Video has unmatched authority in learning — a real person, a real environment, a real demonstration. It's also the most expensive way to say anything, and the hardest to update. The craft is knowing which mode the message actually needs.
Film when seeing is the content
Physical technique, environment, tone. A first-person demonstration of a hands-on skill communicates in forty seconds what three paragraphs can't: hand position, sequence, pace, the visual cues that signal "this is going wrong." The same goes for human moments — a difficult conversation modelled on camera lands differently than a transcript of one. If the learner must eventually see it in real life, show it on screen first.
Animate when the real thing is invisible
Processes, systems, flows of money or data, anything microscopic or abstract — animation exists for what a camera can't capture. Motion design also survives change better: when a step in the process updates, you re-render a segment instead of re-booking a shoot.
Skip it when text is faster
Reference material, definitions, policy detail — anything a learner will want to scan, search or revisit — is worse as video. Nobody scrubs a timeline hunting for the one sentence they need. A well-structured page with a diagram beats a talking head reading the same content, every time.
Whatever you make, respect the edit
Length discipline (under three minutes per idea), captions and transcripts as defaults, and on-screen annotation that points at what matters — the edit is where a competent recording becomes a teaching instrument. Plan it before you press record, not after.
Video isn't a format decision. It's a budget decision about attention — the learner's and yours.