. Trim page loads and long typing, but keep every click a learner has to make.
- Zoom in on what matters. Enlarge small buttons and fields, and keep enough of the page in view to show where they are.
- Leave a beat after each click, so people see what changed before the next step starts.
- Add captions. They help anyone watching without sound or in a second language. Here's how to add captions to a video.
- Keep one idea per video. If the edit runs long, split it where the learner has finished something.
End on the finished result and hold it for a couple of seconds, so learners can compare it with their own screen.
Make software training videos with AI
For training on software that runs in a browser, an AI training video generator can do the recording and the editing for you. In SayScreen, you give it the page where the task starts and describe the lesson the way your outline has it: the goal, the steps, the mistake to warn about. Its AI uses the tool in a real browser, reads each screen, decides where to click and writes the narration, and a natural AI voice speaks it, in any of 94 languages. You get an MP4 with smooth zooms, an animated cursor and click highlights, captions on the video if you want them, and an .srt file.
It records tools that are reachable on the internet, including ones behind a sign-in: the tool's own sign-in page opens and you sign in yourself, with single sign-on and 2-step codes. Use a training account with sample data, because the AI acts in a real account and can make mistakes. It can't upload or attach files, so leave a step like attaching a receipt out of your description. If a line is wrong, rewrite it, and it's re-voiced and re-timed without recording again. When the tool changes, make the video again from the same prompt. There are example prompts on the employee training videos page.
Turn your outline into a training video
$1 for your first 7 days, then $49 a month for 60 minutes of video, up to 8 minutes each. Cancel anytime.
Publish it where people learn, and keep it current
Put the video where people already go when they're stuck, not in a folder nobody opens:
- Your LMS, for courses with a set order, quizzes or completion records.
- Your intranet or wiki, on the page about the process, next to the written steps.
- Onboarding, in the checklist or the email for a new hire's first week.
Upload the MP4 with its caption file and title it with the task: “How to submit an expense claim”. Put the steps as text under the video for people who'd rather skim, and add an owner and the date it was last checked.
Training goes out of date quietly: a button moves, a form gains a field. Keep the script next to the video, check the videos that show a screen whenever it changes, and re-record the steps that changed, or the whole video if the look of the tool changed.
A checklist before you publish
- The title names the task, and the first sentence says what the learner will be able to do.
- Every step in the outline is in the video, in order.
- The labels you say match the labels on the screen.
- No real names, email addresses or account numbers appear in any frame.
- Text is readable without going full screen.
- Captions are on and read through, product names included.
- The video ends on the finished result.
- It has an owner and the date it was last checked.
Questions
Pick one task and the person who has to do it, then choose a format: a screen recording for software, a filmed demonstration for hands-on work. Write an outline with the goal, the steps and how to check the work. Record in short takes, cut the waiting, add captions, and publish it where your team already looks for help, such as your LMS or wiki.
As short as the task allows. As a rule of thumb, one software task takes one to three minutes, and a process with a few parts up to about five. If a topic needs more, make a series of short videos that each end with something the learner can do, rather than one long video.
It depends on the format. For software training, you need a screen recorder, a microphone and a video editor that can trim, zoom and add captions, and a recorder that runs in your browser is enough to start. For a presenter or a demonstration, you need a camera on a tripod and a microphone close to the speaker.
Yes, in two ways. Some AI tools turn a script into a video of a generated presenter reading it. Others record software: you describe the task, and the AI uses the app in a real browser, writes and voices the narration and edits the recording. Either way, watch the whole video against the real process and fix anything wrong before you publish.
Use a training or sandbox account with realistic sample data instead. It keeps real customer and employee details out of the video, and you can click Submit without changing anything that matters. Whatever account you use, check every frame for names, email addresses and account numbers before you publish, including in sidebars and notifications.
Keep each video's script or outline next to it, with an owner and the date it was last checked. Whenever a screen changes, look at the videos that show it and re-record the steps that changed. If the look of the whole tool changed, record the video again from the script. Short videos about one task each keep this a small job.