How long should a microlearning video be? Length and AI

There is no fixed standard length for a microlearning video: make it as short as its one learning objective allows, and keep it to 6 minutes or less. That ceiling comes from research on course videos: the median engagement time never went past 6 minutes, whatever the video's length.

The research figures below are quoted from a peer-reviewed guide, Effective Educational Videos by Cynthia J. Brame (CBE—Life Sciences Education, 2016), read on 2026-09-28. It summarizes a study of 6.9 million video-watching sessions in four edX MOOCs. The AI avatar limits come from Sume's Generate avatar video docs.

Is there a standard microlearning video length?

No. Microlearning is defined by its unit, one objective or topic per piece, not by minutes. Research on how long students keep watching educational video points the same way: videos under 6 minutes tended to be watched to the end, and longer ones were not.

How do I decide how long mine should be?

Work from the objective, not from a target number:

• Write the single objective as one sentence: what the learner can do after watching.
• Script only what serves that sentence: the setup, the steps or rule, one example, a recap.
• Read the script aloud and time it. That is the video's length.
• If it runs past 6 minutes, or past the time your learners actually have, such as a break between shifts, split it into two objectives and two videos.

How long can an AI avatar microlearning video be?

On Sume, one talking video takes a script that Sume estimates at 4–60 seconds, so you shorten a longer script or split it across several jobs. The current code estimates 2.8 words per second; Script for an AI avatar turns that into a word budget.

A 60-second clip fits a tight objective on its own. A lesson that needs two or three minutes is several clips joined into one video, which AI avatar video longer than 60 seconds walks through.

Inside one clip, ordered video_inputs scenes let you plan a hook, a demo and a silent beat, as long as the whole plan still lands in the 4–60 second window. The presenter speaks English only in the current code.

How do I plan a microlearning series with an AI presenter?

Treat the series as many small videos that share one presenter:

• Create one avatar once and reuse its handle for every lesson, with the same aspect ratio, so the series looks like one course.
• Keep one objective per video and name files by objective, so a learner or an LMS can find the one they need.
• When a procedure changes, re-render only the lesson it touches; the rest of the series stays as it is.
• For the full script-to-module workflow, including captions and cost, see AI training video generator.