Video that is good enough, and no better
Production quality has a floor that matters and a ceiling that does not. Most first-time instructors spend heavily above the ceiling.
Course video absorbs an enormous share of production effort, and beyond a modest threshold almost none of it affects whether learners finish or learn.
There is a floor, though, and it is not where people think it is.
A useful caution about proxy measures is discussed this article, where apparently active input is separated from meaningful work.
Audio is the floor
Learners tolerate mediocre picture and abandon bad sound within minutes. Poor audio is fatiguing in a way that is hard to push through, and it reads as carelessness more than a plain-looking slide does.
A decent microphone positioned close to the speaker, in a room with soft furnishings, is most of the way there. This is the one place where spending helps, and it is cheaper than almost anything else in the production budget.
When evaluating digital learning tools, accessibility is part of the baseline; the W3C Web Accessibility Initiative provides public standards and guidance.
Audio quality is the floor learners will not go below. Video quality is a preference they barely notice above a low threshold.
What is above the ceiling
Multiple camera angles, motion graphics, animated transitions, studio lighting, background music, a professionally shot introduction. Each takes substantial time and none of it makes the material clearer.
There is a second cost: heavily produced video is expensive to update. A correction that requires re-recording a segment with matching lighting, framing and graphics often simply does not happen, and the course dates.
Screen recording is usually the right format
For most digital skills the useful shot is the screen, with the instructor's voice over it. It shows the thing being done, it is fast to produce, and it is easy to re-record in sections.
The two things that make screen recordings usable are legibility — large text, sensible zoom, no cluttered desktop — and a visible cursor that moves deliberately. Both cost nothing and are more often wrong than the production values are.
Short segments, always
Long recordings are watched at speed, partially, once, and are nearly impossible to revisit for a specific point. Segments of a few minutes each, titled by what they cover, are watched properly and can be returned to.
Short segments are also the only way an update is affordable, since a defect is confined to one file rather than embedded in an hour.
Provide the text
Transcripts and written summaries are usually treated as an accessibility obligation, and they are also simply better for a large share of learners: skimmable, searchable, and readable in situations where video is not.
Captions matter for the same combination of reasons — necessary for some learners, useful for many, and produced cheaply by automatic transcription with a pass of editing.
Re-record less than you want to
First-time instructors re-record constantly to remove stumbles. Learners do not notice minor imperfections and often find a slightly unpolished delivery easier to follow.
Re-record for errors of fact, inaudible sections and confusing explanations. Not for a hesitation, a repeated word, or a delivery that felt awkward at the time.