With a training video, the test is whether people can do the job after watching it. Nobody in the audience chose to press play, an expert has to confirm every step, and a year later the software or the policy will have changed.
This is for production companies and in-house teams, and most of it happens before the shoot.
In short
- Make one module per task, with an objective you could watch someone do.
- Settle at kickoff who signs off on the facts (the expert) and who signs off on the look.
- Put captions in the quote from the first conversation and deliver them as a separate SRT or VTT file.
- Keep anything that dates out of the spoken lines, and number versions per module.
One module per task, with the objective written first
Before the script, write down what the viewer should be able to do when the video ends, phrased as something you could watch them do.
“A new store manager can close the till and file the end-of-day report” works. “Understand the closing process” doesn’t, because nobody can check whether it happened.
If the objective needs an “and” between two different jobs, you have two modules. Closing the till and ordering stock happen at different times, and the manager who needs one of them next March won’t want to scrub through the other.
Our post on corporate video length puts training at 3 to 6 minutes per module, and splitting by task usually gets you there.
Name each module after the task, in the words employees use. “Closing the till” is easier to find in a learning platform later on than “Module 4: Financial procedures”.
List the modules with the client before you quote. A brief that says “an onboarding video” often turns into twelve modules once someone writes the objectives down, and twelve modules is a different budget.
Get the facts from the expert, and agree who signs off on what
The subject-matter expert knows the process and usually can’t write for the ear, so don’t ask them for a script. Ask for the steps, in a half-hour call or while they walk you through the task with a recorder running, and have your writer turn that into lines a person can say.
The expert then checks the script for accuracy before the shoot, while a wrong step still costs a few words instead of a pickup day.
Settle two sign-offs at kickoff and write both names in the brief. The expert signs off on the facts, meaning the order of the steps, numbers, the names of fields and buttons and anything regulatory. The client’s training or HR lead signs off on the look, the tone and the length.
When the HR lead wants a step cut for time, the expert decides whether it can go. Without that split, each of them expects you to follow their own notes on the same cut.
Our post on conflicting feedback from stakeholders covers what to send when it happens anyway.
The expert should watch the cut too, looking only for mistakes, like an old version of a screen or a step shown out of order. Say so up front, or you’ll also get notes on the music.
Expert on camera, or a voice-over on the screen
For software training, a screen recording with a voice-over usually teaches more than a person on camera, because the viewer needs to see where to click.
Record at the resolution you’ll deliver, close everything that isn’t part of the task and use a demo account with made-up data. A real customer’s name on screen is a privacy problem.
For physical tasks, like operating a machine, film the hands and the equipment up close while the expert explains. A wide shot of someone talking next to the machine gives the viewer nothing to copy.
A face on camera earns its place at the start and the end, saying why the task matters and who to ask. It’s also the part that dates fastest.
If the expert leaves the company, every module with their face needs a reshoot, while someone else can re-record a voice-over without touching the picture. For the shoot itself, see making employees comfortable on camera.
Captions are part of the spec
Put captions in the quote from the first conversation. People watch training at a desk in an open-plan office, often without headphones, sometimes in their second language.
If the client’s accessibility policy is based on WCAG, Success Criterion 1.2.2 asks for captions on prerecorded video at Level A, the lowest level. Those captions identify who is speaking and include the sounds that matter, like the alarm in a safety module.
The same standard asks for audio description at Level AA (Success Criterion 1.2.5) unless the audio already carries everything important on screen. In a screen recording that mostly means saying “click Save, top right” instead of “click here”.
Deliver captions as a separate SRT or VTT file unless the video is going to social. The learning platform can then show or hide them, and a typo gets fixed without a new export. Ask the client which format their platform takes.
Line length and reading speed are in our subtitle guidelines, and how to make an SRT file covers the file.
Plan for the module that will change
A new screen in the software, a new form, a policy rewritten after an audit. With a course split by task, each of those means redoing one module. With one 25-minute video, it means re-editing and re-checking the whole thing.
A few habits make that one module cheaper:
- Keep things that go out of date out of the spoken lines where you can: prices, dates, version numbers and the name of whoever runs the department belong in on-screen text, which is easier to swap than a voice.
- Record the steps separately from the intro and outro, so an update doesn’t touch the shots with a face in them.
- Keep the project files, graphics templates and the voice artist’s contact for as long as the course is in use.
Agree in the quote how updates are priced. A line like “one module updated” can carry a price, while “changes to the course” leaves the scope open.
Names and versions for a course with many modules
Twelve modules with three or four cuts each, plus captions, is fifty-odd files. The pattern in our video file naming convention works here, as long as the module number lives in the title and stays out of the production code.
Course order changes. Halfway through, the client decides “Closing the till” should come before “Ordering stock”, and module 4 becomes module 3. If file names and codes depend on module order, everything gets renamed and somebody sends the wrong file.
With a production code that never changes and the module number in the title (“M03 Closing the till”), you edit the title, and the code and the file history stay where they were.
Number versions per module, too. “v3 of the course” means nothing when module 2 is on its fifth cut and module 9 was approved on the first.
How this works in RecReview
In RecReview, the new task window has a Create several at once button next to the title field. Paste the module list, one title per line and up to 100 at a time, and set the client, project, assignees, delivery deadline and internal deadline once for all of them.
Each module becomes a task numbered with the client’s code, in the order you pasted. The number counts per client and is never reused, so a module keeps its code when the course order changes.
Each module’s task holds its own script, as a file or a Google Drive or OneDrive document, and you can email it to the client from there. If the expert reads to camera, each video block in the teleprompter can take the task’s latest script, keeping only the spoken lines.
A cut uploaded with the module’s code in the file name goes in as a new version of that module.
The expert reviews the cut through a link, with a name and an email instead of an account. If you turn on contact roles, only the client’s Managers can give the final approval, and the expert, as a Reviewer, can comment and ask for changes.
Each version can be transcribed from its Transcript tab. Subtitles are built from the transcript with up to 42 characters per line and two lines. Your team fixes the text on the player and exports SRT or VTT.
The search box also looks inside transcripts, so a line you remember brings up the video it was said in. The project management and transcription pages have more.
